Chapter

Compression of High-Resolution
Satellite Images Using Optical
Image Processing
Anirban Patra, Arijit Saha, Debasish Chakraborty
and Kallol Bhattacharya

Abstract

This chapter presents a novel method for compressing satellite imagery using
phase grating to facilitate the optimization of storage space and bandwidth in satel-
lite communication. In this research work, each Satellite image is first modulated
with high grating frequency in a fixed orientation. Due to this modulation, three
spots (spectrum) have been generated. From these three spots, by applying Inverse
Fourier Transform in any one band, we can recover the image. Out of these three
spots, one is center spectrum spot and other spots represent two sidebands. Care
should be taken during the spot selection is to avoid aliasing effect. At the receiving
end, to recover image we use only one spectrum. We have proved that size of the
extracted image is less than the original image. In this way, compression of satellite
image has been performed. To measure quality of the output images, PSNR value
has been calculated and compared this value with previous techniques. As high-res-
olution satellite image contains a lot of information, therefore to get detail informa-
tion from extracted image, compression ratio should be as minimum as possible.

Keywords: image compression, grating, Fourier transform, image retrieval,

LISS- III sensor

1. Introduction

Compression of images is an important application in the field of satellite image

processing as it is suitable for optimization of storage space and sharing over inter-
net with optimum bandwidth utilization. For compression of satellite images, it is
performed either directly from the image or from transformed part of the images.
As discussed in [1] the compression of satellite images is based on Block Truncation
Coding (BTC) technique. It first converts RGB satellite image into HSV planes.
After that, each of the H and S planes are encoded using block truncation coding
with quad clustering and V plane is encoded with BTC based bi-clustering or tri
clustering depending on the edge information present in the plane. This method
is better than previous BTC methods compared to visual quality of the output
image. [2] discussed the image compression method based on evidence theory and
k-Nearest Neighbor (KNN) algorithm. The main drawback is that the information
loss is large [2]. To improve the quality of the output image, Fourier Transform and
Huffman Coding is used for modification the previous technique. In both method,

1
Satellite Systems - Design, Modeling, Simulation and Analysis

visual quality of satellite image is poor. [3] discusses the use of integer wavelet
regression by increasing temporal correlation, which consequently improves the
compression gain. [4] discusses a satellite image compression technique using dis-
crete wavelet transform for noise removal to compress satellite images. [5] has dis-
cussed the use only hardware-based solutions in this lossless compression technique
of X Sat images. [6] has analyzed the use of Discrete Wavelet Transform in their
lossy image compression work and performance of different wavelets for satellite
image compression. [7] has used the conventional Discrete Cosine Transform sys-
tem for lossless image compression. [8] proposes an image compression technique
using multiplexing and encryption by optical grating method [9–12].
In this chapter, we proposed a scheme to compress multiple high-resolution
satellite images by using phase grating. Each image is modulated by applying high
value of spatial frequency and a fixed orientation angles. For each image, multiple
bands have been generated due to modulation which are placed in the same spec-
trum plane (only three bands are clearly visible). The spectrum is encoded and fil-
tered using Gauss filtering. To detect the location of maximum image information,
an intensity graph has been plotted in the decrypted plane. All stored images can be
securely and efficiently retrieved by applying inverse Regional Fourier Transform
operation. This proposed technique is simple and suitable for optimization of stor-
age space and bandwidth in satellite communication.
The chapter is organized as follows:

• Section I describes the location and data used in this chapter

• Section II discusses the proposed compression method

• Section III presents proposed methodology of our research work

• Section IV provides result of our research work and PSNR value of the
extracted images

• Section V concludes the paper and discuss why low compression ratio is desired
for land cover analysis.

2. Description of location and data used

Images which are used in our research work, collected from Regional Remote
Sensing Centre (East). The images are satellite picture of different areas in Kolkata
Metropolitan Area.
All satellite images used in this paper are captured by LISS III 23 m sensor.
LISS- III sensor is an optical sensor working in four spectral bands (Green, Red,
Near Infrared and Short-Wave Infrared). It covers a 141 km- wide swath with a
resolution of 23 meters in all spectral bands [13, 14].

3. Proposed methodology

3.1 Frequency and orientation angle selection for phase grating

According to rule of phase grating, value of the grating frequency (𝑢0) should be
high. Low grating frequency is creating aliasing problem and therefore it would be
very difficult to reconstruct the original image. In this chapter, we propose to select

2
Compression of High-Resolution Satellite Images Using Optical Image Processing
DOI: http://dx.doi.org/10.5772/intechopen.94147

𝑢0 = 1400, which is sufficient for filtering. In grating, value of the orientation angle
(θ) varies from 0 to 3600. We have worked with 0 deg orientation angle.
The diffraction gratings used are illustrated in Figures 1 and 2.

3.2 Mathematical expression

Let us assume that image, denoted by f1 ( x, y ) , is modulated by two sinusoidal

  m 
phase gratings along the reference abscissa and of the form exp  i   sin2π u1 x 
2 
where u1 is the spatial frequency and m is the phase contrast. In such a case, the
first modulated object may be expressed as:
  m 
s1 ( x, y ) = f1 ( x, y ) exp  i   sin2π u1 x  (1)
2 

Invoking the well-known convolution theorem, the Fourier transform of this

modulated object is given by,

 q=+α
 m   q= +α
 m 
S=
1 ( u, v ) [ F1 ( u, v )  ∑ q J δ ( u − =
qu1 , v )   ∑ J q   F1 ( u − qu1 , v )  (2)
 q=
−α 2   q=
−α 2 

where J q is the qth order Bessel Function of the 1st kind. The diffraction order is
represented by the parameter ‘q’.
The diffraction pattern, as given by Eq. (2) is a series of diffraction spots each
containing the object spectrum. Considering the zero order and the first two orders
of the spectrum, Eq. (2) may be represented by,

 
)  J0  2  F1 ( u, v) + J−1  2  F1 ( u + u1 , v) + J+1  2  F1 ( u − u1 , v)
m m m
S1 ( u, v= (3)
       

3.3 Retrieval of images from spectral band

The reconstruction of images is achieved by decrypting the encrypted plane

followed by plotting an intensity graph of the spectrum horizontally. This intensity
graph is generated using the intensity values with respect to its location (pixel
position). In the intensity graph, peak values indicate the brightest spots, where
information of images are maximum. To avoid any human intervention, system
automatically finds the intensity level horizontally, vertically and diagonally. In
addition, regional Inverse Fourier Transform is applied around the peaks by using a

Figure 1.
Sinusoidal phase grating along x axis.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 2.
Spectrum of an image due to modulation.

fixed cut-off frequency. As only one sideband is enough for image reconstruction,
hence we select one side-band. This spectrum is filtering using a cut-off value of
120. Proper operating frequency is selected and there is no overlapping during the
band selection, so the extracted images are free from aliasing problem. As center
band contains the information of all images, therefore filtering is not applied across
center band.

3.4 Quality checking of filtered image

PSNR value is calculated using Mean Square Error (MSE) approach:

1 m n
∑∑  f ( x,y ) − b ( x,y ) 
2
=PSNR (4)
mn =y 1=x 1

Where f ( x, y ) and b ( x, y ) are original image and retrieved image respectively

and m, n denotes size of images.

4. Results and discussions

Satellite Images chosen for testing the algorithm described in Section III are
shown in Figure 3(a)-(c). The dimensions of the selected images are 512 x 512.

Figure 3.
Three high Resolution Satellite Images.

4
Compression of High-Resolution Satellite Images Using Optical Image Processing
DOI: http://dx.doi.org/10.5772/intechopen.94147

Figure 4.
Three extracted images.

Image Original image size (kB) Compression ratio (compressed PSNR

image/original image)
𝑓1(𝑥, 𝑦) 242 0.84 32.5
𝑓2(𝑥, 𝑦) 240 0.81 31.73

𝑓3 (𝑥, 𝑦) 254 0.86 31.6

Table 1.
Summary of PSNR calculation.

During zonal filtering operation, images 𝑓1(𝑥, 𝑦), 𝑓2(𝑥, 𝑦) and 𝑓3 (𝑥, 𝑦) have
been extracted by Regional Inverse Fourier Transform taking upper spectrum from
horizontal direction. Extracted images are shown in Figure 4(a)-(c), respectively.
A summary of the PSNR calculation is presented in Table 1.

5. Conclusion

In this chapter, phase grating technique has been proposed for compressing
the high- resolution satellite images in frequency domain. The original image is
retrieved by applying Inverse Fourier Transform from the respective spectrum
of the image. As presented here, since we have taken only few coefficients from
the spectrum, the size of the output image is less than the main original image. It
should be mentioned that in high resolution satellite image, compression should be
as minimum as possible. The main reason for the requirement of low compression
ratio is mainly due to large geographical area representation (as these images con-
tain a lot of information). High compression ratio is not suitable for accurate land
cover analysis. To maintain the same dimension with original image and to avoid
aliasing effect, spectral area is carefully selected. Compared with earlier methods,
visual quality of the selected satellite images is very good as it is captured by optical
LISS-III sensor. Our proposed technique is simple and suitable for optimization of
storage space and bandwidth in satellite communication.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Author details

Anirban Patra1*, Arijit Saha2, Debasish Chakraborty3 and Kallol Bhattacharya4

1 Department of ECE, JIS College of Engineering, Kalyani, West Bengal, India

2 Department of ECE, BPPIMT, Kolkata, West Bengal, India

3 RRSC (East – A Unit of ISRO), Kolkata, West Bengal, India

4 Department of Applied Optics and Photonics, University of Calcutta,

West Bengal, India

*Address all correspondence to: [email protected]

© 2021 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

6
Compression of High-Resolution Satellite Images Using Optical Image Processing
DOI: http://dx.doi.org/10.5772/intechopen.94147

References

[1] B. Balakrishnan, S. H. Darsana1, J [9] A Patra, A Saha, K Bhattacharya ;

Mathews, M S. Nair; “Satellite/Aerial “Multiplexing and encryption of images
Image Compression Using Adaptive using phase grating and random phase
Block Truncation Coding Technique”; mask” ; Optical Engineering 59 (3),
Journal of Indian Society of Remote 033105-033111; 2020
Sensing, July 2018
[10] A Patra, D Chakraborty, A Saha,
[2] K Sahnoun, N Benabadji ; “Satellite K Bhattacharya ; “ Compression of
Image Compression Technique Based Satellite Images using Sinusoidal
On The Evidence Theory”; Advanced Amplitude Grating”: ; IJEE 11(1); 2019
Computing: An International Journal
(ACIJ), Vol.5, No.1, January 2014 [11] A Patra, S Bandyopadhyay, D
Chakraborty, A Saha ; “ A Novel
[3] K Sahnoun, N Benabadji ; “Satellite Approach to Compression of Satellite
Image Compression Algorithm Based Images Using Butterworth Filtering”;
On The FFT” ; The International Journal Springer LNSS; 2020
of Multimedia & Its Applications
(IJMA) Vol.6, No.1, February 2014 [12] A Patra, D Sutradhar, S Nath, S
poddar, R Nandi ; “ Study on Video
[4] Md. Al Mamun and Md. Ali Compression and Reconstruction using
Hossain ; “Satellite Image Compression Downsampling and Upsampling”;
Using Integer Wavelet Regression” ; International Journal of Advance
International Conference on Electrical, Research in Science and Engineering
Computer and Communication 9(5); pp-12-17; 2020
Engineering (ECCE) ; 2017
[13] Madhubala M , S.K.Mohan Rao,
[5] K Sahnoun, N Benabadji ; “Satellite G. Ravindra Babu; Classification of
Image Compression Technique Using IRS LISS-III Images by using Artificial
Noise Bit Removal and Discrete Wavelet Neural Networks; IJCA Special Issue on
Transform”; International Journal of “Recent Trends in Image Processing and
Imaging and Robotics; Volume 15, Issue Pattern Recognition” RTIPPR, 2010.
Number 3 , 2015
[14] R Naidu, M.V.S.S Giridhar ;
[6] R.M. Susilo , T.R. Bretschneider; Un-Supervised Classification of
“On the real time satellite image Rice Crop using IRS LISS III Satellite
compression of X-Sat” ; Fourth Images for Wazirabad Command Area;
International Conference on International Journal Of Engineering
Information, Communications and Development And Research
Signal Processing, 2003

[7] T Memane and S D Ruikar ;

“Selection Of Wavelet For Satellite
Image Compression Using Picture
Quality Measures”; International
Conference on Communication and
Signal Processing ; 2014

[8] I. Hacihaliloglu and M. Kartal ; “DCT

and Wavelet Based Image Compression
In Satellite Images”; Recent Advances in
Space Technologies, 2003. RAST’03.

7
Chapter

Overview of Existing and Future

Advanced Satellite Systems
John Nguyen

Abstract

This chapter presents an overview of legacy, existing, and future advanced

satellite systems for future wireless communications. The overview uses top-down
approach, starting with a comparison between a typical commercial regular satellite
system and a high-throughput satellite (HTS) system, following by a discussion on
commonly used satellite network topologies. A discussion on the design of satellite
payload architectures supporting both typical regular satellite and HTS with associ-
ated network topologies will be presented. Four satellite payload architectures will
be discussed, including legacy analog bent-pipe satellite (ABPS); existing digital
bent-pipe satellite (DBPS) and advanced digital bent-pipe satellite using digital
channelizer and beamformer (AdDBPS-DCB); and future advanced regenerative
on-board processing satellite (AR-OBPS) payload architectures. Additionally, vari-
ous satellite system architectures using AdBP-DCBS and AR-OBPS payloads for the
fifth-generation (5G) cellular phone applications will also be presented.

Keywords: high-throughput satellite, analog bent-pipe satellite, digital bent-pipe

satellite, digital channelizer and beamformer, advanced regenerative on-board
processing satellite, cellular phone

1. Background and introduction

Recently, the space industry has pointed out that in the past 5 years, the com-
mercial market has been driving the advancement of satellite technology. Lockheed
Martin is building commercial satellites (e.g., Hellas-sat series) with advanced
on-board processing capabilities for the Saudi Arabian [1]. Hellas satellites probably
will be the first commercial HTS with a very advanced digital processor on-board.
The focus of this chapter will be on commercial satellite systems for communication
applications, and a comparison study between commercial HTS and typical satel-
lites systems conducted by Inmarsat will be provided [2].
For communication applications, commercial satellite systems have been catego-
rized as mobile satellite services (MSSs), fixed satellite services (FSSs), broadcast
satellite services (BSSs), and high-throughput satellite (HTS) services. Depending
on the services, satellite payload architecture will be designed to meet the specified
requirements for that service. Basically, satellite payload architecture can be classi-
fied into four categories: (1) analog bent-pipe satellite (ABPS); (2) digital bent-pipe
satellite (DBPS); (3) advanced digital bent-pipe satellite using digital channelizer
and beamformer (AdDBPS-DCB); and (4) advanced regenerative on-board pro-
cessing satellite (AR-OBPS). This chapter provides an overview of these payload

1
Satellite Systems - Design, Modeling, Simulation and Analysis

architectures and presents two satellite system architectures using AdBPS-DCBS

and AR-OBPS payloads for the fifth-generation cellular phone (5G) applications.
The chapter is organized as follows: Section 2 provides a comparison between
commercial HTS and typical satellite systems; Section 3 discusses the typical satel-
lite network topologies; Section 4 presents an overview of legacy ADPS transpon-
der, existing DBPS transponder, AdBPS-DCBS transponder, and AR-OBPS satellite
system; Section 5 discusses the use of AdBPS-DCBS transponder and AR-OBPS
payloads for the fifth-generation cellular phone (5G) applications; and Section 6
concludes the chapter with a summary and brief discussion of way forward.

2. Typical commercial satellites and HTS comparison

Typical and regular commercial satellites are operating in C-band, Ku-band, and
Ka-band with downlink frequencies approximately at 4, 12, and 40 GHz, respec-
tively. For C-band, Ku-band, and Ka-band, the spectrum bandwidths available by
geostationary orbital position are 500 MHz, 500 MHz, and 3.5 GHz, respectively.
Typical antenna types for these regular commercial satellites are pointed antenna
type with a single beam. Typical diameters for these pointed antennas are (a)
greater than 1.8 m for C-band; (b) 0.9–1.2 m for Ku-band; and (c) 0.6–1.2 m for
Ka-band satellite. Figure 1(a) illustrates a typical regular commercial satellite.
Typical HTSs are usually also operating in Ku-band and Ka-band with the same
downlink frequencies as the regular satellites except that they employ multiple
pointed beam as oppose to a single-pointed beam. Figure 1(b) describes a multiple
beam HTS system. The salient feature of multiple beams is the frequency reuse.
The frequency reuse is defined as the number of times a satellite can reuse the
same spectrum and frequencies. However, high frequency reuse factor can cause
potential cochannel interference or an increase in carrier-to-interference power
ratio (CIR or C/I). IMMARSAT has reported that a reuse factor of 5–30 is possible
with multiple spot beams employed by commercial HTS. Depending on the number
of beams implemented on-board of the satellite, the cost for HTS can be twice of
the cost for a regular satellite. But, the cost per bit for HTS is much lower than the
regular satellite. HTS is a preferred option for point-to-point services, for example,
beyond line-of-sight (BLOS) cellular phone services. Table 1 provides a summary
of the comparison of HTS and regular commercial satellites [2].

Figure 1.
Typical commercial satellites and HTS configurations.

2
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

Comparison Typical regular Typical Remark

factor commercial satellite high-throughput
satellite (HTS)
Operational C-band, Ku-band, Ku-band, Ka-band It should be noted that for
frequency band Ka-band data presented here, all
satellites and supply are not
Throughput ~1–10 ~5–300+ (with
equal; various technical,
capability (Gbps) frequency reuse in
regulatory, and commercial
multiple spot beam)
parameters come into play
Typical cost ~200–300 ~300–500 (cost can when comparing the two-
including launch be twice of regular type satellites. Data collected
(USD) satellite) from IMMARSAT. Source:
see [2]
Advantages Wide coverage; Higher bandwidth/
preferred solution for lower cost per bit;
point-to-multipoint preferred option for
communication point-to-point services
Disadvantages Limited supply Higher upfront costs;
available; lower difficult to find
spectrum efficiency enough customers to
for an equivalent fill each of the beams
frequency

Table 1.
Comparison of typical commercial satellites and HTS.

3. Typical commercial satellite network topologies

This section describes the most commonly used satellite network topolo-
gies, namely “Star” satellite network (Section 3.1) and “Mesh” satellite network
(Section 3.2).

3.1 Typical “star” satellite network

A typical commercial satellite network topology consists of an uplink from a

central anchor station (aka satellite Gateway or satellite Hub) to a satellite and a
downlink from the satellite to users. Users can be mobile or fixed users. Mobile users
can be located in an airplane, a boat, or a car. Fixed users can be located in a build-
ing or a cellular base station. The “star” satellite network is derived from a spoke-
hub distribution paradigm in computer networks, where one central hub serves
as a conduit to transmit messages among network users [3]. Thus, for star satellite
networks, all communications will be passed through a satellite gateway. As shown
in Figure 2, if Mobile User 1 wants to talk to Mobile User 2, Mobile User 1 needs to
send its messages to the satellite gateway (yellow lines), and satellite gateway relays
that messages to Mobile User 2 (red lines).

3.2 Typical “mesh” satellite network

The “mesh” satellite network topology is derived from a local network topol-
ogy, where the network nodes are corrected to each other directly, dynamically,
and nonhierarchically to as many other nodes as possible [4]. In this network
topology, the network nodes can cooperate with one another to route data from
one user to another user efficiently. Hence, for mesh satellite network, Mobile
User 1 can talk to fixed user directly without going through the satellite gateway
(solid lines), and Mobile User 2 can also talk to the fixed user directly (dash lines).

3
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 2.
Typical “star” satellite network.

Figure 3.
Typical “mesh” satellite network.

Any one of the user within the network can send the messages to a terrestrial
network through the red lines representing uplink and downlink between the
satellite gateway and the satellite (Figure 3).
Star satellite network topology does not require advanced satellite payload pro-
cessing on-board and multiple beam, but mesh satellite network requires advanced
on-board processing and multiple beam allowing one user to communicate to
another user automatically and effectively. Section 4 discusses various satellite
payload architectures used in regular satellite and HTS for star and mesh satellite
network applications.

4. Legacy, existing, and advanced satellite payload architectures

This section presents an overview of legacy, existing, and advanced satellite

payload architectures. Section 4.1 presents legacy ABPS payload architecture,
Section 4.2 provides a description of a typical existing DBPS payload architecture,

4
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

Section 4.3 discusses AdDBPS-DCB payload architecture, and Section 4.4 provides
an overview of AR-OBPS payload architecture.

4.1 Legacy analog bent-pipe satellite (ABPS) payload architecture

A typical legacy ABPS payload architecture is depicted in Figure 4, where

the payload has multiple beam antennas (MBAs) using parabolic dishes. For this
architecture, the RF signal is received at the satellite payload and amplifies by a low
noise amplifier (LNA) for increased received signal-to-noise power ratio (SNR).
The RF signal with increased SNR is downconverted (D/C) to an intermediate
frequency (IF) and processed by an IF filter to clean up the signal from adjacent
interference and out-of-band noise. The clean-up signal is then (a) routed to the
proper downlink port by an IF analog switching circuit and upconverted (U/C)
to RF, (b) combined by a multiplexer (MUX), and (c) amplified by a high-power
amplifier (HPA) for downlink transmission.
As illustrated in Figure 5, there are two options for the D/C, namely Option 1
(see Figure 5(a)) is a double downconverter using two local oscilators (LOs) to
downconvert RF signal to IF signal with stable and low phase noise, and Option 2
(see Figure 5(b)) is single downconverter using a LO downconverting RF signal
directlty to an IF signal. Option 1 is being used in many legacy, existing, and
advanced satellite payloads. Option 2 is mostly used in advanced satellite payloads.
Figure 5(c) shows commercial-of-the-shelf (COTS) phase noise characteristics
for typical LOs operating at X-band, Ku-band, and Ka-band. X-band, Ku-band, and
Ka-band illustrated in this figure correspond to 7–11.2, 12–18, and 26.5–40 GHz,
respectively. The main advantages of Option 2 using single downconversion are its
low cost, small size, and low power consumption (also known as small SWAP-C).
This option uses the smallest number of external components as compared to
Option 1 using double downconversion, which is also known as super heterodyne
receiver [5]. However, Option 2 suffers amplitude and phase imbalances caused
by imperfect references associated with I-Q components, direct current (DC)
signal due to self-mixing, and flicker noise.1 Option 1 does not suffer from these
problems and offers excellent selectivity and sensitivity, that is, better rejection of
adjacent interferences. Option 1’s disadvantages are the integration complexity and
high SWAP-C.
In satellite electronic communications, MUX is a multiplexer, which is a device
that selects several (multiple) analog (or digital) input signals and outputs a single
signal. Figure 6(a) describes a functional MUX (aka multiplexer) circuit. On the
contrary, Figure 6(b) depicts a DEMUX (aka demultiplexer), which is an electronic
device that sends a single input signal to multiple signal outputs.

4.2 Existing digital bent-pipe satellite (DBPS) payload architecture

Figure 7 presents an existing DBPS payload architecture using on-board digital

channelizer. Similar to analog payload, there are two options for the RF-to-IF down-
conversion process. Double-downconversion process is typically used for digital
bent-pipe payload architecture.
Figure 8 depicts typical RF-to-IF (or baseband) downconversion and digitiza-
tion and sampling processes for a commercial DBPS payload architecture. The
RF-to-IF process shown in this figure uses Option 1, double downconversion,
and the digitization and sampling process employing bandpass sampling with

1
Flicker noise is a type of electronic noise with a 1/frequency power spectral density.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 4.
Legacy ABPS payload architecture.

Figure 5.
Options for RF downconversion and associated LO’s phase noise.

Figure 6.
Functional block diagrams of MUX and DEMUX.

digital quadrature technology [6]. The RF bandwidth (BW) associated with the
RF bandpass filter (BPF) is selected to match with an over channel bandwidth
(e.g., a maximum of 500 MHz for Ku-band). The automated gain control (AGC)

6
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

Figure 7.
Existing DBPS payload architecture.

Figure 8.
Typical R/F downconversion and digitization processing approach.

is designed to maintain a constant power over the specified channel bandwidth.

There are several advantages associated with bandpass sampling with digital
quadrature techniques, including (a) no phase and amplitude imbalances;
(b) digital finite impulse response (FIR) filters are flexible and computational
complexity with linear phase introducing a constant group delay; (c) only one
A/D converter is required (less weight and power); and (d) when the sampling
period is set at one-quarter of the carrier frequency, the reference in-phase and
quadrature components reduce to an alternating sequence between I-channel and
Q-channel [6].
As shown in Figure 9, the key design issue associated with the digitization
and sampling processing is the selection of required number of bits of the analog-
to-digital (A/D) conversion to (1) achieve optimum loading factor (LF) and
(2) minimize the quantization noise. The LF is defined as the root mean square
(RMS) of the total input signal voltage-to-A/D converter saturation voltage
ratio. The total input signal voltage includes desired signal voltage (S) plus noise
voltage (N) plus interference voltage (I). Figure 10 illustrates an optimum LF as
a function of number of bit of a typical A/D converter. As an example, for 4-bit,

7
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 9.
Existing digitization and sampling processing using bandpass sampling with digital quadrature technique.

Figure 10.
Optimum LF as a function of number of bit of A/D converter.

the optimum LF is about 0.4. In conjunction with LF, the number of bit should
be selected to maximize the signal-to-quantization noise ratio (SQNR) using the
following relationship:2

SQNR ≈ 1.761 + 6.02. N dB (1)

As an example, when N = 4 bits, signal-to-quantization noise ratio is about

25.84 dB.
The key feature of DBPS payloads is the flexibility of the digital channelizer.
Current digital technologies allow for the implementation of robust and reconfigu-
rable digital channelizer adapting to require the number of users and associated
users’ data rates. A typical flexible digital channelizer using polyphase/discrete
Fourier transform (DFT) technology is shown in Figure 11.
As shown in Figure 11, the heart of a typical digital channelizer is a polyphase-
filter network (or simply a polyphase network) and a DFT processor. A typical
polyphase network with a DFT processor is described in Figure 12. The polyphase
network consists of a set of NC digital filters with transfer function H0, H1..., HNc-1,
which is obtained by shifting a basic low pass complex filter function along the
frequency axis [7]. As an example, for a typical 500 MHz channel bandwidth,
assuming for a typical user data rate of 4 MHz and a guardband of 1 MHz, digital
channelizer, NC = 500/(4 + 1) = 100, that is, the number of filter is 100, and each has
a total of 5 MHz bandwidth. A change in sampling frequency by a factor of NC can

2
Quantization (signal processing). Available from: https://en.wikipedia.org/wiki/
Quantization_(signal_processing).

8
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

Figure 11.
Typical digital channelizer using polyphase/DFT technology.

Figure 12.
Typical Polyphase/DFT Technology.

be introduced, thus allowing the circuit in different paths of the polyphase network
to operate at lower frequency than the original sampling frequency. A practical
implementation of a high-throughput low-latency polyphase channelizer can be
found in [8, 9].
Figure 12 shows an example of five input signals, namely S1, S2, S3, S4, and S5,
and the channelizer will select signal interest by filtering out the other signals. As an
example, the signal line with the filter transfer function of H0 filters out S2, S3, S4,
and S5 and sends S1 as an output signal.

4.3 Advanced digital bent-pipe satellite using digital channelizer and

beamformer (AdDBPS-DCB) payload architecture

For a typical commercial HTS system architecture, it usually requires on-board

multiple beam phase array (PA) antenna with associated adaptive digital beam-
former network (DBF) for spot beamforming and frequency reusing of the spot
beams when the beams are not located near each other. Figure 13 describes a typical
AdDBPS-DCB payload architecture, where the digital channelizer is combined with
a DBF to make a “digital channelizer and beamformer” (DCB) [10–12]. For this
payload architecture, the key feature that differentiates this architecture with the
ones discussed above is the combined digital channelizer using polyphase network/
DFT processor and DBF (PolyN/DFT-DBF).
As pointed out in [10–12], DCB architecture shown in Figure 13 can be designed
to (1) form individual beams for each active receive and transmit communication
channels; (2) adaptively generate channel beam steering weights to dynamically
vary the bandwidth, location, and shape of each beam based on traffic demands
and the locations of other, potentially interfering beams avoiding adjacent chan-
nel interference; (3) use digital beamforming weight calibration to compensate
for the temporal and thermal phase and amplitude response variations inherent
in analog multibeam phased array antennas; and (4) adjust the gain of individual

9
Satellite Systems - Design, Modeling, Simulation and Analysis

receive-and-transmit channel beams automatically to compensate for propagation

path and analog payload response variations. In general, there are two possible DCB
implementation approaches, namely DCB Approach 1 and DCB Approach 2 [13].
Figure 14 describes the DCB Approach 1 for processing the uplink signals, where
the uplink signals are individually processed by the digital channelizer (i.e., PolyN/
DFT processing) and DBF independently and separately. DCB Approach 1 requires
a larger computational load because each DBF processes all the user link bandwidth
(e.g., S1, S2, S3, S4, and S5 in Figure 12) at all times to form multiple beams.
DCB Approach 2 is shown in Figure 15, where DCB utilizes an unified process-
ing approach with each DBF processes only the bandwidth corresponding to a
beam (S1 in Figure 12) at normal times. During anomaly operation condition
(e.g., natural disaster event), when the bandwidth has to be reassigned to specific
areas, the arithmetic load on DBF can be reduced by implementing multiple DBFs,
with each capable of processing a bandwidth narrower than that assigned to a beam
(i.e., smaller channel unit). This approach enables a reduction in wasteful arithme-
tic resource usage on bandwidth.
If one defines the number of multipliers, D implemented in each Tx/Rx DBF as
C/fop, where C is the computational load of a DBF (multiplications/sec), and fop
is the operation frequency of the multiplier. Let us compare D calculations between
DCB Approach 1 and DCB Approach 2. Let us assume the following parameters: n
is the number of array elements, m is the number of beams, an userlink processing
bandwidth of 28 MHz, 5 frequency repetitions of the userlink, and an operating

Figure 13.
AdDBPS-DCB payload architecture.

Figure 14.
DCB Approach 1: PolyN/DFT and DBFN individual processing.

10
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

Figure 15.
DCB Approach 2: Unified and combined PolyN/DFT and DBFN individual processing.

frequency of multiplier of 256 MHz. Using these values, D for the DBF/channelizer
of the DCB Approach 1 configuration becomes [13]:

( n × 4 × m × 28 ∗ 106 [multiplications /s ]) (2)

(256 ∗ 106 [multiplications /s ])
and that for DCB Approach 2 configuration becomes [13]:

( n × 4 × m × 28 ∗106 /5[multiplications /s ]) (3)

  multiplications  
 256 ∗ 106    × 2
  s 
The latter calculation assumes an ideal case in which DBF network (DFBN)
processing is performed on a channel-by-channel basis. The complexity of DCB
Approach 2 configuration is 10 times less complex than DCB Approach 1.
As pointed out in [12], the DBFN when coupled with a digital channelizer (aka
DCB) offered more capabilities with many advantages. Nguyen et al. [14] devel-
oped a computer simulation model of a typical DBFN in MATLAB and presented
simulation results for X, Ku, and Ka BFNs using 60-element, 104-element, and
149-element, respectively. Figure 16 is an extracted Ka-band BFN result showing
the achievable antenna gain of 45.5 dB at 3-dB beamwidth of 0.9°. For practical
applications, the DBFN will shape the beam size depending on the coverage area
and desired number of beams. Nguyen et al. [14] pointed out that for 2.5° coverage
area and the desired number of beams of 7, the minimum 3-dB beamwidth of 1.1°
is required. Nguyen et al. [14] also pointed out that DCB can provide a significant
increase in frequency reuse, where the frequency reuse is defined as the number of
times a satellite can reuse the same spectrum and frequencies. High frequency reuse
factor can cause potential cochannel interference (CCI) that results in a decrease
in carrier-to-interference power ratio [aka (C/I) CCI]. As pointed out in [14], for
dynamic allocation using real-time allocation of beams so that the coverage radius of
a cell is equal to the satellite pointing error, assuming satellite pointing error of 0.02
degree pointing error, the (C/I)CCI is about 25 dB for frequency reuse factor 40 [14].

4.4 Future advanced regenerative on-board processing satellite (AR-OBPS)

payload architecture

Figure 17 depicts a potential future AR-OBPS payload architecture [10]. The

payload includes (1) a typically set of digitized analog multiple beam antenna
(MBA) input signals, digitally frequency division demultiplex each input signal to
produce single carrier per channel (SCPC) signal data and demodulate and decode

11
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 16.
Antenna beamwidth and gain of a notional Ka-band DBFN with 12-bit quantization [14].

Figure 17.
AR-OBPS payload architecture.

individual traffic channels to recover the original information bits transmitted on

the uplink; (2) a set of digitized analog multibeam phase array antenna (MB-PAa)
input signals, digitally frequency division demultiplex each input signal to pro-
duce SCPC signal data and demodulate and decode individual traffic channels to
recover the original information bits transmitted on the uplink; and (3) fast packet
switches are typically employed at the AR-OBPS payload’s core to realize statisti-
cal multiplexing gains by efficiently packing and moving data through the switch
and onto the downlink in bursty uplink transmission applications. Moreover, the
digital bandwidth (in Hz) through the AR-OBPS switch is at least 25 times less3
than that supported by an equivalent (pre-demodulation) digital baseband switch
at the center of a DC- or DCB-based system. AR-OBPS payload can also support
digital beamforming, following the frequency division demultiplexing operation, if
a phased array is employed in place of the analog MBA. On the secondary (output)
side of the switch, each user’s binary information is channel encoded and modu-
lated onto a carrier. The modulated carrier data thus produced are multiplexed,
3
Assumes 1 bps/Hz modulation efficiency, 10 bit signal data quantization, and 2.5× practical Nyquist
sampling rate.

12
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

digital-to-analog converted, and passed through an analog reconstruction filter to

generate output signals for the transmit portion of the communication payload. The
channel codes and modulations employed on the uplink (input) communication
channels clearly do not need to be the same as the channel codes and modulations
used on the transmitted downlink channels. Hence, an AR-OBPS payload can
serve as a “translator” facilitating single-hop communications between terminals
employing different link protocols. However, if either the digital multichannel
demultiplexer (DMCD), demodulator, decoder, or digital multichannel multiplexer
(DMCM) encoder modulator, multiplexer (MCEM2) functions are implemented
in ASICs to minimize size-weight-and-power (SWaP), then the AR-OBPS sys-
tem becomes somewhat inflexible, unable to support either uplink or downlink
terminals, respectively, using communication protocols differing from those for
which the AR-OBPS was specifically designed. For this reason, AR-OBPS systems
are typically employed in support of “private networks” in which the communica-
tion satellite service provider only accommodates terminals designed to work on
the provider’s network. Iridium and Spaceway are two examples of commercial
AR-OBPS-based communication satellite systems.

5. Satellite system architectures for 5G cellular phone applications

Sections 5.1 and 5.2 present a notional satellite system architecture using AdBPS-
DCBS satellite payload and AR-OBPS satellite system architecture for 5G cellular
phone applications, respectively.

5.1 AdBPS-DCBS satellite system architecture for 5G applications

AdBPS-DCBS satellite payload can be used to support 5G users. There are

potentially two satellite system architecture options for using AdBPS-DCBS satellite
payload to support 5G mobile user equipment (aka 5G-UE), namely AdBPS-DCBS
Option 1 and AdBPS-DCBS Option 2. For AdBPS-DCBS Option 1, the AdBPS-DCBS
satellite provides communication services directly to 5G-UEs. While in AdBPS-
DCBS Option 2, the satellite provides services to 5G-UEs through the 5G relay nodes
(RNs). Figure 18 illustrates the AdBPS-DCBS satellite system architecture for (a)
AdBPS-DCBS Option 1 and (b) AdBPS-DCBS Option 2 [15].
Figure 18(a) shows that the AdBPS-DCBS satellite requires new radio (NR)
interfaces between (1) AdBPS-DCBS satellite and terrestrial gateway (GW) and (2)
AdBPS-DCBS satellite and 5G-UEs. In addition, it is also required a 5G narrow-band
(gNB) processing station to process the 5G signals from the next generation core
(NGC) network before passing the 5G data to public data network.

5.2 AR-OBPS satellite system architecture for 5G applications

Similar to AdBPS-DCBS satellite payload, AR-OBPS satellite payload can also be

used to support 5G users. There are also two satellite system architecture options for
using AR-OBPS payload to support 5G mobile user equipment, namely AR-OBPS
Option 1 and AR-OBPS Option 2. For AR-OBPS Option 1, the AR-OBPS satellite
provides communication services directly to 5G-UEs. For AR-OBPS Option 2, the
satellite provides services to 5G-UEs through the 5G RNs. Figure 19 describes these
two AR-OBPS architecture options, namely (a) for AR-OBPS Option 1 and (b) for
AR-OBPS Option 2. For these two system architecture options, the gNB processing
is now incorporated into the AR-OBPS satellite payload and no longer required for
the ground system. The GW now can pass the 5G data directly to the NGC. The

13
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 18.
AdDBPS-DCB satellite system architectures for supporting 5G users.

Figure 19.
AR-OBPS satellite system architectures for supporting 5G users.

decoding-demodulation and encoding-modulation processing on-board of the

satellite will be designed to align with the 5G waveform specifications, including 5G
modulation and coding schemes.
Figure 19(a) shows that the AR-OBPS satellite also requires NR interfaces
between (1) AR-OBPS satellite and GW and (2) AR-OBPS satellite and 5G-UEs.
Similar to AdBPS-DCBS satellite system architecture options, the NR interfaces
between the AR-OBPS satellite and 5G-UEs are new. Since the gNB processing is
now placed at AR-OBPS satellite payload, the NR interfaces between AR-OBPS
satellite and 5G-UEs are not the same as the AdBPS-DCBS satellite and 5G-UEs. To
show the differences between the two, Figures 19(a) and (b) use Sat-NG-C and Sat-
NG-U to indicate the new radio interface between (1) terrestrial GW-NGC-and-AR-
OBPS satellite and (2) AR-OBPS satellite-and-terrestrial GW-NGC, respectively.

14
Overview of Existing and Future Advanced Satellite Systems
DOI: http://dx.doi.org/10.5772/intechopen.93227

6. Conclusion

This chapter uses a top-down approach for providing an overview of legacy,

existing, and future advanced satellite payload architectures for future wireless
communication applications. The chapter focuses on the commercial satellite
technologies based on the research results presented in [1, 2]. Section 2 provides the
comparison results performed by Inmarsat describing the technical characteristics
and associated advantages and disadvantages between commercial HTS and typical
satellite systems currently available in commercial satellite market. In Section 3,
two most commonly satellite network topologies used by existing commercial
satellite networks are presented, and the concept of satellite uplink and downlink
associated with star satellite network and mesh satellite network is discussed. The
satellite network topologies presented lead to Section 4, where four satellite payload
architectures are discussed. The legacy analog ABPS payload architecture is shown
to be more appropriate for star satellite network than mesh network. Existing
digital DBPS and AdDBPS-DCB payload architectures are designed for support-
ing mesh satellite network with large number of mobile users. Future advanced
digital satellite payload architecture, namely AdDBPS-DCB, is also presented in
this section. With decoding-demodulating and encoding-modulating processing
on-board of the satellite, AR-OBPS allows for packet switching on-board and higher
quality of service (QOS) than existing DBPS and AdDBPS-DCB at the expense of
higher SWAP-Cost (SWAP-C). Section 4 of the chapter discusses the applications of
AdBPS-DCBS and AR-OBPS payloads for supporting 5G users. Four satellite system
architecture options are presented for supporting the future 5G users.

Conflict of interest

The preparation of this chapter was not funded by Gulfstream, and it was done
by the author using his own time and resources; thus, it does not represent the
Gulfstream’s view on the results presented in this chapter.

Notes/Thanks/Other declarations

The author wishes to thank his wife, Annie Luu-Nguyen, for her immense
patience and support.

Author details

John Nguyen
JohnDTN Consulting Services, Huntington Beach, California, USA

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

15
Satellite Systems - Design, Modeling, Simulation and Analysis

References

[1] Magnuson S. Commercial Satellite Communications, Technical

space—Not military—Driving Paper. Mitsubishi Electric Advance
satellite innovation. National Defense (Technical Journal published by
Magazine. 2018. Available from: https:// Mitsubishi); 2014. pp. 7-10
www.nationaldefensemagazine.org/
articles/2018/4/17/commercial-space- [10] Butash TC, Marshall JR. Leveraging
not-military-driving-satellite-innovation digital on-board processing to increase
communications satellite flexibility and
[2] Revillon P. Fundamentals effective capacity. In: 28th AIAA ICSSC.
and Dynamics of the Satellite AIAA 2010-8715; Anaheim, CA; 30
Communications Business. Presented August–2 September 2010
at the Euroconsult for Inmarsat Capital
Markets Day in 2016. Available from: [11] Sichi SF, Ziegler HE. Beamforming
www.euroconsult-ec.com architectures for advanced MSS
network deployment. In: 29th AIAA
[3] Star Network. Wikipedia. Available ICSSC; AIAA 2011-8021; Nara, Japan;
from: https://en.wikipedia.org/wiki/ 28 November–01 December, 2011
Star_network
[12] Freedman JB, Marshack DS,
[4] Mesh Network. Wikipedia. Available et al. Advantages and capabilities of
from: https://en.wikipedia.org/wiki/ a beamforming satellite with a space-
Mesh_networking based digital processor. In: 32nd AIAA
International Communications Satellite
[5] Superheterodyne Receiver. Systems Conference (ICSSC); AIAA
Wikipedia. Available from: 2014-4321; San Diego, CA; 4-7 August
https://en.wikipedia.org/wiki/ 2014
Superheterodyne_receiver
[13] Komiyama N, Miura A, Orikasa T,
[6] Sadr R, Shahshahani M. On Sampling Fujino Y. Development of resource
Band-Pass Signals, TDA Progress Report allocation re-construction technology
42-96. Pasadena, California: NASA-Jet (digital beam former and digital
Propulsion Laboratory; 1988 channelizer). Journal of the National
Institute of Information and
[7] van der Veldt K, van Nieuwpoort R, Communications Technology. 2015.
Varbanescu AL, Jesshope C. A polyphase pp. 151-163
filter for GPUs and multi-core
processors. In: Proceedings of the [14] Nguyen TM, Guillen A,
Workshop on High-Performance Matsunaga S. Practical achievable
Computing for Astronomy; Delft, capacity for advanced SATCOM
Netherlands; 18 June, 2012. pp. 33-40 on-the-move. In: 2016 IEEE MILCOM
Conference Proceedings; 2016
[8] Kim SC, Bhattacharyya SS.
Implementation of a high-throughput [15] Guidotti A, Vanelli-Coralli A,
low-latency polyphase channelizer on Conti M, Andrenacci S, Chatzinotas S,
GPUs. In: EURASIP Journal on Advances Maturo N, et al. Architectures and
in Signal Processing; 2014. Available key technical challenges for 5G
from: http://asp.eurasipjournals.com/ systems incorporating satellites. IEEE
content/2014/1/141 Transactions on Vehicular Technology.
2019. pp. 2624-2639
[9] Ejima F, Akita M, Fujimura A. Digital
Channelizer for High Throughput

16
Chapter

Satellite Control System:

Part I - Architecture and Main
Components
Yuri V. Kim

Abstract

This chapter provides introductory material to satellite control system (SCS). It

is based on the author’s experience, who has been working in areas of SCS develop-
ment, including designing, testing, operating of real SCS, as well as reviewing and
overseeing various SCS projects. It briefly presents SCS generic futures and func-
tional principles: tasks, architecture, basic components and algorithms, operational
modes, simulation and testing. The chapter is divided into two parts, namely, Part I:
SCS Architecture and Main Components and Part II: SCS Simulation, Control
Modes, Power, Interface and Testing. This chapter focuses on Part I. Part II will be
presented as a separate chapter in this book.

Keywords: satellite control, attitude and orbit, determination, estimation, sensors,

actuators, coordinate systems, reference frame, state estimation and Kalman
filtering, earth gravity, magnetic fields

1. Introduction

Satellite control system (SCS) is a core, essential subsystem that provides to

the satellite capabilities to control its orbit and attitude with a certain performance
that is required for satellite mission and proper functioning of satellite payload
operation. However, the first mandatory task for SCS is assuring satellite safe
functionality; providing sufficient electric power, thermal and communication
conditions to be able for nominal functioning during specified life time at different
sun lightening conditions (including potential eclipse periods), protecting against
life critical failures proving to satellite safe attitude in Safe Hold Mode (SHM).
Without SCS or satellite guidance, navigation and control (GN&C) system, any
Earth-orbiting satellite could be considered just as artificial space body, demon-
strating the launcher capability for the satellite launch. As soon as a satellite is
assigned to perform a certain space mission, it has to have SCS and a kind of special
device (s)-payload (s), performing scientific, commercial or military tasks that are
dedicated to this mission. Today, the widespread satellite and SCS design philo-
sophy [1–3] is based on the concept that satellite is a platform (bus or transportation
vehicle) for the very important person (VIP) passenger, which is the payload, and
this platform is aimed just to deliver and carry it in space. This approach has been
proven as successful or, at least, satisfactory from the commercial point of view.
However, the first Soviet satellite “Sputnik” and further Soviet/Russian satellites

1
Satellite Systems - Design, Modeling, Simulation and Analysis

were built and launched under the different philosophy that satellite is the main
“personage” performing a space mission and the payload (unlikely the ballistic
rocket war head (s)) is just one of the satellite subsystems that should be integrated
into the satellite board under the satellite chief designer guidance, who is responsi-
ble for the mission performance. From the author’s point of view, this approach has
certain advantages following from the Aerospace System Engineering, integration
and distribution functions, and responsibilities between the space mission partici-
pants. In this chapter, SCS is presented from this point of view, integrating con-
ventionally separate satellite GN and C subsystems and devices into the joint
integrated system, attitude and orbit determination and control system (AODCS).
The main principles and features of this system are presented in this chapter.

2. Earth-orbiting satellites and the role of the control system

The first human-made Earth-orbiting satellite (Soviet Sputnik), Simplest Satel-

lite (SS-1), was launched on October 4, 1957. This satellite was launched following
the development of the Soviet intercontinental ballistic rocket R-7 (8 K71). Never-
theless, it started a new era of space human exploration (Figure 1).
SS-1 technical characteristics are as follows [4, 5]:

• Mass 83.6 kg; sealed from two identical hemispheres with a diameter of 0.58 m;
life time 3 months; payload, two 1 W transmitters (HF, 20.005 and VHF,
40.002 MHz) with four unidirectional deployable antennas (four 2.4–2.9 m
metallic rods); electrical batteries, silver-zinc; sufficient for 2 weeks.

• Orbit: perigee 215 km, apogee 939 km, period 96.2 min, eccentricity 0.05,
inclination angle 65.10 deg.

• Inside, the satellite sphere was filled by nitrogen, and the temperature was kept
within 20–23 deg. C with automatic thermoregulation-ventilation system
(thermometer-ventilator).

• The satellite had no attitude control and was free rotated around its center of
mass in orbit, keeping initial angular speed, provided by the separation pulse
after the separation from the launch rocket. However, thanks to the four rod

Figure 1.
Soviet designers-creators of the first earth-orbiting artificial satellite SS-1.

2
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

antennas that provided unidirectional radio transmission in the two-radio

bends, HF and VHF, SS-1 evidently indicated its presence in space for all
people over the world. Even amateur radio operators with amateur receivers
could receive famous now signals: BIP, BIP, BIP … !! (Figure 2).

Figure 2.
SS-1, assembled (left). Open two semispheres (right).

Since SS-1, about 8378 satellites were launched to year 2018 [6]. Early satellite
launches were extraordinary events and demonstrated tremendous achievement of
the launched state, the USSR (4 Oct. 1957, SS-1), the USA (31 Jan. 1958, Explorer 1)
and Canada (29 Sep. 1962, Alouette, launched by Thor-Agena, a US two-stage
rocket), but with time, satellite launches became ordinary and usually pursue a
certain military or civil mission.
Among the civil missions (satellites), the following types can be determined as
already conventional: navigation, communication, Earth observation, scientific,
geophysics and geodetic, technology demonstration and developers training. These
satellites are usually equipped with a kind of payload system(s) (radio/TV trans-
mitter/transducer, radar, telescope or different scientific instrument, etc.) to per-
form certain dedicated space mission(s). For example, the first Canadian Earth
observation satellite RADARSAT-1 (Nov 4, 1995–May 10, 2013; Figure 3) was
equipped with a side-looking synthetic aperture radar (SAR) on board the Interna-
tional Space Station (November 1998, ISS; Figure 4) was installed a Canadian
robotic arm for its assembling and maintenance.

Figure 3.
The first Canadian earth observation satellite RADARSAT-1.

According to the satellite altitude (h), their orbits can be classified as low-
altitude (LEO), 200–2000 km; medium-altitude (MEO), 5000–20,000 km; and
high-altitude (HEO), h > 20,000 km; according to eccentricity as: close to circular
e < 0.01; elliptical 0.01 < e < 0.3; highly elliptical 0.3 < e < 0.8.
There are satellites with special type of orbit such as polar (i = 90 deg), equatorial
geostationary (GEO, i = 0 and h = 35,800 km) and Sun-synchronous provide orbital
precession equal to Sun annual rate (i depends on satellite period) (Figure 5).

3
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 4.
International Space Station (ISS).

Figure 5.
Satellite orbit types (“tundra” and “Molniya” are Russian communication satellites in highly ecliptic orbits).

Miniaturized low-cost satellites are as follows: small satellites (100–500 kg),

microsatellite (below 100 kg) and nanosatellite (below 10 kg).
A large diversity of satellites serving for different missions is in space now. A
widespread point of vew is that all of them are transportation platforms delivering
and carrying in orbit dedicated to the planned space mission payload system, like a
VIP passenger. For example, it could be the postman for the postal horse carriage for
many years ago. Namely, the satellite with its control system (SCS) provides to the
payload all conditions required for the mission performance (orbit, attitude, power,
pressure, temperature, radiation protection and communication with ground mission
control center (MCC)). That is why from the mission integration point of view, the
SCS can be seen as the space segment integration bases that set their development and
operation process in corresponding order. In turn, SCS as satellite subsystem also can
be reviled and established in satellite onboard equipment architecture, combining the
group of subsystems that are dedicated to orbit and attitude determination and
control tasks. It could be done rather from the System Engineering than from the
commercial practice point of view and would significantly streamline satellite devel-
opment order and the degree of responsibility of all the developers.
It should be mentioned that such group of aircraft equipment in aviation has been
named as GN&C Avionics; hence, for space, it can be named as the Spacetronics, and
the heritage of system development and integration wherever it is possible should be
kept. Essential difference with Avionics for the Spacetronics is that it should work for
specified life time in space environment (dedicated orbit) after mechanical start-up
impacts (overload, vibration) connected to the launch into the orbit. The verification
of this capability is usually gained in special space qualification ground tests that
imitate launch impact and space environment with thermo-vacuum and radiation
chambers, mechanical load and vibration stands [7, 8].

4
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

3. Satellite control system architecture and components

3.1 SCS architecture

Today, for many satellites, GN&C onboard equipment can be presented by the
following subsystems, performing related functions listed below:

• Global Positioning System (GPS)—onboard satellite orbit and time

determination

• Propulsion system—orbit/attitude control system

• Attitude Determination and Control System (ADCS)—satellite attitude

determination and control

Integration of these subsystems can be named as attitude and orbit determination

and control system or Spacetronic system. Typically, AODCS includes the following
components:

• Onboard computer system (OBCS) or dedicated to AODCS electronic cards

(plates) in Central Satellite Computer System (e.g., command and data
handling computer (C&DH))

• Sensors

• Actuators

Basic AODCS architecture is presented in Figure 6.

OBCS, onboard computer system; TLM, telemetry data and commands; PL,
payload; PS, propulsion system; RW, inertia reaction wheels; MTR, magnetic
torque rods; GPS, satellite navigation Global Positioning System; MAG, 3-axis
magnetometer; SS, 2-axis Sun sensor; HS, horizontal plane sensor; ST, star tracker;
RS, angular rate sensor; EP, electric power; TR, temperature regulation; VP,
vacuumed protection; RP, radiation protection.

Figure 6.
Satellite AODCS system.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Depending on required reliability and life time, each component can be a single
or redundant unit. Unlike airplanes, satellite is an inhabitant space vehicle that is
operated from the ground. The operation is usually performed via a bidirectional
telemetry radio link (TLM) in S-band (2.0–2.2 GHz). Payload data downlink radio
link (unidirectional) is usually performed via X-band (7.25–7.75 GHz;). For both
links, usually the same data protocol standards are applied Figure 7.

Figure 7.
Satellite communications with ground stations.

Two subsystems can be allocated in AODCS architecture, namely, orbit deter-

mination and control subsystem (ODCS) and attitude determination and control
subsystem (ADCS). Practically both subsystems are dynamically uncoupled; how-
ever, orbital control requires the satellite to have a certain attitude (as well as orbital
knowledge itself), and attitude control requires orbit knowledge also. Hence, orbit
(its knowledge) is essentially continuously required on satellite board where it is
propagated by special orbit propagator (OP). Due to orbital perturbations (residual
atmospheric drag, gravity and magnetic disturbances and solar pressure), satellite
orbit changes over time and OP accumulates errors; its accuracy is degraded.
Before the application of satellite onboard GPS receivers, the satellite position
and velocity were periodically determined on ground by the ground tracking radio
stations (GS, dish antenna), and calculated on-ground orbital parameters were
periodically uploaded to satellite OBCS to correct OP, to provide available accuracy.
Now with GPS satellite, orbit can be calculated onboard autonomously, and OP can
propagate data only during relatively short GPS outage periods. For some applica-
tions, orbital data uploaded from the ground still can be used, at least, for fusion
with GPS-based OP.
For newly developed satellites with GPS, orbit maneuvers (correction, deorbiting,
collision avoidance, special formation flying and orbit servicing missions) can be
executed autonomously onboard at planned time or from ground operators using
orbital knowledge and TLM commands to activate satellite orbit control thrusters.

3.2 AODCS components

Below AODCS components are presented to show their generic principles that
can help for the system understanding and modeling. Generic design requirements
are presented in [3]. Some design examples can be found in many sources, for
example, [1, 9–12].

3.2.1 Sensors

AODCS sensors are designed to measure satellite orbital and attitude position
and velocity. From the most general point of view, they can be considered as the

6
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

vector measuring devices (VMD). The device can measure in space a physical
vector Rm that can be known (referenced) in a reference coordinate system Rr .
Three parameters can be measured: vector module R and two angles of its
orientation Az and El (Figure 8).

Figure 8.
Vector R in the Cartesian coordinate system XYZ.

Vector module and its orientation can be expressed as functions of its

projections Rx , Ry , Rz as follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R¼ R2x þ R2y þ R2z
1 Rx
Az ¼ tan
Ry (1)
1 Rz
El ¼ tan qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R2x þ R2y

It can be noted that measurement of referenced vectors can be used for the
determination of satellite position or angular orientation. A minimum of three vectors
is required to determine satellite position and two to determine its attitude. If more
vectors are measured providing informational redundancy, then such statistical esti-
mation methods as least square method (LSM) and Kalman filter (KF) can be applied.
Satellite velocity and angular rate can be derived by the differentiation of its position
and attitude applying a kind of filter recommended by the filtering and estimation
theory [13–15]. It should also be noted that if vector orientation is measured for the
position determination, then satellite attitude should be known and vice versa.
Especial autonomous satellite navigation system (sensor) is the inertial navigation
system (INS/inertial measurement unit (IMU)). It can be used for the determination
of satellite position, velocity, orientation and angular rate simultaneously. INS is
based on measuring with linear accelerometers and angular rate sensors (“gyros”) the
two vectors: satellite linear active acceleration a and angular rate ω. After integration,
the system provides satellite position, velocity, attitude and angular rate. It is also
assumed in INS theory that the vector of Earth gravity acceleration g is not measured
by the system accelerometers, but it is computed from referenced mathematical
Earth gravity field model. Essential INS disadvantage is that its errors grow with time.
That is why, it has to be periodically corrected by such navigation aids as a pair of
VMD used for the direct attitude determination. A detailed system description is out
of this chapter’s scope and can be found in many publications [16–18]. Only the use of

7
Satellite Systems - Design, Modeling, Simulation and Analysis

angular rate sensors (“gyros”) for determination of satellite attitude is briefly consid-
ered below.

3.2.1.1 Determination of satellite position and velocity (GPS)

Today, satellite GPS can provide onboard accurate data about position, velocity
and time [19] (Figure 9).
Accuracy: position, 15 m (2σ); velocity, 1.5 m/s (2σ); time, 1 μS.

Figure 9.
Satellite GPS SRG-10. Double redundant with a pair of zenith and nadir antennas.

GPS receiver is a radio range measuring device that measures distance from the
desired satellite to navigation satellite constellation (NAVSTAR, USA; GLONASS,
Russia; and GALILEO, Europe) and computes its position and velocity. GPS mea-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sures the distance R (R ¼ R2x þ R2y þ R2z ) of the vector from the desired satellite to
the navigation satellite, and this system is invariant of the system orientation (sat-
ellite attitude). The distance between the desired satellite and navigation satellite is
measured by measuring the time delay Δt between the time ts of the radio pulse
transmitted by navigation satellite and the reception time tr of its reception by GPS
receiver installed on the desired satellite Δt ¼ tr ts . Measuring the distance allows
to determine the desired satellite relative position (relatively to navigation satellite),
and using known navigation satellite position that is continuously received by the
receiver for every tracking satellite in the navigation message (NM) converts it in
absolute position.
A minimum of three navigation satellites should be simultaneously traced by the
receiver to determine position and velocity. Then satellite position is the cross-point
of three spherical surfaces of the position equation Ri ¼ const, i ¼ 1, 2, 3. If more
tracked satellites are available, then redundant information can be used to calibrate
the onboard clock (using the four satellites) and to use least square method or
Kalman filter. Four nonlinear algebraic equations (pseudo-range measurements)
are usually used to determine satellite position with GPS receiver:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i
R ¼ ðx xi Þ2 þ ðy yi Þ2 þ ðz zi Þ2 þ cτ
(2)
i ¼ 1, 2, 3, 4

where Ri is the distance (pseudo-range) to the ith tracked navigation satellite,

xi , yi , zi are the navigation satellite Cartesian coordinates received in the NM, x, y, z
are the Cartesian coordinates of the desired satellite, c ¼ 299 792 458 km=sis the
speed velocity and τ is the GPS receiver clock bias. Satellite position can be found by
solving it (1) numerically. It could also be linearized by using redundant

8
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

measurements (t > 4) with LSM or KF. Satellite velocity can be determined by the
differentiation of its position. Finally, GPS receiver can provide to AODC OBC
current satellite position and velocity in the reference (e.g., Earth-centered inertial
(ECI) [14] frame x, y, z, V x , V y , V z and synchronized (by GPS) onboard time ts ).

3.2.1.2 Determination of satellite attitude and angular rate

3.2.1.2.1 TRIAD method (MAG, SS, HS)

The TRIAD method [10] is applied when two different vectors are measured.
They usually can be any of the three pairs combined with the following three
vectors: Earth magnetic induction vector B (measured with three-axis MAG), Sun
vector S (measured with two-axis SS), and local vertical r (perpendicular to the
local infrared radiation temperature surface, measured with the HS). At least two
different not collinear vectors (their orientation) should be measured to deter-
mine satellite attitude that here is considered as satellite directional cosine matrix
(DCM) and related three Euler angles of the certain order of rotations (e.g., 3-2-1)
[9, 10] .
Let us assume that two different physical nature not collinear vectors U ¼ S and
V ¼ r are measured U m , V m by two vector measuring devises (SS and HS) installed
on the satellite board and both these vectors are referenced in the reference frame as
U r , V r . Let us choose U as the main vector and V as an auxiliary vector. Then an
orthogonal coordinate system (frame) with basis unit vectors, q, r and s can be
defined as follows [10, 20]:
!
! U
q¼ !
∣U ∣
! !
! UV (3)
r ¼ ! !
∣U  V ∣
! ! !
s ¼qr

These unit vectors expressed at a given time by measured values in measured

frame or body frame and reference values in a reference frame define two rotation
matrixes, Cm and Cr , as follows:
 
Cm ¼ qm ; rm ; sm
  (4)
Cr ¼ qr ; rr ; sr

where vectors q, r, s are written in the matrix form as matrix columns.

Rotation matrix Cbr that defines attitude in the body frame with respect to
reference frame is determined by the following formula:

Cbr ¼ Cm
Cr (5)

Three Euler angles of rotation, roll (ϕ), pitch (θ), and yaw (ψ), can be expressed
through the elements of the matrix Cbr . Certain trigonometric formulas depend on
the agreement about the order of the body rotations. For the order 3-2-1, the matrix
Cbr is as follows [9, 10]:

9
Satellite Systems - Design, Modeling, Simulation and Analysis

2 3
cθcψ cθsψ sθ
Cbr ¼ 4 sϕsθcψ cϕsψ sϕsθsψ þ cϕcψ sϕcθ 5 (6)
6 7

cϕsθcψ þ sϕsψ cϕsθsψ sϕcψ cϕcθ

where c and s stand for cosine and sine angle. Then, formulas for Euler angles
can be derived from (6) as:
C231
ϕ ¼ tan ,
C33
θ ¼ sin 1 C13 , (7)
C12
ψ ¼ tan 1
C11

Vector measured sensors

If a pair from the three vectors (B, S, r-write as vectors) is measured, then following
VMD in the pair can be used: SS (Figure 10), HS (Figure 11) and MAG (Figure 12).

Figure 10.
S-vector sensor Bradford fine sun sensor, accuracy, 0.2 deg. (2σ).

Figure 11.
r-vector sensor HS CMOS/SRAM-modular infrared horizon sensor, accuracy, 0.4 deg. (2σ).

Figure 12.
B-vector sensor MAG TFM100-S, accuracy, 10mG (2σ).

3.2.1.2.2 LSM method for star tracker (ST)

If more than two vectors are measured and available for attitude determination,
then LSM-BATCH method [10] can be applied to use informational redundancy for

10
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

increasing the stochastic estimation accuracy. This method basically can be applied
for any set of VMD but is specifically convenient for the star tracker (ST), when
some number (n) of navigation stars are in the device field of view (FOV) and are
detected and tracked simultaneously, providing measured vectors Rm to these stars
that are referenced in the device space catalog Rr (Figures 10–13).

Figure 13.
Star direction R-vector measured sensor (optic and computer units). Advanced stellar compass, accuracy,
200 1600 ð2σ Þ.

Let us consider the transformation of the referenced vector Rr in satellite body

frame, where it is measured with the ST
Rm ¼ CRr (8)

where C is the DCM of the rotation from the reference frame to satellite body
frame and vectors Rr and Rm are written in the matrix form as matrix columns.
If the ST is in the tracking mode keeping in its FOV some n detected navigation
stars, then it can be assumed that C is a small-angle matrix that is independent of
the rotation order and can be expressed as follows:
2 3
1 αz αy
C ≈ 4 αz
6 1 αx 75 (9)
αy αx 1

where αx , αy , αz are small angles of satellite rotation about X, Y, Z axis, respectively.

Then subtracting from (8) Rr , the following equation can be written:

δR ¼ δCRr (10)
2 3
0 αz αy
where δR ¼ Rm Rr , δC ≈ 4 αz
6 0 αx 75.
αy αx 0
Transforming in (10) matrix product and taking into account random measure-
ment errors, this equation can be represented in the following form:

δR ¼ Rr δC þ V (11)

Rrz Rry Vx
2 3 2 3 2 3
0 αx
2 3
δRx
where δR ¼ 4 δRy 5, Rr ¼ 4 Rrz 0 Rrx 7
5, δC ¼ α ¼ 4 αy 5, V ¼ 4 V y 5,
6 7 6 6 7 6 7
δRz Rry Rrx 0 αz Vz

11
Satellite Systems - Design, Modeling, Simulation and Analysis

V is random measurement error vector (considered as the white Gaussian noise,

having covariance matrix R ¼ rI).
Then this equation can be considered as a “standard” linear algebraic equation:

zi ¼ hi x þ Vi
(12)
i ¼ 1, 2, ::n number of measured vectors
2 3
Rrz Rry Vx αx
2 3 2 3
0
4 Rrz
where, zi ¼ δRi, hi ¼ Rri ¼ 6 0 Rrx 5 , Vi ¼ 4 V y 5 x ¼ δC ¼ α ¼ 4 αy 5.
7 6 7 6 7
Rry Rrx 0 i Vz i αz
If n ¼ 1, only a single vector is measured, then dethi ¼ Rrx Rry Rrz þ
Rrx Rry Rrz  0, and thus, determining all three angles of satellite attitude is impossi-
ble. If it takes informational redundancy, then optimal estimate (providing mini-
mum of standard deviation of satellite attitude errors) can be found with the
following LSM formula [13]:

^ ¼ Kz
α (13)
 1
where K ¼ HT R 1 H HT R 1 H, H ¼ ½h1 h2 h3 ::hn ŠT .


3.2.1.2.3 Determination of the angular rate

Direct measurement
Satellite angular rate ω (vector absolute angular velocity) can be measured directly
by three-axis rate sensor (RS) that could have mechanical, optical, or microelectro-
mechanical systems (MEMS) design [21, 22] (Figure 14). Traditionally, independent
of the design type, these RS are usually named “gyros,” paying respect to their
historical appearance for aerospace vehicle control purposes as a mechanical
gyroscope (Figure 15).
Measured angular velocity vector ω can be used to determine satellite attitude by
the integration of matrix kinematic Poisson’s Equation [9, 10]:
^
C_ ¼ ωC, Cð0Þ ¼ C0 (14)

Figure 14.
Measurement of satellite angular velocity ω with three rate sensors RSx , RSy , RSz .

12
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 15.
ω-vector sensor RS, BEI QRS-11 single-axis body rate sensor, accuracy, 7deg=h ¼ 0:0019deg=s ð2σ Þ.

where C ¼ Cib C ¼
2 Cib is the DCM between
3 the inertial frame ECI and satellite
0 ωz ωy
^
body frame and ω ¼ 4 ωz
6 0 ωx 75 is skew symmetric matrix measured in
ωy ωx 0
the satellite frame components of vector of satellite absolute angular velocity. After
determination of the DCM, satellite attitude in three Euler angles can be derived with
Eq. (7) above. Unfortunately, gyro drift causes unlimited growing up errors in
integrated attitude that require periodic corrections from two VMD, measuring the
attitude directly (Figures 10–12).
Body rate estimator
Often, specifically for attitude stabilization (keeping or aka pointing) mode,
satellite angular rate is estimated by using the so-called body rate estimator and is not
measured directly by the RS. Indeed, using for attitude keeping mode small angles and
linear approximation, we can simplify satellite attitude dynamics model [9] to three
single-axis state equations and present it with the stochastic influences as follows:
8
<ω
> _ ¼w
α_ ¼ ω (15)
>
z¼αþv
:

where ω is the angular velocity, α is the satellite deviation angle from the desired
direction and z is the satellite deviation angle measurement with random Gaussian
white noise error v. Realistically, it is a wide spectrum correlated process that
components have spectral density ri ¼ 2σ 2vi T vi , i ¼ x, y, z (σ vi is standard deviation of
the random error vi , T vi is vi ðtÞ correlation time), wi is exciting angular acceleration
noise with spectral density qi ¼ 2σ 2wi T wi (σ wi is standard deviation of the random
angular acceleration wi ¼ MJi i , Mi is exciting external random torque, J i is satellite
moment of inertia, T wi is wi ðtÞ correlation time).
The linear KF can be applied to synthesize the estimator for the optimal
estimation of the vector angle α and the vector of angular velocity ω, using noisy
measurements z [9]:
(
^_ ¼ k12 ðz α
ω ^Þ
(16)
^_ ¼ ω
α ^ þ k22 ðz α^Þ

where α
^ and ω
^ are the optimal estimates of the angle α and the angular velocity
ω correspondingly.

13
Satellite Systems - Design, Modeling, Simulation and Analysis

Matrix KF (16) is separated in three independent scalar channels for X, Y, Z axis.

Its weight coefficients k12 and k22 can be determined by solving KF Riccati Equation
[13–15] for each of these three separate channels independently.
It can be shown that in the considering case, the steady-state (t ! ∞) KF
coefficients are determined by the following formulas (identically for X, Y, Z axes,
i = 1,2,3):
( pffiffiffiffi
k12i ¼ ξi
pffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffi (17)
k22i ¼ 2ξi ξi

q
where ξi ¼ rii is the ratio of spectral densities of satellite disturbing torque noise
to measured attitude error noise (assuming that both are white Gaussian noises).
This parameter can be considered as the filterability index. Eq. (16) can be
represented in the transfer function (Laplace operator) form as a second-order
differential equation unit:
8 s
> ω
^ i ¼ 2
zi
T i s2 þ 2di T i s þ 1
>
>
<
(18)
> 2dT i s þ 1
^¼ zi
>
:α
>
2 2
T i s þ 2di T i s þ 1

where s is the Laplace operator, T i is the time constant, and di is the specific
damping coefficient that is determined by the following formulas:

1 1
8
> T i ¼ pffiffiffiffiffiffiffi ¼ p ffiffiffiffi
k12i
> 4
>
>
< ξi
pffiffiffi (19)
k 2
>
22
>
: di ¼ pffiffiffiffiffiffiffi
i
¼ ¼ 0:707
>
>
2 k12i 2

or in other words, the time constant is in inverse proportionality to the filter-

ability index (in ¼ degree) and the specific damping coefficient is conventional for
such a second-order unit 0.707 for each of the three channels.

3.2.1.2.4 Multisensory sensor unit (MSU)

As it can be seen from the consideration above, the use of directly measuring
devices (e.g., ST and RS) for attitude and body rate determination has a disadvan-
tage. The random noises are at the devices output, and they have to be filtered in the
closed control loop of satellite attitude control that puts some constraints to choose
the control law coefficients. However, using indirect body rate measurement, the
state estimator (filter) unavoidably introduces additional phase delay in the control
loop because of the consecutive inclusion of this filter in the control loop. To use the
RS (gyro) and the integrator for body rate and attitude determination autono-
mously for a long time is not possible because of the accumulated attitude errors
caused by the integration of the gyro drift. The following scheme (that is common
in Aviation) can be considered as free from the disadvantages above. Let us assume
that satellite attitude is determined in two ways: continuous integration of RS
angular velocity (IMU) and using VMD, for example, ST. Then this ST is used to
correct the attitude derived by the integration of RS output. The idea of MSU is
shown in Figure 16.

14
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 16.
Integration of multisensory sensor unit (MSU) single-axis channel.

In integrated IMU attitude (IMU = RS + integrator) as in Figure 14 above (three

identical channels), αi (inertial angle) has been growing with time deterministic error
δα due to the integration of RS bias δω and ST has random noisy error ΔαΔα. The
difference of these signals is equal to the difference of system errors z ¼ δα Δα. This
difference is used for estimation of errors of IMU with a kind of filter, and after the
compensation at the system output, the estimates (^ α and ω^ ) can be used for satellite
control. As it can be seen, in this scheme the filter is not connected in the control loop,
and consequently, it does not introduce additional phase delay; however, the scheme
still performs its job to filter the noise and estimate RS bias. This scheme can be very
effective in practice. It can be presented similar to Eqs. (15) and (16) as follows:

< δω_ ¼ w1
8
>
IMU model equations : δα_ ¼ δω (20)
>
z ¼ δα þ Δα
:

where δω is the RS error of measuring the angular velocity; δα is attitude error

after the integration of the RS signal; zis IMU attitude error measurement with
random Gaussian white noise error v Δα ¼ v, having spectral density r1 ¼ 2σ 2Δα T Δα
(σ Δα is the standard deviation of the random error Δα, T Δα is its correlation time);
and w1 is the exciting noise of RS random drift with spectral density q1 ¼ 2σ 2w1 T w1
(σ w1 is the standard deviation of the random drift, T w is the wðtÞ correlation time).
(
δω^_ ¼ k12 ðz δ^ αÞ
KF equations : (21)
^
δα_ ¼ δ^ω þ k22 ðz δ^
αÞ

where KF coefficients k12 and k22 are determined by (17), substituting there q1
and r1 instead of q and r.

3.2.2 Actuators

3.2.2.1 Propulsion system (PS)

Satellite propulsion system [9, 10] is usually designed for satellite orbital and/or
angular control. In the first case, PS is commanded from the ground OC by TLM
commands in some cases when satellite orbit has to be changed (orbit correction,
deorbiting, collision avoidance), in the second controlled automatically from
onboard AODCS. It consists of such typical elements as orbital and attitude
thrusters (number and installation scheme depending on certain application), pro-
pulsion tank with associated pipes, valves, regulators, and electronics. General
principles of PS act independently of the type (ion thrusters (0.01–0.1 N), liquid
propellant and solid motor (100–10,000 N), cold gas (1-3 N)).

15
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 17.
Satellite control with PS thruster principles.

Figure 15 illustrates the satellite control with PS thruster principles. The princi-
ple of the formation of the propulsion jet force can be presented by the following
equation of variable mass body dynamics that from Russian sources, for example,
[23], is known as Prof. I. Meshchersky’s equation:
dV dm
M ¼F Vp (22)
dt dt

where M is the mass body, F is the external force, V P is the propellant exhaust
velocity and m is the propellant mass. The term V p dmdt is the propulsion force
(propulsion thrust) (Figure 17).

dm
T¼ Vp (23)
dt

In Section 3.2.2.1.2, it is always dm

dt < 0, then T > 0. Usually for any propulsion
V
system, let us introduce parameter specific pulse Isp , where Isp ¼ gjTm_ j ¼ gp then V p ¼
Isp g m
_ and T ¼ Isp g m.
_ If the satellite thruster is installed such as satellite point of
the center of mass (CM) is located on the line of the action of the force T (red
thruster in Figure 17), then the thruster can serve for satellite orbit correction, and
the pulse of the control thrust causes increment of satellite velocity dV ¼ M1 Tdt. If
satellite thrusters are installed in such a way that having an arm L from the CM,
then they create rotating torque (green reversible thruster pair in Figure 17) and
can be used for attitude control (control torque is T r ¼ LT). And the increment in
attitude angle for the time dt of the pair of thruster activation will be dα ¼ TJr dt.
The expelled propulsion mass Δm can be calculated with the K. Tsiolkovsky
formula [9, 24] that follows from Eq. (22)
h dV i
Δm ¼ m0 1 e gIsp
(24)

where m0 is the propellant initial mass. Photos of I. Meshchersky and K.

Tsiolkovsky are presented (Figures 18 and 19).
Discrete pulse modulation control is usually used to minimize the consumption
of the propellant for attitude control [9]. Examples of the gas thruster and the tank
are presented in Figures 20 and 21.

3.2.2.2 Magnetic torque rods (MTR)

Magnetorquers are essentially sets of electromagnets. A conductive wire is

wrapped around a ferromagnetic core which is magnetized when excited by the
electric current caused by the control voltage applied to the coil. The disadvantage
of this design is the presence of a residual magnetic dipole that remains even when

16
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 18.
Prof. I. Meshchersky (1859–1935).

Figure 19.
K. Tsiolkovsky (1857–1935).

Figure 20.
Cold gas GN-2 thruster, nominal thrust 3.6 N (230 psi), specific impulse 57 s.

the coil is turned off because of the hysteresis in the magnetization curve of the
core. It is therefore necessary to demagnetize the core with a proper demagnetizing
procedure. Normally, the presence of the core (generally consisting of ferromag-
netic) increases the mass of the system. The control voltage is controlled by AODCS
control output (Figures 18–20). The magnetic dipole generated by the
magnetorquer is expressed by the formula:

M ¼ niS (25)

17
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 21.
60 liter propulsion gas tank.

where n is the number of turns of the wire, i is the current provided, and S is the
vector aria of the coil. The dipole interacts with the Earth magnetic field, generating
a torque whose expression is:

Tm ¼ M  B (26)

where M is the magnetic dipole vector moment, B is the Earth magnetic field
induction vector and T m is the generated magnetic torque vector. This equation in
the scalar form is as follows:

T mx Bz By Mx
2 3 2 32 3
0
7 6 B Bx 7
4 T my 5 ¼ 4 0 54 My 5 (27)
6 6 7
z
T mz By Bx 0 Mz

Typically, three coils are used; the three-coil assembly usually takes the form of
three perpendicular coils, because this setup equalizes the rotational symmetry of the
fields which can be generated; no matter how the external field and the craft are placed
with respect to each other, approximately the same torque can always be generated
simply by using different amounts of current on the three different coils (Figure 22).
It can be seen from Eq. (26) that MTR cannot generate the magnetic torque in
the direction that is parallel to Earth magnetic field B (MkB) and it always is
perpendicular to the Earth magnetic field vector T m ⊥B. Unfortunately, from
Eq. (27), the required magnetic moments cannot be found, because it has zero
determinant, and we cannot invert it.

Bz By 
 
 0

Δ ¼  Bz 0 Bx  ¼ Bx By Bz þ Bx By Bz ¼ 0 (28)
 
 By Bx
 
0 

However, the following approach can be used to find required vector M [9].
When we take cross-product of B with both sides of Eq. (26) and take into account
that it is useless to apply M in parallel direction to B, and hence we can require that
M⊥B and M
B ¼ 0, then the following formula can be derived:

18
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 22.
3D orthogonal magnetic torque rods.

B  Tm
M¼ (29)
B2

Another MTR control method is the so-called B-dot control [25].

M¼ kB (30)

where k is control gate coefficient.

As the result of (29) control, the satellite will reduce its body rate and is finally
slow rotated along Earth’s geomagnetic field line (vector B). Eventually it achieves
its threshold of capture by the gravity gradient effect.
If the redundancy is required, it is provided by additional (redundant) coil with
the same core (Figure 23).

Figure 23.
Magnetic torque rod SSTL MTR-30, magnetic moment, M ¼ 30 Am2 .

3.2.2.3 Reaction/reaction-momentum wheels (RW/RMW)

Reaction wheels (RW), aka momentum exchange devices [9] or reaction-

momentum wheels (RMW), have massive rotated rotor with big axial moment of
inertia with respect to the axis of rotation. They are electrically controlled by the
electric motors and the rotor is installed on the rotating motor shaft. The controlled
voltage, applied to the control winding of the motor, controls its rotor angular
speed. The product of the rotor angular acceleration Ω _ multiplied by its axial inertia
I is the RW generated inertia control torque Ti ¼ IΩ _ that is applied to the satellite
body in opposite to the acceleration direction. Special embedded angular speed

19
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 24.
Three orthogonal reaction wheels (RW).

sensor (tachometer) measures the motor rotor angular speed and allows organizing
RW inertia torque control proportional to the applied control voltage. At least three
RW, as in Figure 24, are required to produce control torque vector in three-
dimensional spatial, having desired value and pointed in the desired direction.
Sometimes redundant unit of three redundant wheels or one with 4th redundant
skewed wheel is applied to meet reliability requirements.
RW can generate control inertia torques Ti only when they are accelerated or
decelerated. With this torque, they cannot compensate a permanent disturbance
torque Td ¼ const applied to satellite for a long enough time and come eventually to
some maximum/minimum available speed aka the saturation speed. At that state T i
becomes zero. That is why RW is usually applied with other types of actuators such as
MTRs or gas thrusters to de-saturate the RW to use them as the source of Ti again.
In general case, RW can be run around in some nominal angular speed Ω0 . In
this case they can be named as reaction-momentum wheel (RMW) and then just
RW. Mathematically RW/RMW dynamics can be presented as follows. Let us con-
sider satellite with RW unit angular momentum:

Hs ¼ H þ h (31)

where H ¼ Jω is the satellite absolute angular momentum vector (column matrix),

J xx J xy J xz
2 3

ω is the vector of satellite absolute angular velocity, J ¼ 4 J yx J yy J yz 7

5 is the
6
J zx J zy J zz
satellite inertia matrix, h ¼ IΩ is the RMW relative angular3momentum vector, Ω is the
Ixx 0 0
2

RW relative rotation speed vector, and I ¼ 4 0 Iyy 0 5 is the RW inertia matrix.

6 7

0 0 Izz
Then differentiating (31) in rotating with angular velocity satellite axis ω and
using Euler’s rigid body dynamics formula [9], we can get the following equation:

h_
^ ^
_ þω
H H¼ ωh þ T (32)
2 3
0 ωz ωy
^
where ω ¼ 4 ωz
6 0 ωx 75 is the satellite angular velocity matrix and T is
ωy ωx 0
the vector of the external torque applied to the satellite. In the right side of Eq. (32),

20
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

we can see two terms that have meaning of torques applied from the RMW unit to
the satellite body: h_ ¼ Ti the inertial torque and ωh ¼ Tg the gyro torque. They
^

are RMW generated torques that can be used for the satellite attitude control.
Eq. (32) presents satellite attitude dynamics under the action of RMW torques. The
RW dynamics can be presented similar to Eq. (32):

h_ w þ ωhw ¼ Mw þ Mf
^
(33)

where hw ¼ h þ hc is the absolute RW momentum, hc ¼ Iω is the carrier

component of RW absolute momentum, QUOTE Mw is the RW motor control
torque, and Mf is the RW friction torque.
Eq. (33) can be represented in the following form:

h_ ¼ IΩ
^
_ ¼ Iω
_ ωðh þ hc Þ þ Mw þ Mf (34)

where Mf ¼ ke Ω Mdf sgn Ω is the friction torque, ke is the motor natural

damping coefficient, Mdf is the dry friction torque, and Mw is the control motor torque.
The torque Mf usually consists of two components: the viscous torque
(counter electromotive voltage in the control coil) ke Ω and dry or Coulomb
friction torque in the RMW bearings—Mdf sgn Ω. The control torque Mw can be set
as follows:

Mw ¼ ðkΩ þ ke ÞðΩ Ω0 Þ kT Tc (35)

where kΩ is the motor control damping coefficient, Ω0 is the desired angular

RMW speed, kT is the control torque coefficient, and Tc is requested from AODCS
OBC control torque. Taking into account Eq. (35), Eq. (34) can be rewritten in the
following form:
^
_ þ ðkΩ þ ke ÞΩ ¼ ðkΩ þ ke ÞΩ0 þ Tc
IΩ Iω
_ ωðh þ hc Þ þ Mf (36)

In the operator Laplace s-form (transfer function), Eq. (36) can be rewritten as
follows:

ΩðsÞ ¼ WðsÞ Ω0 þ kw1 Tc

  ^ 
Isω ωðh þ hc Þ þ Mf (37)
2 3
1
ðT wx þ 1Þ 0 0
1
where WðsÞ ¼ 6
6 7
T wy þ 1
  7 is the RW matrix
4 0 0 5
0 0 ðT wz þ 1Þ 1
transfer function, T wx ¼ kΩxIþk
x
ex
, T wy ¼ kΩyIþk
x
ey
, T wx ¼ kΩzIþk
x
ez
is the RW time con-
stants, and.
2 3
ðkΩx þ kex Þ 1 0 0
6  1 7
kΩy þ key
 7 is the RW control speed
kw ¼ 6 4 0 0 5
1
0 0 ðkΩz þ kez Þ
coefficient.
It should be noted that sometimes the RW control loop is more sophisticated.
Special integrators could be connected into the loop to memorize and compensate
the dry friction torques acting in bearings. Some small nominal rotating speed can
be set for all three RMW to eliminate the dry friction torque having a pike when the
wheel speed is zero Ω ¼ 0.

21
Satellite Systems - Design, Modeling, Simulation and Analysis

However, more representative case is when Ω0 is set (usually in one direction as

in Figure 23—axis Y) to provide to the satellite a gyro-stiffness (gyro-stabilization
capability) with respect to another two axes X and Z.
_x ¼Ω
Indeed, if we put in (32) that ω_ x ¼ ω_ y ¼ ω_ z ¼ Ω _y ¼Ω
_ z ¼ 0, and ωx ¼ ωz ¼
Ωx ¼ Ω _y ¼Ω _ z ¼ 0, ωy ¼ ω0 ¼ const, Ωy ¼ Ω0 ¼ const, then it can be written as:

ωz H y þ h y ¼ T x
(  
(38)
ωx H y þ h y ¼ T z
 

where Hy ¼ J y ω0 , hy ¼ Iy Ω0 . In this case carrier orbital angular velocity ω0 helps

to increase satellite momentum Hsy ¼ Hy þ hy . Eq. (38) is known as three degrees of
freedom of a free gyroscope precession [21] and represents the gyro-stabilization
effect: that gyroscope vector Hsy being free of disturbing torques can keep its
direction and value in inertial spatial (Figure 25).

Figure 25.
Reaction/momentum wheel HR-0610, torque, 75
10 3 Nm; momentum, ð4 12Þ Nms.

3.2.3 AODCS OBCS

Independently of sytem arhitecture; it is separate dedicated to AODCS com-

puter, or a special AODCS card within central satellite C&DH computer, it is the
integration element of AODCS [1, 11]. AODCS system may consist of the computer
(computer card) itself (OBC) and auxiliary intercommunication electronic units
(electronic cards) AEU carrying DC/DC electric power conversion and I/O (analog
and digital) interface and commutation functions.
OBC can be divided into two parts: the hardware (HW, power convertor, pro-
cessor, input/output [I/O] convertors, non-volatile and volatile memory) and the
software (SW, operation system [OS] and vital or functional software [VS/FS])
(Figure 26).
What makes the satellite OBC essentially different for the airplane OBC is that
its SW can be uploaded and updated from the ground and during operation and
scheduled maintenance. OS OBC includes generic computer programs: program of
I/O interface, time schedule (dispatcher), embedded test, timer and standard
mathematic functions. Satellite SW often is considered as satellite SW subsystem that
is verified during development (with mathematical high-fidelity Matlab/Simulink
simulators and semi-natural processor-in-the-loop (PIL) simulators). SW
subsystem should be tested to meet SW requirements [26, 27]. The flight version of
the SW subsystem is supported with operation real-time satellite simulators (RSS)
[1, 11] located in operation center. It should be mentioned that only final AODCS

22
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 26.
Satellite OBCS, MAC-200 (C&DH unit with AODCS card) comprises of two OBC: Prime and redundant
(cold reserve).

(OBC (HW + SW), sensors, and actuators) functional test [2] that should be
performed in the Space Qualification Laboratory [7] during satellite Space Qualifi-
cation and Acceptance campaign can really minimize the risk of launching a not
ready satellite and prevent against AODCS refinishing in orbit during commission-
ing and operation.
VS can be separated in two parts, ODCS SW and ADCS SW. For both parts, I/O
interface with sensors and actuators is determined in special interface control
document(s) (ICD), describing type, certain connectors, and electrical parameters
of the exchanging data. These data before using them for functional tasks are
pre-processed in OBC with special algorithms.

3.2.3.1 Satellite sensors/actuators data preprocessing

This group of algorithms performs the following common tasks:

• Convert data into required physical parameters and units, taking into account
certain sensor input–output scale function.

• Transform data in certain device frame and compensate device misalignment,

bias and scale function errors if it is possible, monitor device state, establishing
“on/off,” “work/control,” “data bad/good” flags.

• Transfer to C&DH TLM data about sensor/actuator state and their data.

• Perform some other auxiliary functions if they are required.

• Main functional tasks ODCS SW and ADCS SW can be listed as below.

3.2.3.2 ODCS SW

3.2.3.2.1 Satellite orbit propagation (OP)

To understand the idea of propagation of satellite orbit in Earth gravity field to

the simplest, Keplerian motion propagator based on spherical Earth gravity field
model might be used [9]; however more realistic results can be obtained with more
accurate propagator, taking into account the second zonal harmonic J 2 in the
function of approximation of Earth gravitational potential. The following
equations of motion of satellite center of mass in Earth gravitational field can be
considered [9]:

23
Satellite Systems - Design, Modeling, Simulation and Analysis

x xz2 x
x
€¼ μ þ AJ2 15 3 ,
r3 r7 r5
y yz2 y
€y ¼ μ þ A J2 15 3 ,
r3 r7 r5
(39)
3
z z z
€z ¼ μ þ AJ2 15 7 9 ,
r 3 r r5
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r ¼ x 2 þ y2 þ z 2 ,

where x, y, z are the Cartesian coordinates of satellite center of mass in inertial

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
frame ECI, r ¼ x2 þ y2 þ z2 is the module of the radius vector from the center of
Earth to satellite center of mass, μ ¼ 3:986004418
1014 ½m3 =s2 Š is the Earth gravi-
tational constant, AJ2 ¼ 21 J 2
R2e is a constant, J 2 =0.00108263 is the second zonal
harmonic coefficient in the raw of Earth potential function, and Re ¼ 6378137:00 m
is the mean radius of the Earth at the equator.
These equations can propagate satellite position and velocity (x, y, z and x, _ y_ , z_ )
in the inertial Cartesian ECI coordinate system if the initial parameters are initially
set x0 ¼ xð0Þ, y0 ¼ yð0Þ, z0 ¼ zð0Þ and x_ 0 ¼ x_ ð0Þ, y_ 0 ¼ y_ ð0Þ, z_ 0 ¼ z_ ð0Þ. They can be
periodically determined from GPS or MCC TLM information. The propagation
credibility time depends on orbit perturbations [9, 10, 27] and required accuracy.
The most accurate and common ground propagators are NORAD Simplified Per-
turbation Model (SGP) propagators. NORAD SGP is used for proving to users two-
line element (TLE) satellite orbital data. For the low Earth orbit (LEO), having
altitude below 6000 km (period about 225 min), they provide position accuracy
about 1 km within a few days that for many users is accurate enough and needs to
be updated once or twice per week. Currently almost every satellite is equipped
with GPS and its onboard propagators are practically continuously corrected with
GPS (and sometimes MCC TLM) data that provide position within 10–100 m and
velocity within 0.01–0.1 m/s accuracy range. Only some short periods of GPS data
outage require orbit propagation. In addition to satellite position and velocity, OP
calculates conventional orbital parameters (Figure 14) that can be computed with
the following formulas [9]:
μ
a¼
μ V2
2
r 2
hzi
i ¼ cos ‐1
h
hxi
Ω ¼ tan 1 (40)
hyi
r
V
1
ν ¼ tan
p r
zi
u ¼ sin 1
r sin i
ω ¼ u‐θ

where h ¼ r  V is the satellite orbital linear momentum vector, a is the satellite

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
orbit semi-major axis, h ¼ h2xi þ h2yi þ h2zi is the linear momentum module, p ¼
að1‐e2 Þ is the satellite orbit focal parameter, i is the satellite orbit inclination angle, Ω

24
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

is the satellite orbit right ascension of ascending node angle (RAAN), uis the
argument of latitude angle, ν is the satellite true anomaly angle, and ω is the satellite
orbit argument of perigee angle (Figure 27).

Figure 27.
Satellite orbit in the inertial ECI (XYZ) coordinate system.

3.2.3.2.2 Orbital thrusters control

If orbit maneuver is required, then it can be commanded by AODCS SW auton-

omously, or special control commands TLM (uploaded command tables) are send-
ing to satellite AODCS, and in predetermined time they are executed activating at
scheduled time for the calculated period ðΔtÞ the orbital thrusters that provide for
the required orbital correction/maneuver pulseðFΔtÞ.

3.2.3.3 ADCS SW

3.2.3.3.1 Satellite attitude and angular velocity estimation algorithms

This group of algorithms was presented above in 3.2.1.2 and can be used here.
For example, let us consider single-axis stabilized satellite that should keep one
axis (e.g., Z) permanently pointed to the Sun as in Figure 28. Only two angles of the
satellite deviation from this direction and their angular velocities are required to
know to point and keep it in this direction. The satellite has two-axis Sun sensor that

Figure 28.
Satellite pointed by the Z-axis to the sun.

25
Satellite Systems - Design, Modeling, Simulation and Analysis

can measure two angles αx , αy of satellite deviation from the sun direction. Its axes
coincide with the satellite axes xyz. The axis z is the sensitivity axis that nominally
should point into the Sun’s direction (center of brightness), and xy is the focal plane.
The Sun vector is referenced in the Sun frame as Sr ¼ ½0 0 SŠT and is measured in
T
the Sun sensor measured frame as Sm ¼ Sx Sy Sz .


In Figure 28, xyz is the satellite body frame, xs ys zs is the Sun reference frame, SS
is the two-axis Sun sensor and αx and αy are the turn angles of satellite x and y axis
accordingly.
The following formula represents the mathematical transformation of the Sun
vector from the reference into the body frame:

Sm ¼ Cbs Sr (41)
T
where Sr ¼ ½0 0 SŠT , Sm ¼ Sx Sy Sz , and Cbs is the DCM between the refer-


ence (Sun) and measured (satellite) frames. Let us consider that the order of
rotation from the Sun to the satellite frame is 3-2-1 (αz , αy , αx); the DCM matrix Cbs is
as follows [9]:
2 3
cos αy cos αz cos αy sin αz sin αy
Cbs ¼ 4
6 cos αx sin αz þ sin αx sin αy cos αz cos αx cos αz þ sin αx sin αy sin αz sin αx cos αy 7
5
sin αx sin αz þ cos αx sin αy cos αz cos αx cos αz þ cos αx sin αy sin αz cos αx cos αy
(42)

Then from (41), (42) can derive the following formulas:

< Sxm ¼ S sin αy

8
>
Sym ¼ S sin αx cos αy (43)
Szm ¼ S cos αx cos αy
>
:

From (43), desired angles and can be derived that can be used for satellite
attitude control.

Sym
8
αx ¼ tan 1
>
>
Szm
>
>
<
1 Sxm (44)
>
> α y ¼ sin q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
S2 þ S2
>
>
ym zm
:

3.2.3.3.2 Angular velocities (body rates)

Let us also assume that the satellite does not have angular velocity sensors RS and
its angular velocities should be derived from the measured angles αx and αy . Simple
low-frequency first-order differentiating fitters can be applied for this purpose.
Laplace operator s-form (transfer functions) of these filters are presented below:’
8
^ s
< α_ x ¼ T s þ 1 αx
>
>
fx
s (45)
: ^_y ¼
>α αy
T fy s þ 1
>

where T fx , T fy are filter time constants, typically, T f < ð3 10Þ s.

26
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

3.2.3.3.3 Satellite control algorithms

If not the optimization criterion to characterize the control quality [13] is

required, then conventional negative feedback closed control loop with linear PID
(proportional, integral, and damping) control law [9] that provides a good perfor-
mance for many practical satellite control applications can be used to satisfy the
requirements. They are typical for any automatic control system requirements: such
as transfer process decay time and overshooting, residual static error caused by the
permanent external disturbance, etc. Today, attitude control system performance
can be verified mainly on ground with simulation. If we try to evaluate it in flight,
then only onboard attitude sensors TLM data can be used for postprocessing, and it
should be taken into account that mainly sensors that detect high-frequency noise
(perceived errors) will be observable and low-frequency components (sensor
biases) are compensated in the closed control attitude stabilization loop. Simple
example of single-axis satellite attitude stabilization control loop is presented below.
It is a simplified linear model; however, it presents the stabilization principle and
essential features. Let us assume that a simple, positional, and damping control law
is used to stabilize satellite axis Z in Sun direction S as in the example C1 for attitude
and angular rate determination above. Let us assume that only the one-axis control
channel X is considered; small angle αx ð tan αx ≈ αx Þ is measured with SS (44), and
its rate is derived with linear differentiating filter (45). Then requested PD control
torque can be presented by the following formula:

s
T cx ¼ kpx αxm kdx αxm (46)
T fx þ 1

where kpx is the position control coefficient, kdx is the damping control coeffi-
cient, s ¼ dtd is the Laplace operator, T fx is the differentiating filter time constant,
αxm ¼ αx þ Δαx is the measured angle αx ,αx is the true value and Δαx is the
measured error. Let us assume that this torque is generated by only one MTR Y.
Eq. (27) is determined as:

T cx ¼ Bz My ¼ Bz kTRy uy (47)

where Bz ¼ const is the local Z component of Earth magnetic induction vector,

My ¼ kTRy uy is the Y MTR magnetic moment, kTRy is the Y MTR control gate and iy
is the control voltage applied to Y MTR winding. Then as it follows from (46), (47)
requested from the AODCS OBC control voltage to the winding of Y MTR is:

s
uy ¼ K px αxm K dx αxm (48)
T fx þ 1

k
where K px ¼ Bz kpxTRy and K dx ¼ BzkkdxTRy are position and damping magnetic control
coefficients.
Let
2 us take a 3ball-shaped satellite with the inertia matrix as follows:
Jx 0 0
J ¼ 4 0 J y 0 5 where J x ¼ J y ¼ J z J x ¼ J y ¼ J z . Then in inertial spatial, its linear
6 7

0 0 Jz
angular dynamical equations for the axis X can be approximately written as follows:

J x s2 αx ¼ T cx þ T dx (49)

27
Satellite Systems - Design, Modeling, Simulation and Analysis

where s ¼ dtd is the Laplace operator, T cx is the control torque, and T dx is the
disturbing external torque (satellite residual and induction magnetism torque,
atmosphere drug torque, solar pressure torque). Then substituting in Eq. (49)
Eqs. (47) and (48), we can rewrite it as follows:
s
J x s2 αx ¼ kpx αxm kdx αxm þ T dx (50)
T fx s þ 1

Let us divide all terms in Eq. (50) by the coefficient kp and substitute αxm value,
then it can be represented in the following form:
!
s 1
T 2x s2 þ 2dx T x þ 1 αx ¼ T dx Δαx (51)
T fx s þ 1 kpx
qffiffiffiffiffi
Jx
where T x ¼ kpx is the X control channel time constant and dx ¼ pkffiffiffiffiffiffiffi
dx ffi
is the
2 kpx J x
Xcontrol channel-specific damping coefficient. Eq. (51) is a third-order linear differ-
ential equation and could be analytically analyzed. In particular its stability can be
analyzed with algebraic Hurwitz criterion [13]. However, more simple and general
results can be obtained with the following approximate consideration. If T fx < < T x ,
then by the filter time constant, T fx can be neglected, and (51) can approximately be
considered as a standard second-order control unit, presented by the second-order
linear time invariant (LTI) differential equation and rewritten as follows:

1
T 2x s2 þ 2dx T x s þ 1 αx ≃ T dx
 
Δαx (52)
kpx

As it follows from Eq. (52), steady-state error in attitude stabilization can be

calculated with the formula:

1
T 2x s2 þ 2dx T x s þ 1 αx ≃ T dx
 
Δαx (53)
kpx

where T dx0 ¼ const and Δαx0 ¼ constΔαx0 ¼ const (ATT sensor bias).
pffiffi
For Eq. (52), the optimal damping coefficient is dx ¼ 22 ¼ 0:707 [13].
Numerical example
Let us evaluate satellite time constant T x. Let us assume that for the LEO satellite
magnetic field induction vector B has the following value of the projection on the
Sun direction.
and MTR has the following parameters: maximal magnetic moment My max ¼
35Am2 , maximal control current Iy max ¼ 100 mA, winding resistance R ¼ 280 Ohm;
My max
and maximal control voltage uymax ¼ R
Iy max ¼ 28 V. MTR gate is kTRy ¼ uy max ¼
2
35Am
28V ¼ 1:25Am2 =V. Then maximal available magnetic torque is T cx max ¼ 10 Nm. If 3

maximal linear zone for this control channel is αy max ¼ π2 rad, then the position
control coefficient kpx can be calculated with the following formula:

T cx max
kpx ¼ (54)
αmax
0:001
For the data above, it has the value of kpx ¼ 3:14
0:5 ¼ 6:366
10 4 Nm=rad.
Let us consider example of the first Soviet satellite “Sputnik” (SS-1) that had the

28
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

mass m ¼ 83:6 kg and the radiusR ¼ 0:29 m and take the assumption that its
mass was homogeneously distributed within its spherical volume of V ¼ 43 πR3 ¼
0:1 m3 , and then its inertia ( J ¼ J x ¼ J y ¼ J z Þ can be calculated with the following
formula:

2
J ¼ mR2 (55)
5

as for a homogeneous sphere. Substituting into Eq. (55) the data above, we can
calculate that for SS-1 J ¼ 2:82 kgm2 , then its time constant with the considered
MTR control might be:
sffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Jx 2:82kgm2
Tx ¼ ¼ ¼ 66:56 s:
kpx 6:366
10 4 kgm2 s 2

Now damping coefficient can be calculated with the following formula:

kdx ¼ 2dx T x kpx (56)

4
It has the following value: kdx ¼ 2
0:707
66:56
6:366
10 s
Nm=rad ¼
0:06 Nms=rd.
k
Finally, K px and K dx can be calculated. They are as follows: K px ¼ Bz kpxTRy ¼
6:366
10 4 Nm=rad
2:86
10 5
1:25T
A
m2 =V
¼ 17:8 V=rad and K dx ¼ BzkkdxTRy ¼ 2:86
10
0:06Nms=rad
5

1:25T
A
m2 =V
¼ 1:678

3
10 sV=rad.
When T x is determined, then the time constant for the differencing filter T fx can
be chosen from the condition that T fx < < T x . In our example T x ¼ 66:56 s, let us
take that T fx ¼ 5 s.
Simulation
Eqs. (51) and (52) were simulated using Simulink (see Figure 29).

Figure 29.
Satellite single-axis attitude control Simulink block scheme.

29
Satellite Systems - Design, Modeling, Simulation and Analysis

Blocks in the pink color present the satellite model, the dark green color is for
control law blocks, the cyan blocks are registration oscilloscopes, and the display and
the orange color are the disturbances. The red manual switch allows to implement the
differentiating filter, transforming the scheme from the approximation (52) to the
accurate presentation (51). Disturbing external torque Md is constant; attitude sensor
error is represented by the constant value ALP0 and limited range white noise V that
has spectral density SV ¼ 2σ 2V T v (σ v is the standard deviation (SD), T v is the
correlation time). The results of the simulation (Figure 29) with and without the
differentiating filter (with the assumption that α_ x α_ x is directly measured without any
errors) are presented below in Figures 30–34 (left A, (52), without the filter; right B,
(51), with the filter). The numerical data for the simulation are as follows:
sffiffiffiffiffiffiffi
Jx kdx
J ¼ 2:82 kgm2 , T x ¼ ¼ 66:56 s, dx ¼ ¼ 0:707,
kpx 2T x kpx
4
kpx ¼ 6:366
10 Nm=rad, kdx ¼ 0:06 Nm
s=rad, T fx ¼ 5 s,
Md ¼ 10 5 Nm, Δαx0 ¼ dltALP0 ¼ 0:1∘ , Δαx0 ¼ dltALP0 ¼ 0:1∘

Simulation of ACS (Figure 29) is presented in Figures 30–34. Units: vertical

axis (deg), horizontal axis: (s).

Figure 30.
Response to initial deviation angle αx0 ¼ 10 . (a) without dif. filter and (b) with dif. filter.

Figure 31.
Response to initial angular velocity α_ x0 ¼ 0:01deg=s.

Figure 32.
Response to external disturbance torque T d ¼ 10 5 Nm. Static error αx∗ ¼ 0:898∘ .

30
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

Figure 33.
Response to attitude sensor bias Δx0 ¼ 1∘ plus white noise σ V ¼ 0:1∘ , T V ¼ 1 s. Satellite attitude stabilization
errors, ALP.

Figure 34.
Satellite attitude measured errors ALPm.

Decay time: τ ¼ 195s, αx ð3τÞ ¼ 5%αx0 (deg), (s) τ ¼ 275s.

As it can be seen, measured noise is filtered effectively in the control loop, and
stabilization error is equal to the sensor bias with opposite sign.
In Figure 34, we can see that the measured (perceived) errors that TLM data
provide to ground after the decay time do not present sensor bias and present only
measured noise. It is because satellite stabilization error with opposite sign com-
pensates the bias. In general, it can also be seen that the simulation of the approxi-
mate second-order model (52) is very close to the accurate model (51). Hence, at
least for the analytical representation, (52) can be successfully used.

4. Conclusion

Part I of this chapter presents an overview of practical satellite control system,

satellite guidance, navigation and control equipment. The work presented here is
based on the author’s point of view of integration of this GN&C equipment in the
integrated AODCS system (satellite GN&C Spacetronics System). Main work prin-
ciples, architecture, and components of the satellite control system were briefly
highlighted.
The chapter can serve to a wide pool of space system specialists as an introduc-
tion to satellite control system development.

Acknowledgements

The author wishes to express his sincere gratitude to the Canadian Space
Agency, where he had the opportunity to learn and possess the knowledge and
experience related to the writing of this chapter. As well, he is very thankful to
many of his colleagues from Canadian Magellan Aerospace Company (Bristol Aero-
space Division) with whom he discussed and analyzed satellite AODCS design pro-
jects and issues that helped him to work out the system analysis and its principal
concepts presented in this chapter. Additionally, he cannot forget that his

31
Satellite Systems - Design, Modeling, Simulation and Analysis

experience and background in Aerospace Technology were also accumulated from

the former USSR (Moscow Aviation Institute, Moscow Aviapribor Corporation,
Moscow Experimental Design Bureau Mars, Institute in Problems in Mechanic of
RAN) and Israel (IAI, Lahav Division and Tashan Engineering Center), where he
could observe and learn from diverse and wealthy engineering and scientific schools
led by great scientists and designers such as Prof. BA. Riabov, Prof. V.P. Seleznev,
V.A. Yakovlev, G. I. Chesnokov, V.V. Smirnov, Dr. A. Syrov, Acad. F. Chernousko,
A. Sadot and Dr. I. Soroka.
This chapter was written as a solo author since his friend and regular coauthor
Prof. George Vukovich from York University of Toronto passed away 2 years ago.
For many years, Prof. Vukovich served as Director of his Department of Spacecraft
Engineering in CSA. He will always keep good memories of Prof. Vukovich who
helped and encouraged him continue his scientific and engineering work.
The author also acknowledges the copyrights of all publishers of the illustrations
that were extracted from the open sources in the Internet.

Note

Dedicated to Prof. G. Vukovich.

Author details

Yuri V. Kim
David Florida Laboratory, Canadian Space Agency (CSA), St. Hubert, Ottawa,
Canada

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

32
Satellite Control System: Part I - Architecture and Main Components
DOI: http://dx.doi.org/10.5772/intechopen.92575

References

[1] Eickhoff J. On Board Computers, on [11] You Z. Space Microsystems and

Board Software and Satellite Operation. Micro/Nanosatellites, National Defence
Berlin: Springer-Verlag; 2012 Industry Press. NY: Elsevier Inc; 2018

[2] Space Systems, Design, Qualification [12] Lee J, de Rouiter A, Ng A,

and Acceptance Tests of Small Lambert C, Kim Y, Yoshihara K.
Spacecraft and Units. International Attitude Determination and Control
Standard, ISO 19683; 2017 Subsystem of JC2Sat-FF Mission. In:
Proceedings of 12th International
[3] Space Engineering, Satellite Attitude Aerospace Conference ISCOPS in
and Orbital Control System (AOCS) Montreal, July 2010, AAS 10-427; 2010
Requirements, ECSS-E-ST-60-30C;
2013 [13] Bryson AE, Yu-Chi Ho J. Applied
Optimal Control. Levittown, PA: Taylor
[4] Development of First Artificial & Francis; 1975. pp. 364-373, 369, 457
Satellites (Rus.). Available from: https://
www.kik-sssr.ru/Hist_1_PS-1_books. [14] Fourati H, Belkhiat D. Multisensor
htm [Accessed: 15 January 2020] Attitude Estimation. Fundamental
Concepts and Applications. NW:
[5] Available from: https://www.space. CRC Press, Taylor & Francis Group;
com 17563-sputnik. Sputnik: The Space 2017
Race’s Opening Shot [Accessed: 17
January 2020] [15] Kim YV. Kalman filter
decomposition in the time domain using
[6] Chaturvedi A. Do You Know How observability index, IFAC, Seoul, July 6–
Many Satellites Are Currently Orbiting 11, 2008. In: International Conference
Around the Earth? Available from: Proceedings. Vol. 2. NY: Curran
https://www.geospatialworld.net/blogs/ Associates Inc.; 2009. pp. 625-630
do-you-know-how-many-satellites-
earth/ [Accessed: 16 January 2020]
[16] Seleznev VP. Navigation Devises,
Rus. Moscow: Mashinistroenye; 1974
[7] About the David Florida Laboratory-
Canada.ca, Available from: https://
[17] Chatfield A. Fundamentals of High
www.asc-csa.gc.ca/eng/laboratories-
Accuracy Inertial Navigation. Reston,
and-warehouse/david-florida/about.asp
VA: AIAA; 1974
[Accessed: 17 January 2020]

[8] Delchamos T, Jonason G, Swift R. [18] Titterton D, Weston J. Strapdown

The Spacecraft Test and Evaluation Inertial Navigation Technology. Reston,
Program, Telestar I, NASA SP-32. VA: The Institute of Engineering and
Washington, DC; 1963. p. 1007 Technology; 1996

[9] Sidi M. Spacecraft Dynamics and [19] Hofmann-Welenghof B,

Control. A Practical Engineering Lichtenegger H, Collins J. GPS Theory
Approach. Cambridge: Cambridge and Practice. New York: Springer-
University Press; 1997 Verlag Wien; 2001

[10] Wertz JR, editor. Spacecraft [20] Shuster M, Oh S. Three-axis attitude

Attitude Determination and Control. determination from vector observations.
Dordrech: Kluwer Academic Publishers; Journal of Guidance & Control. 1981;
1978 4(1):70-77

33
Satellite Systems - Design, Modeling, Simulation and Analysis

[21] Armenise M, Ciminelli C,

Dell’Olio F, Passaro V. Advances in
Gyroscope Technologies. Berlin:
Springer-Verlag; 2010

[22] Arnold R, Maunder L.

Gyrodynamics and its Engineering
Applications. London: Academic Press;
1961

[23] Meshchersky IV. Dynamics of a

point of variable mass. In: Work on the
Mechanics of Bodies of Variable Mass.
2nd ed. GITTL, Moscow; 1952.
pp. 37-188, 280

[24] A Man- and an Equation, ESA.

Available from: http://blogs.esa.int/
rocketscience/2012/10/14/a-man-and-
an-equation/ [Accessed: 19 January
2020]

[25] How to Slow Down Satellite

Rotation? D-dot Algorithm, ESE 3 Sat
Project. Available from: http://www.
ece3sat.com/blog/2018-01-25-how-
to-slow-down-rotation-the-bdot-
algorithm/

[26] Space Engineering-Software,

ECSS-E-ST-40C, Noordwijk; 2019

[27] Vukovich G, Kim Y. Satellite orbit

deacay due to atmospheric drag.
International Journal of Space Science
and Engineering. Inderscience
Publishers; 2019;5(2). Availabe from:
www.inderscience.com/jhome.php?/
jcode=ijspacese. [Acessed: 4 June 2020]

34
CORRIGENDUM Chapter

Satellite Control System

Yuri V. Kim

David Florida Laboratory, Canadian Space Agency (CSA), St. Hubert, Ottawa, Canada

Corrigendum to: Kim YV. Satellite Control System: Part 1-Architecture and Main
Components. In: Nguyen TM, editor. Satellite Systems – Design, Modeling,
Simulation and Analysis. London: IntechOpen; 2020. DOI: 10.5772/intechopen.92575.

The publisher is correcting [1] following an author’s request.

The title of the chapter is changed from “Satellite Control System: Part I– Architecture and Main Components” to
“Satellite Control System- Architecture and Main Components”.
Due to a misunderstanding, the Abstract had reference to the Part II of the chapter. However, Part II is planned for
publishing in the future IntechOpen book. Therefore, the last sentence in the Abstract “Part II will be presented as a
separate chapter in this book.” is changed to “Part II will be presented in one of the future books planned for release
by the publishing house.”

Changes have been made in online and print versions of the chapter.
This corrigendum is published in agreement with the authors.
The publisher and the author regret any inconvenience this might have caused to the readership.

References

[1] Kim YV. Satellite Control System-Architecture and Main components. In: Nguyen TM, editor. Satellite Systems
– Design, Modeling, Simulation and Analysis. London: IntechOpen; 2020. DOI: 10.5772/intechopen.92575.
Chapter

Game Theoretic Training Enabled

Deep Learning Solutions for Rapid
Discovery of Satellite Behaviors
Dan Shen, Carolyn Sheaff, Genshe Chen, Jingyang Lu,
Mengqing Guo, Erik Blasch and Khanh Pham

Abstract

The chapter presents a game theoretic training model enabling a deep learning
solution for rapid discovery of satellite behaviors from collected sensor data. The
solution has two parts, namely, Part 1 and Part 2. Part 1 is a PE game model that
enables data augmentation method, and Part 2 uses convolutional neural networks
(CNNs) for satellite behavior classification. The sensor data are propagated with
the various maneuver strategies from the proposed space game models. Under the
PE game theoretic framework, various satellite behaviors are simulated to generate
synthetic datasets with labels for the training to detect space object behaviors. To
evaluate the performance of the proposed PE model, a CNN model is designed and
implemented for satellite behavior classification. Python 3 and TensorFlow are used
in this implementation. The simulation results show that the trained machine learn-
ing model can efficiently and correctly classify the satellite behaviors up to 99.8%.

Keywords: space situational awareness, satellite characterization, sensor models,

simulated training data, training performance, general-sum games, CNN

1. Introduction

Since space has already been fully utilized, society has become increasingly
dependent on space advantages in various industrial, civil, and commercial appli-
cations. This dependence brings an essential vulnerability, especially shortage of
continuous situational awareness of the space environment to ensure freedom of
movement. Due to the fact that information from space is crucial for key decision-
making, such as urban, agricultural, and responsive planning, space is regarded as a
significant frontier. In addition to real-time and hidden information constrains, the
existence of space object density significantly produces the complexity of the space
situational awareness (SSA). Understanding the position of space objects from low-
level information fusion can support high-level information fusion SSA missions
of sensor, user, and task refinement [1]. In order to implement SSA accurately, it
is possible to coordinate the evaluation of residential space objects (RSO) through
user-defined operation pictures (UDOP) [2].
Space control and SSA are required for space prevalence, which depend on fast
and precise space object behavioral discovery. Developing a theoretical approach
for fast discovery of variation of satellite behaviors is the main task of this book

1
Satellite Systems - Design, Modeling, Simulation and Analysis

chapter. The machine learning methods with novel neural networks are proposed
in this case. However, there are numerous challenges for constructing the tools
because of the following reasons: (i) partially observable movements, (ii) resident
space objects (RSOs), (iii) uncertainties modeling and propagation, (iv) real-time
response, and (v) computationally intractable algorithms.
Space access investigation and mission trade-off considers are imperative for
the victory of space-borne operations. The tracking algorithms of space object can
be measured depending on collecting data to track satellites, debris, and natural
phenomena (such as comets, asteroids, and solar flares). Tacking is related with
sensor administration, which can point sensors to observation points to decide the
circumstance and to aware threatens. SSA improvements consist of models (such as
orbital mechanics), measurements, computational software (such as tracking), and
application-based system coordination (such as situations). For instance, game-
theory methods for SSA can be utilized for pursuit-evasion analysis [3].
This book chapter creates and establishes game theoretic training enabled deep
learning (GTEL) methodologies for fast discovery of the behaviors from the satel-
lites. GTEL is an adaptive feedback adversarial theoretic method, which acquires
data from sensors related to the relationship between resident space objects (RSOs)
of interest and sensing assets from ground and space (GSAs). Thus, a game theory
is modeled instead of a control problem for this circumstance. Game reasoning
uses data-level fusion, random modeling/propagation, on the other hand, RSO
detection/tracking predicting the future RSOs-GSAs relationships. The adversarial
engine also supports optional space pattern dictionary/semantic rules for adaptive
transition in the Markov game. In the event that no existing pattern dictionary is
accessible, GTEL will construct an initial pattern and modify it during the game
inference. The output of GTEL inference consists of two parts of control methods:
(i) measurements processing and (ii) RSOs localization. The two parts establish a
game-equilibrium, one of which is sensing asset management, the other of which is
the estimation of RSO behaviors.
The chapter is organized as the following. Section 2 introduces the comprehen-
sive system design for our methods. Section 3 presents the Markov game theory
with satellite maneuvering. Section 4 proposes the details of our machine learning
model for space behavior detection. And the numerical results and analysis are
displayed in Section 5. At the end, Section 6 draws the conclusion of this chapter.

2. Overall system architecture

The proposed methodology of GTEL is shown as Figure 1. The core piece of the
method for detecting unknown patterns of space objects is Markov Game Engine.
Due to the patterns are obscure, there is no preparing training data accessible.
Therefore, the Markov Game Engine makes use of zero-shot learning [4] and trans-
fer learning to unsupervised classify unavailable target domain data by training
available data from source domain (i.e. simulate data). The knowledge adaptation
or domain transfer is performed through manifold learning to share intermediate
semantic embeddings (such as attributes) between labeled and unlabeled data. On
one hand, the manifold learning can reduce the dimension of sensing data (such
as azimuth angle, elevation angle, range, and range rate). On the other hand, the
manifold learning can also mitigate the difficulties of object tracking and detection
with fast space object behaviors’ detection. Moreover, in order to solve the uncer-
tainties in this task, the ℝ 5 × 𝕊 coordinate system is used with filtering technology
on optimal transport (OT). The essential components of our methodology are shown
as the following:

2
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

1. Markov game – The space conflicting situation is inferred by the Markov game
structure [5, 6], in which system state is represented by distributions rather
than deterministic values.

2. Uncertainty modeling and propagation – The uncertainties are expressed by

the production of Von-Mises and independent Gaussian distribution for both
measurement interference and original state uncertainty that are identified on
cylindrical manifold ℝ 5 × 𝕊 [7]. Convergence is not guaranteed with relaxed
synchronization for uncertainty spread of information.

3. Optimal transport based tracking of space objects in cylindrical manifolds – Compared

with ensemble Kalman filter (EnKF) methods for tracking space objects, optimal
transport (OT) [8, 9] is much more precise and robust. Furthermore, it is way
broader than the algorithms with uncertainties assumption in ℝ n, because OT
is not related to the distribution (since OT is focused on a transformation instead
of importance sampling).

4. Course of actions for behavior modeling – The behavior of RSO is specified a

course of action (CoA), which in turn dictates to what the RSO may appear to
perform or is doing within subsequent a couple of steps.

5. Manifold learning for data level sensor fusion – The raw sensor data is generally
high-dimensional. Considering that the measurement data is reflections of
witnessed satellites (only several parameters will determine their states), so it
can be reasonably assumed that the inherent dimension of the measurement
domain is low. The communication bandwidth can be saved with decreasing
the number of dimensions by using Manifold learning algorithms [10].

The simulated positions of satellite are used to generate sensor measurements,

and the results will be utilized to track several space objects to complete the

Figure 1.
GTEL system architecture.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

estimation of the position. Afterwards, the estimations of the orbits are used for
collision alert and maneuver detection of space objects. The satellite maneuvers are
going to be interpreted as platform commands to perform course of actions and
space object movements.
Figure 1 shows an adaptive feedback approach with the game theory enabled. It
utilizes sensors to obtain information about the relationships between the ground/
space sensing assets and the RSOs of interest [11]. RSOs and GSAs determine the
relations. Thus, rather than saying it is a control problem, it is a dynamic game.
Data-level fusion, game reasoning, RSO detection/tracking, and uncertainty mod-
eling/propagation are combined to predict the RSO-SA relationships in the future.
Optional space behavior dictionary/semantic rule for adaptive transition matrices in
our Markov game is also supported by our game engine. In addition, the game solu-
tion is an equilibrium, which is controlled by both space sensing asset management
and the (estimated) RSO behaviors.

3. Markov game in space situational awareness

Lloyd Shapley has invented a concept of stochastic game [6], which is a dynamic
game played by one or several players focusing on probabilistic transitions. There
are several stages for this game. At first, the game is set in one state. Then, the par-
ticipated players should select an action individually. Based on the current state and
the actions players chosen, each player will receive a reward. Therefore, after the
chosen from each player, the game comes to a new random state, where the previous
state and previous actions chosen by the players determine a distribution of the new
random state. Afterwards, the above action will be repeated again for the new state.
After finite or infinite number of stages of playing, the total reward for each player
is obtained using the discounted sums of each stage reward or the averages of every
stage rewards. In this way, each player gets a reward and the reward is compared
with each other. The aforementioned sequences are the procedures for the sto-
chastic game, which can be generalized by Markov decision processes (MDP) with
repeated games. Our space situational awareness (SSA) would utilize this game tool
for intent prediction [12].
The Markov game engine extracts specific information from each event as the
following: (i) a finite set of players N, (ii) a finite or infinite set of states, S, (iii) a
finite set of accessible actions for each player in N set, Di (the overall action space
is D = ×i∈NDi), (iv) a transition rule q: S × D→∆(S), (where ∆(S) is the space of all
probability distributions over S), and (v) a reward function r: S × D→RN.
Figure 2 shows a visual description of the simple game states with only two play-
ers, who have only two options of actions for each player. The arrows in the graph
indicates the probable transitions from one state to the other state. The states with
red color means that only player 1 changes the approach. On the contrary, the states
with blue color indicates only player 2 changes the approach. The state with green
color indicates both players change the approach.
The GTEL solution uses a two-player Markov game to investigate the sensor
management for tracking space objects. Whether deliberate or unintentional,
some of space objects may cause confusion to observers (sensors) when the orbital
maneuvers are performed. In general, spatial object tracking can be assumed as an
optimal control problem (one side optimization) or a game problem (two side opti-
mization). For the settings of optimal control, the position and velocity of the space
objects will be calculated (filtered) as the states dependent on the measurement
from the sensor. However, this method ignores the possibility that the space objects
may alter their orbits purposely with intelligence. It may cause difficulties for the

4
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

satellite to track the space objects. Therefore, the Markov game method provides
a solution for these difficulties. In this approach, on the one hand, the observed
satellite will utilize the tracking and sensing model to destroy the tracking estima-
tions by confusing the observer. On the other hand, the observer figures out ways to
minimize the uncertainties of tracking, where the uncertainties are dependent on
the entropy of tracking.
In this chapter, the information uncertainty of the GTEL pursuit-evasion (PE)
game method [13] was exercised with a circumstance of two satellites, one of which
is Geostationary Earth Orbit – GEO (observed satellite), the other of which is space
based Low Earth Orbit-LEO satellite (observer satellite). Figure 3 provides an illus-
tration of a space based optical (SBO) sensor measurement model. The angle from
the line from the object to the SBO and the line from the sun with object is defined
as Bistatic Solar Angle, represented by θ. The smaller the angle is, the stronger
the lighting conditions. Therefore, it causes difficulties for observations when the
angle is large because of saturation of lighting. As shown in Figure 4, the scenarios
of light have shown. In the graph, the blue line is for the LEO orbit, the green line

Figure 2.
A diagram of states in a Markov game.

Figure 3.
SBO with a Bistatic solar angle.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 4.
LEO and GEO based on sensor management and maneuver strategies with game theory.

Figure 5.
The performance of tracking using theoretical Markov game strategies.

indicates the GEO orbit with maneuvers, the pink lines are the SBO sensor per-
formed to track the GEO in order to lower the uncertainty, and the red line displays
the sunlight from the sun to the earth.
The research scenario is shown in Figure 4, where the red line indicates the
direction of sun light, green color is for the GEO orbit with maneuvers, blue color

6
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

for LEO orbit, and pink lines indicate when the SBO sensor resource is used to track
the GEO (use the sensor data to lower the uncertainty).
Figure 5 shows the results of tracking based on intermitted measurements.
Both cubature KF (CKF) and extended Kalman Filter (EKF) trackers are shown.
With the increase of the period without measurement due to the Earth blockage,
the tracking errors increased as well. In addition, the tracking errors increased with
the maneuver actions from the satellite. On the contrary, informational entropy
decreased by sensor measures in this process.
Figure 6 top graph displayed the PE game control results, with α and β as angles
of the maneuver thrust. The zoom-in view of the game optimal controls is shown

Figure 6.
The maneuver controls dependent on the PE game solution for the satellite direction.

Figure 7.
Sensor’s game theoretic on–off controls and associated information gains.

7
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 6 below, which exhibited that the observer satellite can keep the tracking
uncertainty within a desired level while saving the resources of sensor. Figure 7
shows the on–off sensor controls and the associated information gains. The results
show that with the larger potential information gain, the sensor would use the
resources to take measures (for the observer’s on–off control, 0 means turning off
and 1 indicates turning on).

4. Machine learning based RSO behavior pattern classification

4.1 The scheme of machine learning

The machine learning (ML) details will be described in this section. The main
purpose of construction this machine learning scheme is to detect the behaviors
of the resident space object (RSO) by fusing sensors data from multiple sources,
including the velocity, orbital energy, angular momentum, and the position of RSO
compared to the station. Similar with other machine learning model, this model is
generally trained off-line by using generated data. Then the generalized weights will
be deployed in the application using TensorFlow deployment. In addition, in order
to improve the robustness of the trained system, we proposed a neural network
scheme with the ability to train the newly-added unknown pattern online with only
tiny modifications of the weights.
As shown in Figure 8, the RSO pattern classification architecture is displayed. It
consists of two separate parts, one of which is offline part as Modeling, the other of
which is online part as Monitoring, to detect the RSO behavior pattern.
As the matter of the offline part, named as Modeling RSO Behavior Pattern, our
neural networks will be trained as a classifier by using the collected simulated data.
The data specifically indicates the different behaviors of the RSO. As shown in the
graph, some useful features are obtained from different sensors in the sessions of
feature extraction. Subsequently, the extracted feature will be sent into the training
model for neural network tuning and training to generate a classifier, which is used
to identify the RSO behavior with fine grained size. On the contrary, the online
part, Monitoring RSO Behavior Pattern, acquires RSO patterns in real-time to distin-
guish the abnormal behaviors. In this way, a warning message would be prompted if
there are any abnormal behaviors detected.

Figure 8.
The scheme of RSO behavior classification.

8
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

Additionally, the classifier generated by the machine learning methods has sev-
eral properties as the following: (a) ability to fuse heterogeneous and complex input
data, (b) scalability, (c) robustness with perturbations relative to the data, (d) high
accuracy, and (e) explicitly. However, it is hard to fulfill all the requirements as
shown above, therefore, some trade off would be considered for our model. Among
different structures of neural networks, convolutional neural networks fit our case
appropriately.

4.2 Framework of convolutional neural networks (CNNs)

The structures of Convolutional Neural networks (CNN) as well as Dense

Neural Networks (DNN) are shown in Figure 9. The DNN neural networks consist
of several hidden layers with hidden neurons. Every neuron in the subsequent layer
is fully connected to the neurons in the previous hidden layer. The neurons contain
the linear functions with activation function for each neuron, which is completely
independent with the other neurons. Finally, after going through several layers,
the input data is generated to the output data as the “output layer”, which utilizes
softmax activation function to produce classification probabilities.
However, the DNN (as top image shown in Figure 9) has several drawbacks, such
as scaling difficulties for large images. For instance, for an image with 18 × 18 × 3
dimensions, DNN for the first layers will have 18 × 18 × 3 = 972 neurons. If the next
DNN layer has 30 neurons, the weight parameters will be 972 × 30 + 30 = 29190. With
a larger size of image, the weight parameters will increase a lot as well. Moreover,
the 30 neurons may not have enough complexity to generalize the accuracy of our
classification model.
On the other hand, the CNN (as bottom graph shown in Figure 9) can solve the
aforementioned issues. A simple CNN [14] consists of several different filters as
convolutional layer. In addition, the pooling layer can decrease the dimensions of
the input data. Meanwhile, since the adjacent data in an image has similar values,

Figure 9.
The framework of the convolutional neural network.

9
Satellite Systems - Design, Modeling, Simulation and Analysis

the filter can extract the features reasonably with few weight parameters. Thus,
using the CNN filters can increase the accuracy of our neural network classifier.
With the development of CNN, it successfully proceeds to be the best model
structure for computer vision and image processing tasks. Moreover, CNN has also
been used by Natural Language Processing (NLP) area. With small filters moving
across the input dataset, the filters, with only small numbers of parameters, are
re-utilized to recognize patterns for the large image. Therefore, with the similar
classification abilities, CNN network is faster to train and predict compared with
DNN. After several filters, the output part is flattened to bypass several dense layers
to produce the softmax activation at the last layer with the sparse cross-entropy loss
function to backpropagation for the behavior of RSO pattern prediction.
Our GTEL PE method utilizes the CNN architecture to classify the RSO pattern
with observed data. Compared with the other conventional methods, the convo-
lutional neural networks can process the RSO observation much faster. Therefore,
we utilize the Python and TensorFlow with Keras [15] as the fundamentals for
code write up. Although, each filter is computationally expensive to be trained,
the overall CNN architecture is faster to be trained to provide similar accuracy
classification.
A typical CNN structure is shown in Figure 10. After training our CNN-DNN
with our training data, the test data (with 10–20%) will be employed to evaluate
the generalization of our CNN-DNN model. Additionally, in order to solve the

Figure 10.
CNN architecture for RSO behavior classification.

10
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

overfitting problem of our model, several dropout layers were added after each
layer. This model outperforms other traditional methods with better accuracy and
higher computationally efficiency.
The revised deep learning neural network structure for 143-sapce-behavior (we
partition the pointing angles of α and β into cells with 15 degrees each and the total
cell number is 143) is shown in Figure 11. At first, there are 3 different dimensions
in our raw data, which are 3 parameters dimension, 15 times (each track consists
of 15 observation measurements) interval dimension, and 1 number of channels
dimension. Therefore, the convolutional neural network (CNN) comprises an
initial 3*15*1 input dimension with 72,000 samples as the raw dataset. In order
to distinguish the 143 different labels of satellite behaviors (15-degree difference
for between each behavior), 128 “2 × 2” filters are utilized with the same padding
for the first convolutional layer. Thus, there are 3*15*128 dimensions of output

Figure 11.
Revised CNN structure for 143-label data.

Layer (type) Output shape Param #

Conv2d (None, 3, 15, 128) 640
Conv2d (None, 3, 15, 128) 82,048
Conv2d (None, 3, 15, 256) 164,096
Conv2d (None, 3, 9, 512) 918,016

Conv2d (None, 3, 3, 256) 917,760

Flatten (None, 2304) 0
Dense (None, 1024) 2,360,320

Dropout (None, 1024) 0

Dense (None, 512) 524,800
Dropout (None, 512) 0
Dense (None, 512) 262,656
Dropout (None, 512) 0
Dense (None, 143) 73,359

Table 1.
The parameters of our 143 classification CNN-DNN model.

11
Satellite Systems - Design, Modeling, Simulation and Analysis

after the first layer. Then the previous output bypasses another two convolutional
layers with 128 and 256 “1 × 5” filters together with the same paddings. After the
first three convolutional layers, the dimension of the data exploded to 3*15*256.

Figure 12.
Model structure for the CNN_DNN network.

12
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

Afterwards, another two convolutional layers are added with the same padding
methods to downsample the dimensions to 3*3*256 for future operation. The two
layers both has 1 × 7 filters with 512, and 256 third dimension respectively. Finally,
the 3*3*256 data is flattened to one dimension to bypass three dense layers with
1024, 512, and 512 neurons, respectively. All these layers utilized 30% dropout
parameters after dense layers. The employment of the dropout layer is used for
solving the overfitting problem during the training process. In order to predict
the 143 labels classifications, there is a 143-degree softmax attached at the last
layer of the CNN-DNN model. Additionally, the cross-entropy loss function with
the Adam Optimizer minimizes the result with the gradient of the next calcula-
tion points.
The details of the parameters of the CNN are shown in Table 1 and Figure 12.
There are 5,303,695 parameters that need to be trained in this large NN for detect-
ing the different behaviors. In order to train the GTEL CNN much faster and more
accurately, a learning rate decay is used during the training process. For the first 75
epochs, the learning rate is 1e−4, which can learn faster after the weight initializing.
After 75 epochs, the exponentially decay of learning rate finds the optimized path
for decreasing the loss for the CNN model.

5. Numerical results and discussion

5.1 Training data generation

In order to capture the training data with the two-line elements (TLEs) from the
space-track.org, we modified the catalog tracking by the addition of the maneu-
vers. The different maneuvers indicate different labels in our training data. As an
example, the maneuver to increase the energy of the orbital is labeled as 1. On the
contrary, the maneuver to decrease the energy of orbital is labeled as 2. In this way,
adding no maneuver is labeled as 3. The details related to adding maneuvers are
shown as the following:

a. The earth-centered inertial coordinates (ECI) is converted from the TLEs at 0

time step;

b. Use the specified methods (Markov game) shown in Section 3 to propagate

the satellites;

c. Convert the 16 waypoints back to azimuth, elevation, range, range rate

relative to a ground site.

Therefore, almost 57,332 tracks were generated for training dataset with the
other 6371 tracks as the testing dataset. The original first 10,000 tracks are shown in
Figure 13 for 143-space-behavior maneuvers.
The data format is list as:

• Column 1: track id

• Column 2: observation id (from 1 to 15)

• Column 3: Azimuth angle (rad)

• Column 4: Elevation angle (rad)

13
Satellite Systems - Design, Modeling, Simulation and Analysis

• Column 5: Range (km)

• Column 6: Training label (from 1 to 143)

5.2 Results and analysis of our CNN_DNN neural networks

After 1000 epoch optimization, the GTEL systems achieved almost 98% training
accuracy with 96% test accuracy as shown in Figure 14. There is a little overfitting
after 400 epochs of training, which can be solved by adding more dropout layers
and other methods. Figure 15 displays the cross-entropy loss during the training
process. The training set loss is always decreasing; however, the loss of validation
set is flat after 200 epochs of training, indicating some overfitting issue after 200
epochs of training of the data.
Figure 16 shows the confusion matrix to evaluate our model’s performance. As
shown, for both training and test datasets, the true labels are almost the same with
the predicted label by the GTEL CNN model. The performance evolution pattern is
shown in Figure 17.

Figure 13.
The first 10,000 training tracks with various space behaviors.

Figure 14.
Training and validation accuracy during the training process.

14
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

Figure 15.
Training and validation loss during the training process.

Figure 16.
Confusion matrix of training data and testing data.

For filter-based normalization, changes are grouped normalized by filters,

which aims to display the change distributions over iterations for individual filters.
For instance, as shown in Figure 18, the filter changes in the convolutional layer
and dense layers are visualized. The changes are drastic in the first several itera-
tions and become relatively small in the later stages (after 400 epochs) for most
of the layers due to learning rate decay as well as the convergence of the GTEL
CNN model.
Notice in Figure 18 (the color maps are shown on the right side of each plot, the
whiter the more stable of the training process), there is no constant deep blue color

15
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 17.
Overview of validation classes prediction accuracy during training epochs.

for all training process, indicating the great performance and reasonable training
configuration during training of the deep learning neural networks. Understanding
the meaning for these weights during training process with further checking is very
important in the CNN model design, which leads to more reasonable explanation in
the deployment of the model.
In addition, we investigated the filter image correlations in a broad overview.
As shown in Figure 19, rows and columns represent layers and image classes,
respectively. A sequential color scheme is used to encode the number of anomaly

16
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

filters to intuitively represent these relationships between satellite behavior labels

and anomaly filters. Using a grid-style visualization, interpretability is possible
where rows and columns represent layers and classes of behaviors, respectively.
The number of rows and columns equal to the number of layers with anomaly
filters and classes with anomaly iterations each class, respectively. In Figure 19,
the darker the color it is, the more anomaly weights and filters appear in that layer,
which are related to that class. From this visualization, it is easy to observe that
the second fully connected dense layer has the most anomaly weights and filters
among all the other layers. Hence, there are some trends for most anomaly weights
and filters during training processes, especially the middle layers of the deep
convolutional neural networks. In addition, the anomaly class seems to have the
same interval in the dataset, indicating more work is needed for these classes and
layers to build a better network. It displays a similar trend compared to the results
as shown in Figure 17.

17
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 18.
The weight changes in filters during iterations. More blue color indicates stronger variation for the filters during
iterations.

Figure 19.
The abstract version of correlation view.

18
Game Theoretic Training Enabled Deep Learning Solutions for Rapid Discovery of Satellite…
DOI: http://dx.doi.org/10.5772/intechopen.92636

6. Conclusions

In this chapter, a machine learning design has been presented and implemented
to discover space object behaviors. Our GTEL methods using CNN models the cir-
cumstance with PE game rather than a control problem. The stochastic modeling/
propagation, RSO tracking, and data level fusion are utilized to predict the relations
for future space by game reasoning. In order to generate the data for training, the
Marko game approach is used with maneuvering strategies. The method provides
a way to solve the SSA using unknown behaviors. Additionally, the unknown
behaviors exist where a satellite employs the tracking and sensing way to corrupt
the tracking estimates to perturb the sensors. On the other hand, the space sensors
decrease the uncertainties during tracking process. Finally, the CNN-DNN is used
to train the numerical results, where the accuracy is 98% for our classification RSO
model with 143 labels.
In the future, a multi-player game theory with adversarial network with SSA
will be employed to enhance the deep learning for sensor management, combined
tracking, as well as secure communications. Methods for diffusion-based coopera-
tive space object tracking [16] and block chain [17] are emerging as methods for
game-theoretical methods.

Acknowledgements

The work was supported under contract FA9453-15-C-0459, FA9453-15-C-0423,

and FA8750-19-C-1000. The views and conclusions contained herein are those of
the authors and should not be interpreted as necessarily representing the official
policies, either expressed or implied, of AFRL, or the U.S. Government.

Author details

Dan Shen1, Carolyn Sheaff 2, Genshe Chen1*, Jingyang Lu1, Mengqing Guo1,
Erik Blasch3 and Khanh Pham4

1 Intelligent Fusion Technology, Inc., Germantown, MD, USA

2 Information Directorate, Air Force Research Laboratory (AFRL), Rome, NY, USA

3 Air Force Office of Scientific Research (AFOSR), Arlington, VA, USA

4 Space Vehicles Directorate, Air Force Research Laboratory (AFRL),

Albuquerque, NM, USA

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

19
Satellite Systems - Design, Modeling, Simulation and Analysis

References

[1] Blasch E, Bosse E, Lambert DA. [10] Shen D, Blasch E, Zulch P,

High-Level Information Fusion Distasio M, Niu R, Lu J, et al. A joint
Management and Systems Design. manifold leaning-based framework
Norwood, MA: Artech House; 2012 for heterogeneous upstream data
fusion. Journal of Algorithms and
[2] Blasch E. Enhanced air operations Computational Technology (JACT).
using JView for an air-ground fused 2018;12(4):311-332
situation awareness UDOP. In: AIAA/
IEEE Digital Avionics Systems [11] Jia B, Pham K, Blasch E, Shen D,
Conference, October. 2013 Wang Z, Chen G. Cooperative space
object tracking using space-based
[3] Shen D, Pham K, Blasch E, Chen H, optical sensors via consensus-based
Chen G. Pursuit-evasion orbital game filters. IEEE Transactions on Aerospace
for satellite interception and collision and Electronic System;52(4):1908,
avoidance. Proceedings of the SPIE. 2016-1936
2011;8044:89-97
[12] Chen G, Shen D, Kwan C, Cruz J,
[4] Palatucci M et al. Zero-shot learning
Kruger M, Blasch E. Game theoretic
with semantic output codes, advances in approach to threat prediction and
neural information processing systems. situation awareness. Journal of
2009 Advances in Information Fusion.
2007;2(1):35-48
[5] Mertens JF, Neyman A. Stochastic
games. International Journal of Game
[13] Shen D, Jia B, Chen G, Pham K,
Theory. 1981;10(2):53-66. DOI: 10.1007/
Blasch E. Game optimal sensor
BF01769259
management strategies for tracking
[6] Shapley LS. Stochastic games.
elusive space objects. In: Proceedings of
Proceedings of the National Academy IEEE Aerospace Conference. 2017:1-8.
of Sciences of the United States of DOI: 10.1109/AERO.2017.7943676
America. 1953;39:1095-1100
[14] Zou H, Lang H, et al. Covert photo
[7] Horwood JT, Poore AB. Gauss classification by deep convolutional
von mises distribution for improved neural networks. Machine Vision and
uncertainty realism in space situational Applications. 2017;28(5-6):623-634
awareness. SIAM/ASA Journal
on Uncertainty Quantification. [15] Available from: https://www.
2014;2(1):276-304 tensorflow.org/

[8] Das N, Ghosh RP, Guha N, [16] Jia B, Pham K, Blasch E, Shen D,
Bhattacharya R, Mallick B. Optimal Chen G. Diffusion-based cooperative
transport based tracking of space space object tracking. Optical
objects in cylindrical manifolds. The Engineering. 2019;58(4):041607
Journal of the Astronautical Sciences.
2019;66:582-606 [17] Xu R, Chen Y, Blasch E, Chen G. An
exploration of Blockchain-enabled
[9] Das N, Deshpande V, Bhattacharya R. decentralized capability-based access
Optimal-transport-based tracking of control strategy for space situation
space objects using range data from awareness. Optical Engineering.
a single ranging station. Journal of 2019;58(4):041609
Guidance, Control, and Dynamics.
2019;42(6):1237-1249

20
Chapter

Future Satellite System

Architectures and Practical Design
Issues: An Overview
Tien M. Nguyen

Abstract

This chapter discusses existing and future trends on the design and build of
“Modular” and “Open” satellite Bus and mission payload along with practical design
issues associated with the use of Modular Open System Approach (MOSA). Existing
modular Bus and mission payload architectures for typical commercial, civilian, and
military satellite systems will be discussed. The chapter provides space industry
views on “Open” versus “Close” interfaces design and addresses the challenges
associated with open interfaces using Open System Architecture (OSA) approach
using MOSA principles. The system interfaces discuss in this chapter include (i)
internal to satellite Bus and mission Payload (PL), (2) between satellite Bus and
mission payload, and (3) external to both satellite Bus and mission payload.

Keywords: open system architecture, Modular Open System Approach (MOSA),

satellite Bus, mission payload (PL), modular architecture, open Interface, close
Interface, modular satellite Bus, modular Mission payload

1. Background and introduction

Typical commercial and civilian satellite systems take about 2–3years to build
and launch [1–4], while military systems take between 7 and 10 years [5, 6]. A
typical production flow for assembling and launching of a space vehicle is presented
in Ref. [6] and redrawn in Figure 1 as introduction steps for better understanding of
the design, build and launch of a satellite system. This chapter focuses on practical
design issues for satellite Bus’ and mission PL’s system/subsystem components
builds, and corresponding interface-design’s challenges associated with satellite Bus
integration, mission PL integration, and satellite system integration. A survey of
existing commercial, civilian and military satellite systems revealed that a typical
satellite Bus includes the following modular components [7–16]:

• Bus Subsystem 1—Bus TX/RX Antenna Subsystem (BAS): Provide Bus’s

Receive (RX)/Transmit (TX) antennas and associated Bus’s antenna beam
control functions;

• Bus Subsystem 2—Bus Communication RF Front-End-Back-End Subsystem

(BCom-RFS): Provide Low Noise Amplifier (LNA), High Power Amplifier
(HPA), satellite Bus Down/Up Radio Frequency (RF)-to-Intermediate

1
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 1.
A typical satellite system production flow (redrawn from [6]).

Frequency (IF) conversion, and Analog-to-Digital/Digital-to-Analog

conversion functions—Note that typical HPAs are Traveling Wave Tube
Amplifier (TWTA) and Solid State Power Amplifier (SSPA), and some
advanced satellite transponders use Linearized TWTA (L-TWTA) or L-SSPA
in the RF Back-End Subsystem;

• Bus Subsystem 3—Bus Command & Data Handling Subsystem (BC&DHS):

On-board computer that interfaces with all Bus components;

• Bus Subsystem 4—Bus Telemetry-Tracking & Command Subsystem

(BTT&CS): Process uplink satellite Bus command data, perform satellite
tracking functions and provide downlink Bus telemetry reporting satellite Bus’s
heath and conditions;

• Bus Subsystem 5—Bus Electrical Power Subsystem (BEPS): Provide and

regulate Bus power;

• Bus Subsystem 6—Bus Thermal Control Subsystem (BTCS): Maintain Bus’

thermal environments;

• Bus Subsystem 7—Bus Altitude and Determination Control Subsystem

(BADCS): Provide satellite stabilization, control and positioning;

• Bus Subsystem 8—Bus Propulsion Subsystem (BPS): Provide propulsion

functions for satellite maneuvering;

• Bus Subsystem 9—Bus Communication Security Subsystem (BCOMSEC):

Provide Bus data encryption and decryption functions to protect data from
intruders. Typically, BCOMSEC is tightly coupled with BTT&CS;

• Bus Subsystem 10—Bus Structure & Mechanism Subsystem (BS&MS):

Provide structure and mechanism to mount all satellite Bus components.

Similarly, our survey also revealed that a typical mission PL consists of the
following modular components [7–16]:

2
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

• PL Subsystem 1—PL PAS: Similar to BPAS but for mission PL;

• PL Subsystem 2—PL Com-RFS: Similar to BCom-RFS but for mission PL;

• PL Subsystem 3—PL Digital Processing Subsystem (PDPS): Provide mission

specific processing functions. For SATCOM missions, specific processing
functions can be dynamic resources control, channelization processing, etc. For
PNT missions, the functions can be time transfer processing functions. For
imaging/sensing missions, the functions can be image preprocessing functions;

• PL Subsystem 4—PL C&DHS: Similar to BC&DHS but for mission PL;

• PL Subsystem 5—PL TT&CS: Similar to BTT&CS but for mission PL;

• PL Subsystem 6—PL EPS: Existing PLs use power supply from satellite BEPS;

• PL Subsystem 7—PL TCS: Maintain PL’s thermal environments;

• PL Subsystem 8—PL ADCS: Existing PLs use ADCS from the satellite BADCS;

• PL Subsystem 9—PL PS:: Existing mission PLs use PS from the satellite BPS;

• PL Subsystem 10—PL COMSEC: Similar to BCOMSEC but for mission PL;

• PL Subsystem 11—PL Frequency & Timing Subsystem (PFTS): Provide

reference frequency and timing functions to meet specific mission requirements;

• PL Subsystem 12—PL Transmission Security Subsystem (TRANSEC): Provide

security functions to combat unintentional and/or intentional Radio Frequency
Interference (RFI) (e.g., frequency hopping/de-hopping, frequency
spreading/de-spreading);

• PL Subsystem 13—PL Specific Mission Suite (SMS): Provide specific mission

PL processing functions depending on whether the mission is Satellite
Communications (SATCOM) mission or Position Navigation and Timing
(PNT) mission or Imaging/Sensing mission [7–16];

• PL Subsystem 14—PL Structure & Mechanism Subsystem (BS&MS): Provide

structure and mechanism to mount all mission PL components.

In practice, the above satellite Bus modular components can be found in the
following typical satellite Busses [7, 11, 14]:

• Loral Satellite Bus 1300 or Loral 1300

• Lockheed Martin (LM) A2100A/AX-Land Mobile/AX-High Power

• Boeing 702HP/HP-GEM/MP/702SP and 502.

For achieving optimum weight and power, existing satellite Bus and mission PL
are tightly coupled together with customized interface design. The industry trends
for the design and build of future satellite systems are moving toward OSA using
MOSA principles, in which the satellite Bus is loosely coupled with the mission PL

3
Satellite Systems - Design, Modeling, Simulation and Analysis

using “Open” and widely accepted interface standards. The key communication
linkage between a satellite Bus and a mission PL is the communication data Bus.
Currently, majority of satellite Busses employ the standard 1553 data Bus for data
communications among Bus components, and between the satellite Bus and mission
PL components. The communications over 1553 data Bus is limited to 1 Mega bit per
second (Mbps). Recently, there was an advanced development effort that was
funded by the U.S DOD to develop new 1553 standards called 1553 Enhanced Bit
Rate (EBR–1553) increasing the speed to 10 MB/s [17]. The EBR-1553 requires a
star/hub topology to provide the higher data rate and additional components to
implement the architecture. For data rates larger than 10 Mbps, space industry
trend is moving toward SpaceWire data Bus that was recently developed in Europe
for use in commercial satellites and scientific spacecraft [18].
The objective of this chapter is three-fold: (1) Provides an overview of existing
modular satellite Bus, mission PL architectures and related communication data
Busses, (2) Discusses future trends on the modular and open design and build of
satellite Bus and mission payload using MOSA principles, and (3) Addresses the
practical design challenges associated with “Modular” and “Open” design for future
satellite Bus and mission PL. The chapter is organized as follow: (i) Section 2
describes existing modular satellite Bus and mission PL architectures and related
communication data Busses; (ii) Section 3 presents industry view on “Open” and
“Close” interfaces for connecting satellite system components and existing popular
standards; (iii) Section 4 discusses the interface design challenges and provides an
overview of MOSA and related DOD Guidance and assessment tools for MOSA
implementation; (iv) Section 5 provides examples how to transition modular satel-
lite Bus and mission PL architectures to modular-and-open architectures using
MOSA implementation approach and tools in Section 4; and (v) Section 6 concludes
the chapter with remarks on the benefits associated with the proposed approach.

2. Existing satellite systems, related interfaces and standards

Figure 2 describes an overview of existing satellite systems, consisting of a

satellite Bus, a mission PL and a typical set of interfaces between the Bus and PL
using a standard data Bus. A typical set of interfaces between the satellite Bus and a
mission PL includes seven interface types, namely: (i) Physical & Mechanical
Interface, (ii) Electrical/Power/Cable Interface, (iii) Grounding Interface, (iv)
Software & Data Interface, (v) Electromagnetic Compatibility (EMC)/ Electro-
magnetic Interference (EMI)/Electromagnetic Pulse (EMP) and Electro Static Dis-
charge (ESD) Interface, (vi) Thermal Interface, and (vii) Frequency & Timing
(F&T) Interface. This section focuses on satellite Bus and mission PL architectures
and the data interfaces between them. Subsections 2.1 and 2.2 describe existing
satellite Bus and mission PL architectures along with related interfaces and industry
standards, respectively. Subsection 2.3 discusses existing standard 1553 data Bus and
the pushes from space industry moving toward military standard 1553-B data Bus
(MIL-STD-1553-B) and high-data-rate SpaceWire data Bus.

2.1 Existing satellite bus, related interfaces and standards

As described in Section 1, existing satellite Bus architecture includes typical 10

modular components, namely, BAS, BComRFS, BC&DHS, BTT&CS, BEPS, BTCS,
BADCS, BPS, BCOMSEC and BS&MS. A functional description for each of these
modular Bus components is also described in Section 1. Figure 3 illustrates a
notional block diagram for existing modular satellite Bus architecture. The figure

4
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

Figure 2.
Overview of existing satellite Systems using standard 1553 data bus.

Figure 3.
Existing notional modular satellite bus architecture.

shows that space industry has used the modular design concept to architect the
satellite Bus, where common functions are group together and then isolate or
separate from the other group of functions. As an example, BAS consists of a
group of antenna components and control functions (e.g., antenna pointing,
beamforming, etc.), which is separated and isolated from BComRFS. It is important
to note that the figure also shows how these satellite Bus components are connected
together, i.e., the lines with arrows connecting them. These lines represent the
interfaces among the Bus components, where the interface can be any of the seven
interface types described above. Below is a list of some of the existing interfaces and
associated standards for existing satellite Bus based on National Aeronautical and
Space Administration (NASA), European Space Agency (ESA), U.S. DOD and
international Consultative Committee for Space Data System (CCSDS) standards
[19–26]:

5
Satellite Systems - Design, Modeling, Simulation and Analysis

• Typical NASA Electrical/Power/Cable Interface Standards [19, 20]:

◦ Satellite Bus shall protect its own electrical power system via overcurrent
protection devices on its side of the interface.

◦ Satellite Bus shall deliver a maximum transient current on any Power Feed
Bus of 100% (that is, two times the steady state current) of the maximum
steady-state current for no longer than 50 ms.

◦ Bus Survival Heaters, which are elements of the Bus thermal subsystem,
shall be required to have power to heat certain satellite Bus components
during off-nominal scenarios when the BEPS power is not fully energized.

• Typical U.S. DOD EMC/EMI/EMP/ESD Interface Standards [21]:

◦ Power line conducted emissions for satellite Bus equipment shall meet the
EMC interface specification specified in SMC Standard Handbook, SMC-
S-008, Section 6, 6.01, 6.02, 6,03, 6.04, 6.05, 6.06, 6.07, and 6.08.

◦ Power line conducted susceptibility for satellite Bus equipment shall meet
the EMC interface specification specified in SMC Standard Handbook,
SMC-S-008, Sections 6, 6.10, 6.11, 6.12, 6.13, 6.14, 6.15, 6.16, 6.17, 6.18
and 6.19.

◦ ESD susceptibility for satellite Bus equipment shall meet the EMC
interface specification specified in SMC Standard Handbook, SMC-S-008,
Section 6, 6.43.

◦ EMP susceptibility for satellite Bus equipment shall meet the EMC
interface specification specified in SMC Standard Handbook, SMC-S-008,
Section 6, 6.45.

• Typical NASA Grounding Interface Standards [20, 22]:

◦ Satellite Bus EPS should ground in a way that reduces introducing stray
currents or ground loop currents into the satellite Bus components.

◦ Satellite Bus ground interface shall follow NASA single-point ground or

multiple-point ground architecture.

• Typical NASA Thermal Interface Standards [19, 20]:

◦ A conductive heat transfer of 15 W/m2 or 4 W shall be considered small

enough to meet the intent of being thermally isolated.

• Typical Software & Data Interface Standards [19, 20, 23, 24, 25, 26]:

◦ Satellite Bus command and telemetry data formats shall be NASA Unified
S-Band (USB)/CCSDS standards or U.S. DOD Space-Ground Link
Subsystem (SGLS) standards. Note that (i) most of NASA and ESA
standards are CCSDS compliance for interoperability purpose, and (ii)
some military systems have both USB and SGLS capabilities.

6
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

◦ Satellite Bus “Safe Mode” is a combined satellite Bus components

hardware and software configuration that shall be designed to protect the
components from possible internal or external harm while making
minimal use of satellite Bus resources (e.g., power).

◦ Satellite Command SAFE Mode shall be required to protect and preserve

satellite Bus components under anomalous and resource constrained
conditions.

◦ Satellite Bus components shall respond to uplink commands from the

Satellite Operation Center (SOC) to suspend and resume the transmission
of the Components’ telemetry data. For commercial satellite systems, SOC
can also control the mission PL.

For military applications, majority of satellite Busses are usually designed using
contractor’s custom designed interfaces and very tightly couple together to reduce
weight, size and power. It is for this reason, current military satellite BTT&CS com-
ponent also include the COMSEC component. For commercial applications, satellite
developers are also concerned with weight, size and power reduction, but they are
also concerned with component refresh and upgrade without redesigning the satellite
Bus, hence commercial satellites tend to use modular Bus components and widely
accepted interface standards to connect the internal Bus components. Industry views
on the “open” and “close” interfaces will be addressed in Section 4.

2.2 Existing mission payload, related interfaces and standards

As pointed out in Section 1, existing mission PL architecture consists of 14

modular components, but there are three PL components that rely on the satellite
Bus’ design, namely, PL EPS, PL ADCS and PL PS. Therefore, the mission PL
architecture usually has 11 modular components, including PL AS, PL Com-RFS,
PDPS, PL C&DHS, PL TT&CS, PL TCS, PL COMSEC, PFTS, PL TRANSEC, PL
SMS and PL S&MS. A functional description for each of these mission PL modular
components is also provided in Section 1. Figure 4 presents a notional block dia-
gram for existing modular mission PL architecture. Similar to the satellite Bus

Figure 4.
Existing notional modular Mission payload architecture.

7
Satellite Systems - Design, Modeling, Simulation and Analysis

design, the space industry has also applied the modular design concept to architect
the mission PL. Below is a list of some of the existing interfaces and associated
standards for existing mission PL leveraged from NASA, ESA, U.S. DOD and inter-
national CCSDS standards [19–26]:

• Typical NASA Electrical/Power/Cable Interface Standards [19, 20]:

◦ Sizing all components of the mission PL power harness, such as the wires,
connectors, sockets, and pins to the peak power level shall be required by
the mission PL equipment in addition to satellite Bus to prevent damage to
the power harnessing.

◦ PL Survival Heaters shall be required to have power to heat certain

mission PL components during off-nominal scenarios when the BEPS
power is not fully energized.

• Typical U.S. DOD EMC/EMI/EMP/ESD Interface Standards [21]: Similar to

satellite Bus discussed above but for mission PL.

• Typical NASA Grounding Interface Standards [20, 22]: Similar to satellite Bus
discussed above but for mission PL.

• Typical NASA Thermal Interface Standards [19, 20]:

◦ The mission PL thermal design should be decoupled from the satellite Bus
at the mechanical interface between the satellite Bus and neighboring
mission payload to the maximum practical extent.

◦ A conductive heat transfer of 15 W/m2 or 4 W shall be considered small

enough to meet the intent of being thermally isolated.

• Typical Software & Data Interface Standards [19, 20, 23, 24, 25, 26]:

◦ Mission PL command and telemetry data formats shall be NASA USB/CCSDS

standards commercial applications or U.S. DOD SGLS standards for military
applications. Some military systems have both USB and SGLS capabilities.

◦ PL “Safe Mode” is a combined mission PL components hardware and

software configuration that shall be designed to protect the PL
components from possible internal or external harm while making
minimal use of satellite Bus resources (e.g., power).

◦ PL Command SAFE Mode shall be required to protect and preserve mission

PL components under anomalous and resource constrained conditions.

◦ Mission PL components shall respond to uplink commands from Mission

Control Center (MCC) to suspend and resume the transmission of the
mission PL components.

◦ Mission PL shall be responsible for on-board mission data storage capabilities.

For most commercial applications, the MCC can be merged with the SOC, and
the mission PL TT&CS (PTT&CS) and PL CD&HS (PCD&HS) components can be

8
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

incorporated into satellite (i) Bus TT&C (BTT&CS) and (ii) Bus CD&HS
(BCD&HS) components, respectively. Similar to the satellite Bus interfaces design,
for military applications, the mission PL components are tightly coupled using
contractor’s custom designed interfaces. For commercial applications, the mission
PL components are loosely coupled using widely accepted open interfaces.

2.3 Existing data busses

Subsections 2.3.1 and 2.3.2 provide an overview of standard 1553 and SpaceWire
communication data Busses, respectively.

2.3.1 Standard 1553 data bus

Existing commercial, civilian and military satellite data Busses have been using
Military Standard 1553B (MIL-STD-1553B) data Bus for communications among
satellite Bus and mission PL components. Figure 5 describes a typical MIL-STD-
1553B System [17, 27, 28]. This figure uses MIL-STD-1553B terminologies: (i) the
Bus Controller (BC) is considered as an Intelligent Terminal (IT) that is located in
the satellite mission computer, which is usually referred to as a Satellite Bus C&DH
component, and (ii) Remote Terminal (RT) is considered as a slave terminal that is
located in satellite platform components, which can be located in any satellite Bus or
mission PL components.
Figure 5 shows a typical commercial satellite system with RTs located in both
satellite Bus and mission PL components. As an Example, the RTs located in satellite
components are BAS, BADCS, BTCS and BTT&CS; and RTs located in the mission
PL components are PAS, PTCS, PDPS, PTRANSEC, and PFTS. For military appli-
cations, the Mission Computer (MC) can be located in both satellite Bus and mis-
sion PL, where the MC in the satellite Bus is responsible for all control functions
associated with the satellite operations and MC in the mission PL is responsible for
all control functions related to the mission PL operations.

Figure 5.
Typical civilian and commercial MIL-STD-1553B satellite Systems.

9
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 6.
Typical civilian and commercial SpaceWire satellite Systems.

2.3.2 Standard SpaceWire data bus

SpaceWire (SpW) is an industry standard with protocol derived from IEEE-1355

and ECSS-E50-12C managing by the international SpW working Group [18, 29, 30].
The SpW standard is a self-managing serial protocol that provides a high-speed data
rates from 2 to 400 Mbps, and low power serial interface using LVDS1 Drivers with
distances up to 30 feet while offering a flexible simple user interface. Figure 6
illustrates typical uses of SpW data Bus with a PCD&HS, a SpW Router and SpW
cables for connecting mission PL components. Some examples of existing satellite
programs employed SpaceWire standard are: TacSat (part of the U.S. Operationally
Responsive Space Program), NASA Lunar Reconnaissance Orbiter (Orbiting the
Moon taking high resolution images), ESA Sentinel-3 (a pair of satellites providing
operational Earth observation services using optical and microwave instruments),
and Japanese NEC NEXTTAR (one of the first spacecraft designed using SpW for all
of its onboard communications).

3. Industry view: open vs. close interfaces and standards

Figure 7 presents the space industry view on open and close interface design.
This view separates the interface design into two categories, namely, Contractor
Proprietary Interface and Contractor Non-Proprietary Interface. Under this view,
the interface standards are then classified into two categories, namely, Preferred
and Non-Preferred Interface Standards. Based on this view, Section 3.1 defines open
interface design, and Section 3.2 defines close interface design. Section 3.3 provides
a list of existing popular open standards widely accepted by space industry.

1
LVDS is defined as Low Voltage Differential Signaling TIA/EIA-644, is a technical standard that
specifies electrical characteristics of a differential, serial communication protocol. LVDS Drivers use 80%
less current than current popular Pseudo Emitter-Coupled Logic (PECL) devices.

10
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

Figure 7.
Industry view on open and closed interfaces design.

3.1 Open Interface design

From Figure 7, the open interface design falls into the contractor non-
proprietary design category. For the interface design to be open, the interface
design shall not be contractor proprietary and that the interface shall use either
popular open interface standards widely accepted by space industry or open inter-
face standards with little market support and narrowly used by space industry.
Thus, a popular open interface design is a non-proprietary design that uses popular
open interface standard that is widely used by space industry. The benefits of open
interface design for the satellite buyers are (i) improving competition allowing
various space vendors (or contractor) to build open satellite Bus and mission PL
subsystem components, (ii) ease of refresh and technology upgrade allowing to
swap subsystem components without impacting the overall system, (iii) ease of
adapting to new requirements and operational threats, (iv) incorporating innova-
tion by allowing operational flexibility to configure and reconfigure a mission PL
quickly to meet rapidly changing operational requirements, (v) enabling cost saving
and cost avoidance during the design and sustainment phases by reusing technology
and Software/Hardware/Middleware (SW/HW/MW) components, and using
existing standardized HW/SW/MW parts and modules, and (vi) improving inter-
operability where severable HW/SW/MW modules can be changed independently.

3.2 Close interface design

As shown in Figure 7, the close interface design shall fall into contractor propri-
etary category. For an interface design to be close, it shall be contractor proprietary
and that the interface shall use either close interface standards with little market
support narrowly used by space industry or popular closed interface standards
widely used by space industry. Thus, a popular close interface design is a contractor
proprietary design that uses popular closed interface that is widely used by space
industry. The key benefits of close interface design are the potential reduction of
weight, size, power and manufacturing cost.

3.3 Popular open standards

Based on Figure 7, the criteria for popular open standards are (i) publicly
available and widely used by both satellite Bus and mission PL vendors, (ii)

11
Satellite Systems - Design, Modeling, Simulation and Analysis

community and/or industry consensus-based that are matured and stable, and (iii)
technically adequate for all future commercial, civilian and military satellite sys-
tems. Following is a list of current popular standard organizations and widely
adopted open standards [18–31]:

• Consultative Committee for Space Data Systems (CCSDS) Standards: is a

multi-national forum for the development of communications and data
systems standards for spaceflight. The goal is to enhance governmental and
commercial interoperability and cross-support, while also reducing risk,
development time and project costs.

• AIAA Space Plug-and-play Avionics (SPA) Standard: SPA is a set of

AIAA standards developed for spacecraft platform, subsystem, and
component (including payload) developers for integrating plug-and-play
characteristics into spacecraft structures, avionics, and hardware and software
components to promote their rapid integration. The SPA community
anticipates adding protocols (e.g., Ethernet as SPA-E) as the PnP capabilities
are normalized.

• MIL-STD-1553 Standard: is a military standard published by the United States

Department of Defense that defines the mechanical, electrical, and functional
characteristics of a serial data Bus.

• SpaceWire Standard: is a spacecraft communication network standard based in

part on the IEEE 1355 standard of communications. It is coordinated by the
European Space Agency (ESA) in collaboration with international space
agencies including NASA, Japanese Space Agency (JAXA) and Russian Federal
Space Agency (RKA).

• NASA/SMC/Aerospace Hosted Payload Interface Design (HPID): this design

guideline provides a prospective Instrument Developer with technical
recommendations to assist them in designing an Instrument or Payload that
may be flown as a hosted payload on commercial satellites flown in Low Earth
Orbit (LEO), or Geostationary Earth Orbit (GEO).

• SQL for databases specified in ANSI ISO/IEC 9075–1, ISO/IEC 9075–2, ISO/IEC
9075–3, ISO/IEC 9075–4, ISO/IEC 9075–5.

• HTML for presentation layer specified in XML 1.0 www.webstandards.org.

• XML for data transfer.

• Web Services for remote system calls.

• U.S. Space and Missile Systems Center (SMC) approved a project for
developing Common Payload Interface Specification (CoPaIS) standard for
satellite-to-payload Command and Data Handling (C&DH) interface intended
for all future SMC procured medium to large satellites [31].

• Other popular standards: MIL-STD-1553B, CAN Bus, RS-422 (TTC-B01

Protocol)/EIA/TIA-422, RS-422 (PC-Protocol).

12
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

4. Interfaces design challenges and MOSA implementation

This section addresses the design challenges and is divided into four subsections,
including: (i) Subsection 4.1 discusses interface design and practical design issues;
(ii) Subsection 4.2 introduces MOSA concept; (iii) Subsection 4.3 presents DOD
MOSA guidance and the U.S. Naval Open Architecture (NOA); and (iv) Subsection
4.4 discusses MOSA tools and approach for MOSA implementation that addresses
the design issues identified in Subsection 4.1.

4.1 Interface design and practical design challenges

The interfaces between satellite subsystem components can be SW, HW or MW

interfaces. The design and build of these interfaces are well incorporated into any
satellite subsystem components “Design Product” and associated “Design Process”.
The Design Product includes System Architecture, Interface Product, Indepen-
dent Verification and Validation (IV&V) Test Plan, Schedule, Design Approach,
Acceptance Criteria, and System-Built Product.
For MOSA, the Design Process is expected to incorporate MOSA into: Architec-
ture Process, Interface Management, IV&V Process, and System Engineering and
Integration (SE&I) Process. The Interface Product and its open interface design
using MOSA along with the Interface Management are the key challenges in the
development of open-and-modular satellite systems. The key design challenges for
the design and build of open-and-modular satellite systems are:

• Challenge 1: Determination of Key Open Subsystem (KOSS): This is also

known as KOSS Selection. Ideally, all modular subsystem components should
be made open. But this is not practical, because some interfaces need to be
customized using close interface design due to weight, size and power
reduction requirements. The key challenge here is to identify a set of criteria
that can be used for KOSS selection. Subsections 4.3 and 4.4 will address this
challenge.

• Challenge 2: Designation of Key Interfaces for the Selected KOSS: Satellite

system designers need to identify a subset of selected set of KOSS components
that can be designated as key interfaces. The key challenge here is to identify a
business case for the designated key interfaces. Subsections 4.3 and 4.4 will
describe selection criteria and tool to address this challenge.

• Challenge 3: Selection of Open Standards for the Designated Key Interfaces:

Selection of the popular and open standards for the designated key interfaces is
also a potential challenge for the designers. The selection should be based on
the cost, required technical specification and availability of “open” products in
the market and their usage by space industry. SubSection 3.3 above provides a
list of some existing open and popular standards. Subsection 4.4 will discuss
how to resolve this challenge by developing a business case to justify the
selection of key interfaces and associated open standards.

• Challenge 4: Management of Key Open Interfaces: Not all identified key

interfaces in Challenge 2 can be designated an open interface standard at the
initial system design phase due to market unavailability. Hence, managing
these interfaces can be a potential challenge ensuring that they will be “open”

13
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 8.
MOSA and U.S. NOA approach.

by the time of Full Operational Capability (FOC) deployment. Section 4.3

discusses DOD guidance for managing key interfaces for military satellite
system development.

4.2 Introduction to MOSA

U.S. DOD recommends OSA design using MOSA principles for future military
satellite system development with a goal to achieve a balance between business and
technical objectives that make a business sense in terms of (i) increase competition
and lower system acquisition cost, and (ii) lower sustainment cost over its life cycle.
MOSA design approach requires to implement five MOSA principles, including two
Business (B) and three Technical (T) principles [1, 2]. Figure 8 captures these five B
and T principles. Recently, U.S. Navy augmented MOSA principles with addition
five Naval Open Architecture (NOA) principles, including two business and three
technical principles as shown in Figure 8 [3].
Subsections 2.1 and 2.2 provide current implementation of the Technical princi-
ple 1 (T1) for the modular design of satellite Bus and mission PL, respectively. The
remaining Subsections 4.3 and 4.4 discuss the implementation of T2, T6, Business
principle 3 (B3), and B4 using DOD guidance for addressing the challenges
presented in Subsection 4.1.

4.3 DOD guidance

MOSA mandated the space system technical requirements be based on the

maximum extent practicable on open standards as indicated in Section 3.2 of the U.
S. DOD Guidebook for Program Managers [1]. The book provides MOSA2 guide-
lines and contract language for generating a Request for Proposal (RFP) [1]. At the
minimum, the RFP shall incorporate the following MOSA tasks that can help to
minimize the MOSA implementation risk in the design, build and test of new
satellite systems:

2
Note that the term Open System Architecture (OSA) has also been used interchangeably with MOSA
by U.S. DOD.

14
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

• Design the open system architecture using open interfaces. Implement the open
interfaces using open standards for connecting HW-to-HW, and SW-to-SW.

• The satellite system design shall accommodate growth and provide open
interface standards to allow future reconfiguration and addition of new
capabilities without large-scale redesign of the system.

• Develop a capability roadmap for the system covering the life of the system
following the completion of the rapid prototyping contract phase.

• Address “Commercial Off the Shelf/Non-Developmental Item and Open

System Software Licenses,” including Open Source Software, Verification of
Open Architecture, Modular Open Systems Approach Metrics to be Reported,
Modular Open System Approach Analysis Report.

• Generate Open System Management Plan (OSMP) to capture all MOSA

activities, technology roadmaps; Define and track MOSA metrics; Update
roadmap. Following is a list of MOSA metrics that should be used to
demonstrate an open satellite system:

Number Open Key Interfaces

◦ Percentage ð%Þ Open Key Interfaces ¼ Total Number Key interfaces

◦ Number and location of private extensions on open interfaces;

◦ Contractor use of company private extensions on open standard

middleware;

◦ Open Software Design Tool Kits/Component Design Tool Kits (OSDTK/

CDTK) will be provided with a minimum of Government Purpose Right
(GPR); Minimal license fees may apply for COTS items;

◦ Percentage of Chief Engineers, IPT Leads and program team members on

architecture, software, logistics and Test & Evaluation trained in Open
Systems Architecture and the MOSA tools;

◦ Future Competition Strategy included in the OA Business plan within the

OSMP;

◦ MOSA (or OSA) requirements flowed down to sub-tier suppliers and

recorded in IBM Rational® DOORS® requirements database or an MBSE
digital model.

◦ Design a system that consists of hierarchical collections of software,

hardware, and firmware Configuration Items (CI’s). Document in the
MOSA Analysis Report its modularization choices for the system design
and any tradeoffs performed in accordance with the OA verification plan.

◦ Document any processes or applications necessary to support MOSA in

the MOSA Analysis Report.

The above U.S. DOD’s guidance encourages the satellite system designers to
consider the above MOSA items in the design and build of the modular and open

15
Satellite Systems - Design, Modeling, Simulation and Analysis

satellite Bus and mission payload for future space systems. The following section
presents a proposal for assisting the satellite system designers to implement these
MOSA items along with assessment tools provided by U.S. DOD.

4.4 MOSA implementation and assessment tools

It is observed that the U.S. DOD, U.S. civilian agencies (e.g., NASA, NOAA, etc)
and U.S. satellite manufacturers/suppliers (e.g., LM, Boeing, Northrop Grumman
(NG), Raytheon, L3, etc) are investigating approaches for the modular and open
design and build of satellite Busses and mission PL’s using MOSA modular and open
design principles. Figure 9 proposes an approach to design and build of future
modular and open satellite Busses and mission PLs, and allowing the satellite buyers
to: (i) Buy the satellite Bus (see Path A of the figure) and mission PL (see Path B)
from different satellite manufacturers/suppliers, (ii) Have an option to choose a
third satellite vendor to integrate the satellite Bus and mission PL (see Path C).
The proposed MOSA implementation approach shown in Figure 4 consists of six
basic steps that are incorporated into three execution paths, namely, Path A, Path B
and Path C:

• Path A is for the satellite Bus manufacturer/supplier. This path has three basic
steps:

◦ Step I-A: Develop Modular satellite Bus Architecture (MoBA). The MoBA
subsystem components are described in Sections 1 and 2 (see Figure 3).

◦ Step II-A: Designate KOSS’s and select open standards for all internal
satellite Bus subsystem components. Open interface standards selection
and designation of KOSS are discussed in Sections 3 and 4.

Figure 9.
Proposed implementation approach for design and build of satellite Systems allowing buyers to acquire satellite
bus and Mission PL independently.

16
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

◦ Step III-A: Design and build Open Modular satellite Bus System (OMoBS).
This step is achieved by identifying all potential KOSS’s from the satellite
Bus to any mission PL’s, i.e., the selected KOSS’s should be independent of
mission types. The satellite Bus manufacturer is responsible for
integrating all Bus components and have the satellite Bus ready for sale.

• Path B is for the mission PL manufacturer/supplier. This path also has three
basic steps that are similar to Path A:

◦ Step I-B: Develop Modular Mission PL Architecture (MoPA). The MoPA

subsystem components are also described in Sections 1 and 2 (see
Figure 4).

◦ Step II-B: Designate KOSS’s and select open standards for all internal
Mission PL subsystem components. Open interface standards selection
and designation of KOSS for mission PL are also discussed in Sections 3
and 4.

◦ Step III-B: Design and build Open Modular Mission PL System (OMoPS).
This step is achieved by identifying all potential KOSS’s from the any
mission PL’s to satellite Bus, i.e., the selected KOSS’s should be
independent of mission types. The mission PL manufacturer is responsible
for integrating all mission PL components and have the PL ready for sale.

• Path C is for the satellite system integrator. This path has additional three new
steps:

◦ Step IV: The system integrator works with satellite Bus and mission PL
manufacturers to develop a satellite system interface specification
specifying all “open” and “close” interfaces between the mission PL-and-
satellite Bus. All open interfaces between the mission PL-and-satellite Bus
shall be selected to meet the business and performance objectives
approved by the buyer. The system integrator performs satellite Bus and
mission PL integration using the approved interface specification.

◦ Step V: System integrator performs system test and verification subject to

buyer’s approval.

◦ Step VI: System integrator delivers the satellite system to the buyer.

DOD has also developed MOSA tools to assist MOSA implementation and
assessment of military satellite Bus and mission PL “Openness”. These tools can also
be used for civilian and commercial applications. The DOD tools include MOSA
Program Assessment and Rating Tool (PART), Open Architecture Assessment Tool
(OAAT), and Key Open SubSystem (KOSS) Tool:

• MOSA PART3: It is being used by DOD as the standard MOSA program

assessment and rating tool for DOD space system programs.

3
PART can be found from: https://www.dau.mil/cop/mosa/Lists/Tools/DispForm.aspx? ID=2&Conte
ntTypeId =0x01002BC08FCA204040449CF11CB472BEEE1800AA6D1BC9926604469A02D
DB936F94D1F

17
Satellite Systems - Design, Modeling, Simulation and Analysis

• MOSA OAAT4: Assist U.S. Navy program managers in assessing the

“openness” of their programs. It aligns to the Open Architecture Assessment
Model (OAAM) as approved by Assistant Secretary of The Navy (ASN) for
Research, Development and Acquisition (RDA), which serves as the Navy
Acquisition Executive. Other DOD agencies have also been using OAAT since
the tool can provide a reproducible and objective method of conducting
program assessments.

• MOSA KOSS Tool5: One of the key MOSA principles is the Business Principle
number 4, namely, Designate Key Interfaces (see Figure 8, B4). The
identification of KOSS’s is an important task in realizing open systems. This
MOSA principle requires the system designers to compromise between cost
and performance by selecting a set of KOSS’s with their associated interfaces
that can be assigned widely used open standards allowing for easy and
affordable update and frequent refresh. MOSA KOSS tool provides guidance
for KOSS’s identification and selection. The tool makes use of system capability
road map, system requirements and Subject Matter Expert (SME), program’s
sponsor and warfighter knowledge to identify the system/subsystem
components expected to have a high volatility over the system life cycle. The
tool specifies the key interfaces as those either side of volatile components. The
tool will help the satellite system designer to identify and rank KOSS’s
components that will meet both programmatic and technical requirements.

5. Future resilient and robust satellite system architectures

This section demonstrates how to use Steps II-A and II-B of the proposed MOSA
implementation approach presented in Section 4.4 for the design and build of
future resilient and robust satellite systems. Subsections 5.1 and 5.2 present poten-
tial modular-and-open satellite Bus and mission PL architectures, respectively.

5.1 A potential modular open satellite bus architecture solution

To demonstrate how to transition the notional modular satellite Bus system

architecture presented in Figure 3 to a modular-and-open satellite Bus architecture,
this subsection provides an example for the transition of three modular Bus sub-
systems, namely, BC&DH (Bus Subsystem 3), BTT&C (Bus Subsystem 4) and
BEPS (Bus Subsystem 5). These modular Bus subsystems are decomposed to
subsystem component-level and analyzed for consideration as potential KOSS’s for
open interface standardization. Table 1 summarizes the decomposition and analysis
results for these three satellite Bus subsystems.
In practice, the preliminary KOSS analysis results shown in Table 1 should be
finalized by the system designer using DOD KOSS tool discussed in Section 4.4. As
shown in Table 1, standardizing the BC&DH data interfaces will probably provide
the biggest return on investment since the BC&DH subsystem interfaces with each
onboard system. Incorporation of the timing interface along with the data interface
will minimize the amount of connections, thus reducing overall system mass. Any

4
OAAT from: https://www.dau.mil/cop/mosa/Lists/Tools/DispForm.aspx?ID=1&ContentTypeId=
0x01002 BC08FCA204040449CF11CB472BEEE1800AA6D1BC9926604469A02DDB936F94D1F
5
KOSS from: https://acc.dau.mil/adl/enUS/317012/file/46502/KOSS%20Overview_FINAL_5Aug09.pdf.

18
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

Modular satellite Modular satellite Bus subsystem Recommendation for open interface
Bus subsystem and component description Standardization (potential KOSS)
component

BC&DHS Bus Command & Data Handing

Component No. Subsystem (C&DHS)

BC&DHS-1 Command Authentication Processing Recommend for Interface standardization

Unit (Sync Word Frame Lock,
Unparsed Command)

BC&DHS-2 System Timing Unit

BC&DHS-3 Fault Management Processing Unit Not recommended for interface

(Execute Stored CMD Sequence, standardization due to many variations
Monitor System Health) between systems

BC&DHS-4 Bus Resource Management Recommend for open interface

Processing Unit (Managing Internal standardization
and External Bus Data)

BC&DHS-5 Memory Storage Unit

BC&DHS-6 Spacecraft Control Processor

BC&DHS-7 Bus Telemetry Conditioning Not required; software driven functions.

Processor Should be considered in software
interface analysis.

BC&DHS-8 Bus Cyber Security Unit Recommend for open interface

standardization

BTT&CS Bus Tracking-Telemetry &

Component No. Command Subsystem (TT&CS)

BTT&CS-1 TT&C Waveforms/MODEM Recommend for open interface

standardization

BTT&CS-2 TT&C Antenna Assembly for S- Not recommended for interface

Band/L-Band standardization.

BTT&CS-3 TT&C RF Front-End and Back-End Recommend for open interface

Assembly standardization

BTT&CS-4 Unified S-Band (USB) RX/TX

Assembly

BTT&CS-5 SGLS S-Band RX/TX Assembly

BTT&CS-6 SGLS Base Band Signal Processing Recommend for open interface
(USB Mode1, 2) standardization

BTT&CS-7 In Band TT&C Processor Located at Not recommended for interface

Private Station standardization due to many variations
between systems

BTT&CS-8 Power Controller Assembly Recommend for open interface

standardization

BEPS Component Bus Electrical Power Subsystem

No. (EPS)

BEPS-1 Solar Array (SA) Recommend for open Interface

standardization.

BEPS-2 Battery Assembly (BA) Not recommended for interface

standardization; Battery size will vary
depending on the mission profile.
Additional batteries could potentially
require customized interfaces to tie all
batteries to power bus.

19
Satellite Systems - Design, Modeling, Simulation and Analysis

Modular satellite Modular satellite Bus subsystem Recommendation for open interface
Bus subsystem and component description Standardization (potential KOSS)
component

BEPS-3 Solar Array Drive Assembly (SADA) Recommend for open Interface
standardization.

BEPS-4 Transient Filter Unit (TFU) Not recommended for interface

standardization

BEPS-5 Bus Power Regulation Unit (BPRU) Recommend for open Interface
standardization.
BEPS-6 Fuse Box Assembly (FBA)
.
BEPS-7 Pyro Relay Assembly (PRA)

Table 1.
Satellite bus subsystems decomposition and potential KOSS.

interfaces that require a significant amount of analysis or Non-Recurring Engineering

(NRE) hours is not a good candidate for standardization. The fault management
processing interface is in this category, and it is not recommended for standardization.

5.2 A potential modular open Mission payload architecture solution

This subsection provides an example for the transition of the notional modular
mission PL architecture presented in Figure 4 to a modular-and-open mission PL
architecture. Table 2 summarizes the decomposition and KOSS analysis results for
four mission PL subsystems, including PAS (PL Subsystem 1), CPCom-RFS (PL
Subsystem 2), PDPS (PL Subsystem 3) and PFTS (PL Subsystem 11).
The mission PL digital processing system is not recommended for interface
standardization due to many variations between systems and subsystems. Multi-RF
Wideband RX Up/Down Converters and Tunable IF Down Converters require a
significant amount of analysis or NRE hours and are also not a good candidate for
standardization. Again, DOD KOSS tool should be used to finalize the KOSS analysis
results presented here for actual design and build of the satellite systems.

6. Conclusion

The chapter provides an overview of existing modular satellite Bus and mission
PL architectures and associated standards for communication data Busses. The
chapter defines open and close interfaces along with industry approved popular
standards and discusses the interface design challenges. Moreover, the chapter pro-
vides an overview of MOSA and related DOD guidance and assessment tools to
address the interface design challenges. Examples for the design and build of
modular-and-open satellite Bus and mission PL architectures are also presented.
The intent of this chapter is to provide an innovative approach for the satellite
system designer to design and build of the next generation satellite achieving a
balance between business and technical objectives that make a business sense for
both the satellite manufacturers and buyers in terms of lower system acquisition
and sustainment costs over its life cycle. The MOSA implementation approach
presented here allows the satellite manufacturers to build the satellite Bus and
mission PL separately for more production, flexibility, and market competition.
Concurrently, the approach also allows the satellite buyers to buy satellite Bus at
high volume with reduced unit costs and less schedule risk. Another benefit for the

20
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

Modular mission Modular mission PL Recommendation for open interface

PL subsystem subsystem and component standardization (potential KOSS)
component description

PAS Component No. PL Antenna Subsystem

(PAS)

PAS-1 PL RF Antenna Not recommended for interface standardization

Configuration (PRAC) due to many variations between systems

PAS-2 Beam Forming Unit (BU)

PAS-3 Antenna Controller (AC) Recommend for Open Interface standardization

PCom-RFS PL Com RF Front-

Component No. End/Back-End Subsystem
(CPCom-RFS)

PCom-RFS-1 PL LNA Component Not recommended for interface standardization

due to many variations between systems/
PCom-RFS-2 PL HPA Component
subsystems
PCom-RFS-3 Multi-RF Wideband
Receiver (RX)

PcCm-RFS-4 PL Up/Down Converters Recommend for Open Interface standardization

PCom-RFS-5 Tunable IF Down

Converters

PDPS Component PL Digital Processing

No. Subsystem (PDPS)

PDPS-1 PL ADC/DAC Not recommended for interface standardization

due to many variations between systems
PDPS-2 FPGA Processor

PDPS-3 PL MOD and DEMO

(Optional)

PDPS-4 Digital Network Switch

(Optional)

PDPS-5 PL System Controller

PTFS Component P/L Frequency and Timing

No. Subsystem (PFTS)

PFTS-1 Atomic Clock Unit (ACU) Not recommended for interface standardization;
ACU and CM&CU will vary depending on
PFTS-2 Clock Monitoring & Control
mission type and mission requirements
Unit (CM&CU)

PFTS-3 Frequency Generation & Recommended for Interface standardization.

Up conversion Unit

PFTS-4 Timing Variation and Not recommended for interface standardization

Frequency Stability

Table 2.
Mission payload subsystems decomposition and potential KOSS.

satellite buyer is the adaptability of changing the requirements on the mission PL

without impacting the satellite Bus.

Acknowledgements

Although the preparation of this work was not funded by The Aerospace Cor-
poration, but the author acknowledges the work presented in this chapter was based

21
Satellite Systems - Design, Modeling, Simulation and Analysis

on his knowledges accumulated over the years from many space programs at The
Aerospace Corporation, Raytheon and Jet Propulsion Laboratory. In addition, the
author would like to express his appreciation to Aerospace’s manager, Ms. Navneet
Mezcciani, for her professional support.

Conflict of interest

The preparation of this chapter was not funded by the Aerospace Corporation,
and it was done by the authors using his own time and resources, thus it does not
represent the Aerospace Corporation’s view on the DOD guidance for MOSA
implementation and the proposed system architecture solutions.

Notes/Thanks/Other declarations

The author wishes to thank his wife, Thu-Hang Nguyen, for her enormous
patience and boundless support during the preparation of this chapter.

Author details

Tien M. Nguyen1,2

1 California State University, Fullerton, USA

2 The Aerospace Corporation, El Segundo, USA

*Address all correspondence to: [email protected]; [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

22
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

References

[1] U.S. DOD Open Systems Architecture ATR-2015-00535. El Segundo, California:

Contract Guidebook for Program The Aerospace Corporation; 2015
Managers, v.1.1, June 2013. Available
from: https://www.dau.edu/ [7] Boeing. Boeing Satellites. 2015.
cop/pm/layouts/15/WopiFrame.aspx? Available from: http://www.boeing.
sourcedoc=/cop/pm/DAU%20Sponsored com/space/boeing-satellite-family/
%20Documents/DoD%20Open%
20System%20Architecture%20(OSA)% [8] BAE Systems. Processors. 2015.
20Contract%20Guidebook%20for% Available from: http://www.baesystems.
20Program%20Managers-version% com/down load/BAES_161071/
201.1-%20June%20%202013.pdf& processors
action=default&DefaultItemOpen=1
[9] GAO. Space Acquisitions – Space
[2] Zimmerman P. Modular open Systems Based Infared Systems Could Benefit
approaches (MOSA) panel on standards. from Technology Insertion Planning.
In: Office of the Deputy Assistant 2015. Available from: http://www.gao.
Secretary of Defense on Systems Engrg, gov/assets/670/669454.pdf
Defense Standardization Program
Workshop. 12 July 2018. pp. 1-7 [10] GPS. Global Positioning System
Constellation Replenishment. 2015.
[3] U.S. Department of Navy, Naval Available from: http://www.gps.gov/
Open Architecture Contract Guidebook congress/reports/2015/2015-02-AF-
for Program Managers, V.2.0. 30 June report-GPS-constellation-replenishme
2010. Available from: http://www.dtic. nt.pdf
mil/dtic/tr/fulltext/u2/a550060.pdf
[11] Martin L. A2100. 2015. Available
[4] Acquiring and Enforcing the from: http://www.lockheedmartin.com/
Government’s Rights in Technical Data content/dam/lockheed/data/space/
and Computer Software Under documents/a2100/A2100brochure.pdf
Department of Defense Contracts: A
Practical Handbook for Acquisition [12] L Martin. Advanced Extremely High
Professionals. 7th ed. Los Angeles, CA: Frequency (AEHF). 2015. Available
Office of the Staff Judge Advocate, from: http://lockheedmartin.com/us/
Space and Missile Systems Center products/advanced-extremely-high-
(SMC), U.S. Air Force Base; August frequency–aehf.html
2015. Available from: https://www.acq.
osd.mil/dpap/cpic/cp/docs/Technical_ [13] Martin L. Global Positioning System
Data_and_Computer_Software_Rights_ (GPS). 2015. Available from: http://
Handbook_7th_Edition.pdf www.lockheedmartin.com/us/products/
gps.html
[5] Revillon P. Fundementals and
dynamics of the satellite [14] Loral. 1300 Series Satellite Platform.
communications business. In: 2016 2015. Available from: http://www.
EuroConsult for Inmarsat Capital sslmda.com/html/products/1300.html
Markets Day. Available from: www.
euroconsult-ec.com [15] National Aeronautical and Space
Administration (NASA). Tracking and
[6] Davis LA, Filip L. How long does it Data Relay Satellite (TDRS) Third
take to develop and launch government Generation Capabilities. 2013. Available
satellite systems? In: Aerospace Report No. from: http://www.nasa.gov/directorates/

23
Satellite Systems - Design, Modeling, Simulation and Analysis

heo/scan/services/networks/txt_ Available from: ftp.estec.esa.nl/pub/ws/

tdrs_gen3.html OBCDS_Workshop/Proceedings.htm

[16] Collins R. Momentum and Reaction [24] Consultative Committee for Space
Wheels. Available from: https://www. Data Systems (CCSDS).
rockwellcollins.com/Data/Products/ Recommendations for Space Data
Space_Components/Satellite_Stabiliza Systems, Packet Telemetry, CCSDS
tion_Wheels/RSI_12_Momentum_and_ 102.0-B-5, Blue Book. 2000. Available
Reaction_Wheels.aspx from: https://public.ccsds.org/Pubs/
102x0b5s.pdf#search=packet%20
[17] Gwaltney DA, Briscoe JM. telecommand
Comparison of Communication
Architectures for Spacecraft Modular [25] Consultative Committee for Space
Avionics Systems. NASA/TM—2006– Data Systems (CCSDS).
214431. June 2006. NASA Center for Recommendations for Space Data
AeroSpace Information Systems, Telecommand – Part 1:
Channel Service, CCSDS 201.0-B-3,
[18] SpaceWire—Links, Nodes, Routers, Blue Book. 2000. Available from:
and Networks, ECSS–E–50–12A. 24 https://public.ccsds.org/Pubs/201x0b3s.
January 2003. European Cooperation for pdf#search=packet%20telecommand
Space Standardization
[26] Space-Ground Link Subsystem
[19] NASA Common Instrument (SGLS). Ground Station, System
Interface Project. Hosted Payload Analysis Summary Report, WDL-
Guidelines Document. Document TR3227-1. Vol. 1. 15 November 1968.
Number: CII-CI-0001, Version Rev. A., Philco-Ford Corporation. Available
4 November 2013. Earth System Science from: https://apps.dtic.mil/
Pathfinder Program Office dtic/tr/fulltext/u2/853122.pdf

[20] NASA Common Instrument [27] Hearn M. MIL-STD-1553B: The Past

Interface Project. Hosted Payload Guide and Future Data Bus. 2019. High
for Proposers. Document Number: Frequency Electronics Magazine.
HPIG0001, Version: Initial. 22 March Available from: https://www.highfreque
2018 ncy electronics.com/index.php?
option=com_content&view=article&
[21] Space and Missile Systems Center. id=1955:mil-std-1553b-the-past-and-f
Electromagnetic Compatibility (EMC) uture-data-bus&catid=161&Itemid=189
Requirements for Space Equipment and
Systems, SMC Standard SMC-S-008. 13 [28] Department of Defense Interface
June 2008. Air Force Space Command Standards for Digital Time Division
Command/Response Multiplex Data
[22] NASA. Electrical Grounding Bus, MIL-STD-1553B. 21 September
Architecture for Unmanned Spacecraft, 1975, Department of Defense,
NASA HDBK-4001. 17 February 1998. Washington D.C. 20360
NASA Technical Handbook
[29] AeroFlex, SpaceWire 101, Webex
[23] Plancke P. Spacecraft Digital Busses Seminar. 15 February 2006. Available
and Interfaces Selection and from: www.aeroflex.com/spacewire
Standardization Status. 19–20
September 2001. On-Board Payload [30] ESA Publications Division.
Data Processing Workshop ESTEC. SpaceWire Standard Document

24
Future Satellite System Architectures and Practical Design Issues: An Overview
DOI: http://dx.doi.org/10.5772/intechopen.92308

ECSS-E-ST-50-12C. The Netherlands; 31

July 2008

[31] Los Angeles Air Force Base

(LAAFB). Space and Missile Systems
Center (SMC), Common Payload
Interface Specification Project
(CoPaIS), SMC18-069. 5 April 2018.
Available from: https://govtribe.com/
opportunity/federal-contract-
opportunity/common-payload-
interface-specification-project-
copais-smc18069

25
Chapter

System Designs of Microsatellites:

A Review of Two Schools of
Thoughts
Triharjanto Robertus

Abstract

Microsatellite has been considered as disruptive technologies in satellite

engineering. Its development cost and time provide advantages for new kind of
Earth observations, telecommunications, and science missions. The increasing
trend of microsatellite launches and operations means that the approach was so suc-
cessful that it could create funding sustainability. Major contributing factors of its
success were due to the system design of the microsatellites. This chapter discusses
two microsatellite system design approaches, namely Technical University of Berlin
heritage and University of Surrey heritage. Both Universities provide approaches for
system design and build of microsatellite systems. The design approaches are being
compared along with lessons learned. The choices of microsatellites to be compared
in this chapter will be those that are manufactured about the same time such that
the technology compared is mostly the same and flown in-orbit. The chapter shows
that the differences between the two system design approaches are on the choice of
main computer and associated link configuration and in the attitude control modes.
Another major different is in the satellites’ structure design. For some satellite’s
components, incoming technologies have made the design choices from the two
schools of thoughts converged.

Keywords: satellite design, system design, microsatellites, TU Berlin,

University of Surrey

1. Introduction

Microsatellite has typical weight between 20 and 170 kg at launch as auxiliary

payload. It is initially made as technology experiment and education tools by univer-
sities. Nowadays, microsatellite becomes a common space platform for commercials
and emerging space nations. The commercial mission is typically Earth observation,
data collecting platform (text-based communication), including ships and aircraft
tracking. Studies done by Swartout [1] show that between 2009 and 2012, about 8–12
satellites with mass above 50 kg as auxiliary payload were launched yearly. The data
also show that the trend seems to be steady. Bunchen and De Pasquale [2] noted that
105 satellites with mass of 11–50 kg were launched between 2000 and 2013.
Surrey Space Technology Limited (SSTL), a subsidiary company under
University of Surrey, is one of the companies that initiated the use microsatellite
technology as commercial Earth observation satellite platform. It built a constella-
tion of five satellites named Disaster Monitoring Constellation (DMC) in 2003, with

1
Satellite Systems - Design, Modeling, Simulation and Analysis

payload of 3-band multispectral imager of 30-m resolution, which was intended for
wide-swath land coverage imaging. After the first constellations decommissioned,
it built the second generation with better resolution (20 m). The first launch of
DMC-2 constellation was done in 2009 [3].
Since 2013, Skybox/Skysat has deployed 15 satellites that carry 1-m panchromatic
imager and 2-m 4-band multispectral imager [4]. Unlike DMC, which mission objec-
tives are to observe wide areas with nadir pointing scanning mode, it aims to provide
frequent repeat very high resolution images using massive numbers of highly maneu-
verable satellites. Another commercial Earth observation microsatellite constellation
mission is prepared by Axelspace. The company planned to have 50 satellites launched
starting 2017. The satellite carries imager with 2.5-m panchromatic and 5-m multi-
spectral [5, 6]. Figure 1 shows the configurations of the Skybox and Grus satellites,
which show that Skybox uses single lens and parabolic data downlink antenna, while
Grus uses two lenses and horn-type data downlink antenna.
In addition to Earth observation missions, microsatellite constellation also
being built for Low Earth Orbit (LEO) telecommunication mission. OneWeb and
Telesat are two companies that will launch hundreds of microsatellites in coming
years [7, 8].
The use of microsatellites for commercial purposes means that the technol-
ogy is mature enough to ensure good return-of-investment. One of the major
aspects that contribute to the success of microsatellite technology is its system
design. Therefore, the objective of this chapter is to provide insight into micro-
satellite system design. The chapter addresses the question related to limitation
in weight and size, and how the satellite designer manages to meet the mission
requirements.
Out of many microsatellites developers, two system designs of microsatellites,
namely Technical University (TU) Berlin heritage and University of Surrey heri-
tage, are selected for comparison in this chapter, due to their very different design
approaches. To be comparable, the choices of microsatellite system to be compared
are the ones that manufactured about the same time, so that the technology avail-
able is mostly the same. The microsatellites also have to have in-orbit experience, so
its design success can be measured. Data mining resulted that the satellite operation
year chosen is between 1999 and to date. For TU Berlin system, the choices are DLR-
TUBSAT, MAROC-TUBSAT, Indonesian LAPAN-TUBSAT, LAPAN-ORARI, and
LAPAN-IPB. Meanwhile, for University of Surrey system, the choices are Korean
KITSAT-3, STSat-1 and STSat-3, as well as Turkish BILSAT-1 and RASAT.

Figure 1.
Google Skybox satellite and Axelspace’s Gruz satellite design.

2
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

This chapter is divided into five sections, with the first section introducing the
background and objectives of the chapter. The second section explains how the
satellite design samples for the University of Surrey heritage were selected, and
what satellite design parameters were used in the comparison. Section 3 displays
the satellite design parameters for TU Berlin heritage. Section 4 provides analysis
from the comparison of the two-design heritage, in term of parameters noted in the
previous two sections. Section 5 summarizes the analysis and provides recommen-
dation for further studies regarding the subject.

2. University of Surrey heritages

University of Surrey is known as one of the pioneers in the design and build of
microsatellite in the 1990s. It started launching microsatellite in 1991 with amateur

KITSAT-3 BILSAT-1
Operation 1999–2003 2003–2006
Bus

Solar panel 3 GaAs (2 deployable) @ 50 × 85 4 GaAs @ 60 × 60 cm (58 W)

cm, (150 W)
Battery NiCd; 10 V; 8 Ah NiCd; 28 V; 4 Ah
Reaction wheel/Gyro 3 + 1 Teldix DR01/FO laser 4 SSTL/MEMS
Thruster — Pressurized gas + resistojet
Star sensor 1 2 Altair

Sun sensor 2 axis 4 × 2 axis

Horizon sensor 2 axis
Magnetotorquer/meter 3-axis air coils/3-axis fluxgate 3-axis air coils/2 × 3-axis
fluxgate
Telemetry, Tracking, and VHF uplink; UHF downlink S-band
Commanding (TTC)
Data TX S-band 3.3 Mbps S-band 8 Mbps
Main computer/link config. 2 × microprocessor/CAN 2 × microprocessor/CAN
Attitude control computer 1 1
Payload data handling Microprocessor based FPGA based

GPS — SSTL SGR

Payload
3-band imager w/ 570-mm lens 2 × 3-band imager w/150-mm
lens

Radiation dose sensor Pan imager w/300 mm lens

High energy particle sensor Store and forward

communications
Scientific class magnetometer 8-band low resolution imager

CMG
Size (cm) 50 × 60 × 85 60 × 60 × 60

Mass (kg) 110 130

Table 1.
Sample for the University of Surrey microsatellite system design.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

radio missions. To simplify the satellite design, the first microsatellite generation
has passive attitude control system, that is, using gravity gradient telescopic boom.
The university provided microsatellite development and building capabilities
to many emerging space countries, including Thailand, Malaysia, South Korea,
Algiers, Turkey, and Nigeria. At the time, such countries started to use remote sens-
ing satellites, mostly from the United States and European, for various land-based
applications. Therefore, they required remote sensing payloads to include in their
satellite missions. Such mission elevates the design requirements to active attitude
control system and higher data rate downlink system.
Thailand’s Mahanakorn University collaborated with the University of Surrey
to jointly develop TMSat that was launched in 1998 [9]. TMSat focuses on remote

STSAT-1 STSAT-3 RASAT

2003–2008 2013–2015 2011–2017
Bus

Solar panel 3 GaAs (2 deployable); 3 GaAs (2 4 GaAs; 52 W

160 W deployabl
e); 275 W

Battery NiCd; 14 V; 12 Ah Li-ion; 20 V; Li-ion; 28 V; 9 Ah

20 Ah
Reaction wheel/Gyro 4 /FO laser 4 /FO laser 4 /MEMS
Thruster — Hall thrust —
Star sensor 1 2 SaTReC 1

Sun sensor 4 panels +2 cell Coarse and fine 4 analog

Horizon sensor — — —
Magnetotorquer/meter 3-axis/3-axis fluxgate 3-axis/3-axis 3-axis/2 × 3-axis
fluxgate
TTC S-band S-band S-band (primary) and
UHF/VHF (emergency)

Data TX X-band 3.2 Mbps X-band 10 Mbps X-band 100 Mbps

Main computer/link Microprocessor/CAN Leon2-FT (triple 2 × microprocessor/
config. redundancy)/CAN CAN and space wire
and space wire
Attitude control 1 1 AIU (attitude 1
computer interface unit)
Payload data handling FPGA based FPGA based FPGA based
GPS 1 1 1
Payload

Far UV imaging 2× Multiband IR Pan imager w/840 mm

spectrograph imagers lens
Space physic sensor Spectrometer 3-band imager
w/420 mm lens
Data collection system
Size (cm) 66 × 55 × 83 102 × 103 × 88 70 × 70 × 55.4
Mass (kg) 106 175 95

Table 2.
Sample for the University of Surrey microsatellite heritage system design.

4
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

sensing and amateur radio mission. Since Thailand did not continue building its
subsequent satellites, TMSat is not selected as satellite design heritage sample in
this chapter.
Singapore’s Nanyang Technology University (NTU) collaborated with
the University of Surrey to jointly develop satellite subsystem for UoSAT-12.
However, the satellite is not a microclass and therefore is not selected as a
sample for the University of Surrey’s satellite system design in this chapter.
The satellite subsystem from NTU is a communication payload with S-band
downlink and L-band uplink, which provides the Internet protocol communica-
tion operating at 1 Mbps. Since the experience with the University of Surrey
only in subsystem design and development, the subsequent NTU satellite,
that is, XSAT, is also not considered as the University of Surrey heritage
satellite [10, 11].
South Korean experience with the University of Surrey satellite design is
when Satellite Technology Research Center (SaTReC), an institution under Korea
Advanced Institute of Science and Technology (KAIST), jointly built KITSAT-1
and KITSAT-2 and launched it in 1992 and 1993. Both satellites have store-forward
communication amateur payload and low-resolution imagers. Since the KITSAT-1
and KITSAT-2 development time does not match with other microsatellite design
sample, only the design of KITSAT-3 is used in this chapter. SaTReC then devel-
oped STSAT series as its second generation microsatellites. Since STSAT-2 expe-
rienced launch failure, only STSAT-1 and STSAT-3 are selected as satellite design
samples [12–15].
Turkey’s experience with the University of Surrey satellite design is when its
space research institute, TUBITAK-UZAY (previously named BILTEN TUBITAK-
ODTU), jointly developed BILSAT-1. The satellite was part of DMC-1 constellation
[16–19]. After BILSAT-1, the institute then built its second generation microsatel-
lite, RASAT. Therefore, both microsatellites are used as sample for the University of
Surrey design heritage [20–23].
Fifteen satellite bus design parameters are selected for the comparison, including
14 mechatronics component parameters in the satellites’ design. For the University
of Surrey satellite heritage, the parameters are tabulated in Tables 1 and 2. Structure
design from four of the five microsatellites is shown in Figures 2 and 3. Payload
parameters also noted in Tables 1 and 2 to explain the similarity (or differences) in
the mission requirements and their impacts to satellite bus parameters. The weight
and dimensions are, in additional of drawings, noted in to explain the satellite

Figure 2.
Mechanical design of KITSAT-3 and STSAT-3.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 3.
Mechanical design of BILSAT-1 and RASAT.

structure design aspects. The satellite operation years are noted in the tables to show
the context of available technology.
As shown in Figures 2 and 3, the University of Surrey heritage satellites use
electronic trays for its satellite bus electronics. The aluminum trays also function as
load bearing structure, so that the rest of the satellites components, such as reaction
wheels and attitude sensors, can be laid out around them. After all components
integrated, the solar panels and/or other outside panels that are made of lighter
materials can be used to cover the satellites.

3. Technical University of Berlin heritages

Technical University (TU) of Berlin had launched six microsatellites between

1991 and 2007. During such time, the university had provided microsatellite devel-
opment capacity building to Morocco and Indonesia. However, only Indonesia
(Satellite Technology Center) had developed its second generation of microsatellites.
Tables 3 and 4 provide samples of microsatellite systems used for the comparison
considering the development and operation time of the satellites. The microsatel-
lite parameters from the TU Berlin heritage shown in Table 3 are from DLR-
TUBSAT and MAROC-TUBSAT, and in Table 4 are from LAPAN-TUBSAT and two
Indonesian built satellites, that is, LAPAN-ORARI and LAPAN-IPB. Additionally,
the four satellite structure drawings are presented in Figures 4 and 5 [24–29] for the
comparison of structural design.
The author should describe the key differences among the four structure designs
presented in Figures 4 and 5.
As shown in Figures 4 and 5, for the TU Berlin satellite heritage, the com-
ponents are laid out in boxes. For DLR-TUBSAT and Maroc-TUBSAT, they are
modular boxes (ACS, payload, power, etc.). Meanwhile, in LAPAN’s satellite
series, the boxes are integrated in lower and upper compartments of the same
structure. The boxes were made from aluminum plates and therefore function as
load bearing structure. The solar panels are directly attached to the outer part of
the boxes.

6
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

DLR-TUBSAT MAROC-TUBSAT
Launch 1999–2007 2001–2006
Bus

Solar panel 4 Si @32 × 32 cm, (14 W) 4 Si @32 × 32 cm (14 W)

Battery NiH2; 10 V; 12 Ah NiH2; 10 V; 12 Ah
Reaction wheel/Gyro 3 IRE 203/FO laser 3 + 1 IRE 203/FO laser

Thruster — —
Star sensor — IRE
Sun sensor 4 panels +1 cell 6 single cell
Horizon sensor — —
Magnetotorquer/meter 1 axis coil + 1 rod/− 1 axis/3-axis sensor

TTC 2 UHF w/omni antennas 2 UHF w/omni antennas

Data TX S-band analog S-band 256 kbps
Main computer/ link config. 32 bit microcontroller/star 32 bit microcontroller/star

Attitude control computer — —

Payload handling Multiplexer Recorder
GPS — —
Payload
B/W video cam. w/16 mm lens NIR imager w/72 mm lens

B/W video cam. w/50 mm lens

B/W video cam. w/1000 mm lens

Size (cm) 32 × 32 × 32 32 × 34 × 36
Mass (kg) 45 47

Table 3.
Sample for the Technical University of Berlin microsatellite system design.

4. Analysis

4.1 Power generation and storage

Tables 1 and 2 show that the Korean satellites have employed deployable solar
panel (which is also shown in Figure 1), since the mission required high power and
used direct energy transfer (DET) mode. Such approach is very much different than
those used by KITSAT-1 and KITSAT-2, which have body-mounted solar panels.
On the other hand, Turkish satellites use body-mounted solar panels and therefore
do not have the requirement of one side of the satellite always facing the sun for
battery charging.
Tables 3 and 4 show that all TU Berlin heritage use body-mounted solar panels. It
uses Si panels for its first three satellites, then opted to higher capacity GaAs panels in
LAPAN-ORARI and LAPAN-IPB. Generally, the power budget for the University of
Surrey heritage satellites is higher than the TU Berlin heritage, even in the ones with
body-mounted solar panels. As shown in Figure 5, in LAPAN-IPB, one of the sides has
two 46 × 26 cm solar panels. The side is projected to be Sun pointing most of the time.
Battery chosen to be used in the early University of Surrey heritage satellite
design is NiCd, while in TU Berlin’s satellite design is NiH2. NiCd batteries require

7
Satellite Systems - Design, Modeling, Simulation and Analysis

LAPAN-TUBSAT LAPAN-ORARI LAPAN-IPB

2007–2013 2015-now 2016-now
Bus

Solar panel 4 Si @43 × 24 cm, 4 GaAs @46 × 26 cm 5 GaAs @46 × 26 cm

(14 W) (30 W) (30 W)
Battery NiH2; 14 V; 12 Ah Li-ion; 16 V; 19.5 Ah Li-ion; 16 V; 36 Ah
Reaction Wheel/Gyro 3 IRE 203/FO Laser 3 + 1 IRE 303/FO Laser 3 + 1 IRE 303/FO Laser
Thruster — — —

Star sensor Vectronics (VTS) VTS, IRE VTS, LAPAN

Sun sensor 4 panels +2 cells 6 single cells 6 single cells
Horizon sensor — — LAPAN (IR camera
based)
Pitch sensor — — LAPAN (CCD based)

Coil/magnetometer 3 axis/− 3 axis/VFMS-51 3 axis/fluxgate scientific

class
TTC 2 UHF w/ omni 2 UHF w/ omni 2 UHF w/omni antennas
antennas antennas

Data TX S-band analog S-band 5 Mbps X-band 105 Mbps

Main computer/link 32 bit microcontroller/ 32 bit microcontroller/ 32 bit microcontroller/
config. star star star
Attitude control — — —
computer
Payload handling Multiplexer Digital and analog FPGA based
switcher + recorder
GPS — VGPS-51 VGPS-51

Payload
Color video cam. Color video cam. 4-band imager
w/50 mm lens w/1000 mm lens w/300 mm lens

Color video cam. 4 M pix cam. 4 M pix cam.

w/1000 mm lens w/1000 mm lens w/1000 mm lens

AIS (ship monitoring AIS (ship monitoring

system) system)
APRS (amateur text
message)
Amateur voice repeater
Size (cm) 45 × 27.5 × 45 47 × 38 × 50 50 × 57.4 × 42.4
Mass (kg) 54.7 74 115

Table 4.
Sample for the Technical University of Berlin microsatellite heritage system design.

charging controller mechanism ensuring that the battery is completely drained

before being charged. This is because partial charging can induce memory effect,
which can decrease the battery capacity to its last partial charge state. For NiH2
batteries, they tend to have large packaging due to its cylindrical shape, as shown
in DLR-TUBSAT and LAPAN-TUBSAT drawing (Figures 4 and 5), but its charg-
ing mechanism is very simple (can do trickle charging). As soon as Li-ion battery
technology available, both designs opted out Li-ion battery for its easy handling
(no memory effect) and higher power-to-mass ratio.

8
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

Figure 4.
Mechanical design of DLR-TUBSAT and MAROC-TUBSAT.

Figure 5.
Mechanical design of LAPAN-TUBSAT and LAPAN-IPB.

4.2 Main computer

On the choice of main computer, the University of Surrey heritage uses

microprocessor, such as 32-bit PowerPC 603, while the TU Berlin heritage uses
microprocessor, such as 32-bit SH series. Advantage of using microcontroller is
having shorter booting time, so that it can recover quickly in the event of latch-up

9
Satellite Systems - Design, Modeling, Simulation and Analysis

and needs to be restarted. The advantage of microprocessor is its ability to handle

more complex and parallel jobs. To anticipate any anomaly in the operation, the
use of microprocessor is usually done by using redundancy (i.e., a second processor
will take over the operation in the event of anomaly). In the University of Surrey
satellite design heritage, the electronic components are connected to main computer
with dual line of controller area network (CAN). Meanwhile, the TU Berlin satellite
design heritage uses star configuration with dedicated line to each component from
the main computer, using RS232 or 422.

4.3 Attitude control subsystem

Tables 1 and 2 show that the University of Surrey satellite design heritage uses
separate attitude control computer that integrates attitude sensors, including sun
and star sensors with all reaction wheels and gyros. This is done so that the attitude
control system can work in closed loop all the time. Such approach is necessary for
the microsatellite design with deployable solar panels, such as KITSAT-3, STSAT-1,
and STSAT-3 since failure of sun pointing could be disastrous for the satellite. As
shown in Tables 3 and 4, in the TU Berlin satellite design heritage, none of the sat-
ellites have separate attitude control computer. In the design, each reaction wheel-
gyro pair directly connected to the main computer, and therefore, closed loop with
star and sun sensors can only be done using the main computer resources.
Differences are also found in the attitude control sensor between the University
of Surrey design heritage. The Korean microsatellites use fiber-optic gyro, while the
Turkish microsatellites use MEMS gyro. Meanwhile, in all TU Berlin microsatellites,
fiber-optic gyros are used.
For attitude control actuators, all the selected satellites use reaction wheels and
air coils for angular momentum dumping/generation. Figures and data showed
that TU Berlin heritage satellites use reaction wheels in 3-axis configuration. For
LAPAN-ORARI and LAPAN-IPB satellites, they used redundant wheel at satellite
major inertia axis that noted as 3 + 1 as shown in Table 4. For the University of
Surrey heritage satellites, only KITSAT-3 uses reaction wheels in 3-axis configura-
tion. The rest of the satellites uses tetrahedral configuration (noted as 4 as shown in
Table 1).
The TU Berlin’s attitude control design was chosen to reduce computational
burden for filtering out reading noise/jitter in the attitude control sensors. The TU
Berlin heritage satellites offer two options for attitude control mode, in addition to
regular closed loop, including (1) interactive mode for the satellite with video cam-
era payload, such as DLR-TUBSAT and LAPAN-TUBSAT, and (2) angular momen-
tum management mode for the satellite with line imagers, such as Maroc-TUBSAT
and LAPAN-A3. The angular momentum management mode is supported by their
structure design, that is, solid aluminum box, which created maximum inertia
properties at 1 axis and very little cross-product inertias [30, 31]. Such design has
been successfully performed highly stable open-loop angular momentum manage-
ment operation as published by Utama [31] and Mukhayadi [32].

4.4 Propulsion subsystem

From a selected set of satellite designs shown in Tables 1–4, only BILSAT-1
and STSAT-3 have thrusters. The objective for BILSAT-1 thruster is to maintain the
satellite orbit separation in the constellation, so that the image coverage could be
optimized. In STSAT-3, the plasma thruster is part of in-orbit qualification process
for the low power plasma thruster technology developed by KAIST.

10
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

4.5 TTC

For Telemetry and Telecommand, the University of Surrey heritage satellite

stopped using low frequency (UHF and VHF) after KITSAT-3. Such usage in RASAT
is only in emergency situation. Meanwhile, in the TU Berlin heritage, UHF TTC is
still used until LAPAN-IPB. The advantage of using low frequency for TTC is on
its omni-directional antenna. Therefore, the satellite can always be contacted by its
ground station, regardless of its attitude. The cost of the satellite’s control ground
station is also much lower. However, the risk for frequency noise for its operation is
also higher.

4.6 Payload

The payload profiles for both satellite design heritages showed that the platforms
are suitable for both Earth observation, science, and low data rate communication
missions. All of the selected satellites, except Korean STSAT-1, are Earth observa-
tion missions, which are considered important by stakeholder of satellite developer
in Korea, Turkey, and Indonesia. KITSAT-3, BILSAT-1, RASAT, and LAPAN-IPB
are for land cover that can be applied for estimating crop yield. The payload data
showed that combining mission is typical for microsatellite applications. The multi-
band infrared (MIRIS) payload in STSAT-3 is used for Earth and space observation.
LAPAN-ORARI has three kinds of missions, including Earth observation, commu-
nication, and ship data collecting platform.

4.7 Mission data downlink

The quantity and quality of the payload in Tables 1–4 showed that mission data
are increasing with time, which increase the required downlink data rate. For the
University of Surrey heritage, the data rate started with 3 Mbps in KITSAT-3 and
increased to 100 Mbps in RASAT. For the TU Berlin heritage (the digital transmis-
sion cases), the data rate started with 256 kbps in Maroc-TUBSAT and increased
to 100 Mbps in LAPAN-IPB. In the early missions, the mission data downlink is
transmitted in S-band, and as the data rate requirement increases, the downlink has
been shifted to X-band.

4.8 Payload computer

Payload computer is typically separated from satellite main computer, which

mainly manage the satellite bus. As the payload data rate increased, the payload pro-
cessing electronics is also evolved, from microcontroller/microprocessor to FPGA
based, which is known to be able provide high computing power with less risk from
space radiation as compared to high capacity microprocessor.

4.9 Orbit determination

None of the microsatellite has ranging system. Therefore, in early missions,

their orbit determination is mainly depending on NORAD’s data. The use of GPS
for Position-Navigation-and-Timing by the University of Surrey heritage satellites
started with BILSAT-1, while for the TU Berlin heritage satellites, it started with
LAPAN-ORARI. The accuracy of orbit determination becomes crucial in Earth
observation mission, as part of the parameters used in satellite image geometric
correction.

11
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 6.
Microsatellites’ weight versus launch year.

Figure 7.
Microsatellites’ density versus launch year.

4.10 System level parameters

Figure 6 shows the weight of each microsatellite sample. It shows that the weight
of TU Berlin heritage satellites grows in time. This is due to the increase in mission
quantity and complexity, which therefore requires more components in the satel-
lites (bigger batteries, more attitude sensors, larger lens for imager payload, etc.).
For the University of Surrey satellites heritage, such pattern is not found. The den-
sity of (weight/volume) the satellites is shown in Figure 7, indicating that the TU
Berlin heritage satellites are more compact than the University of Surrey heritage
satellites. For the University of Surrey satellites heritage, the design uses maximum
volumetric envelope for maximizing the solar panel area.

5. Conclusions

The chapter has discussed the differences between the University of Surrey
design heritage microsatellites and the TU Berlin heritage microsatellites. Five

12
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

sample satellites from each satellite design heritage are compared, including 15 bus
parameters, payload profiles, and satellite weight and volume at launch. From the
comparison, it is found that major differences in the satellite bus are in the choice of
main computers and their associated link configuration and in the attitude control
modes that also affect the design. Another major difference is in the satellites’
structure design, which resulted in much higher density in the TU Berlin heritage
satellites than the University Surrey heritage satellites. In the early design, there are
differences in the choice of satellite’s batteries. However, as soon as Li-ion batteries
became available, both design heritages used such technology. In answering the
increasing needs in payload data handling, both design heritage use FPGA-based
payload data handling and high downlink data rate in X-band. GPS is also the
technology adopted by both design heritages for orbit determination and imager’s
ancillary data.
For further studies on the topic, it is suggested that comparison to be done on
the power budget of the satellites and on the operation performance parameters of
the satellites with similar missions.

Acknowledgements

The author wishes to acknowledge the Center for Aerospace Policy Studies of
LAPAN for funding this publication and also wishes to thank the book editor for
giving substantial advice in writing this chapter.

Author details

Triharjanto Robertus
National Institute of Aeronautics and Space, Jakarta, Indonesia

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

13
Satellite Systems - Design, Modeling, Simulation and Analysis

References

[1] Swartout M. Cheaper by the dozen: [10] Sun W, Sweeting MN. In-orbit
The avalanche of rideshares in the results from UoSAT-12 Earth
21st century. In: 2013 IEEE Aerospace observation minisatellite mission. In:
Conference; Big Sky, MT, USA; 2013 Proceedings of the 3rd International
Symposium of IAA; Berlin; 2-6 April
[2] Bunchen E. 2014 Nano/Micro- 2001. pp. 79-82. IAA-B3-0305P
satellite market assessment. In:
Proceedings of the 28th AIAA/USU [11] Fouquet M, Sweeting M. UoSAT-12
Conference on Small Satellites. Utah; Minisatellite for high performance
2014 earth observation at low cost. Acta
Astronautica. 1997;41(3):173-182
[3] Kramer HJ. DMC-3. 2016. Available
from: https://directory.eoportal.org/ [12] Lim JT, Nam MR, Ryu KS, Tahk KM,
web/eoportal/satellite-missions/d/ Rhee SH, Kim KH. Exploring space on
dmc-3 a small satellite, STSAT-2: A test bed
for new technologies. In: 14th Annual
[4] Kramer HJ. SkySat. 2016. Available AIAA/USU Conference on Small
from: https://directory.eoportal.org/ Satellites; 2000
web/eoportal/satellite-missions/s/skysat
[13] Kim BJ, Lee HW, Choi SD. Three-
[5] Axelspace. GRUS-1: The Next axis reaction wheel attitude control
Generation of Earth-Observation system for KITSAT-3 microsatellite.
Satellites. 2016. Available from: https:// Space Technology. 1997;16(5/6):191-296
www.axelspace.com/en/solution_/grus/
[14] Park JO, Park HY, Lee DH,
[6] Quigley JT. Japanese Startup Raises Lee JH, Rhee SW. STSat-3 (Science
$15M to Build Satellites that Are the Size & Technology Satellite-3) satellite
of your Air Conditioner. 2015. Available missions. In: Proceedings of the 60th
from: https://www.techinasia.com/ IAC (International Astronautical
japan-micro-satellite-startup-axelspace- Congress); Daejeon, Korea; 12-16
series-a October 2009. IAC-09.B4.2.6

[7] Henry C. How OneWeb Plans to [15] Shin GH, Park HY, Seo JK,
Make Sure Its First Satellites Aren’t Chung TJ, Myung NH. Introduction to
Its Last. 2019. Available from: https:// micro-satellite STSAT-3 development.
spacenews.com/how-oneweb-plans-to- In: Proceedings of the 60th IAC
make-sure-its-first-satellites-arent-its- (International Astronautical Congress);
last/ Daejeon, Korea; 12-16 October 2009.
IAC-09.B4.6A.2
[8] Clark S. Telesat Taps Blue Origin to
Launch Broadband Satellite Fleet. 2019. [16] Gomes LM, Yuksel G, Lappas V, da
Available from: https://spaceflightnow. Silva Curiel A, Bradford A, Ozkaptan C,
com/2019/02/04/telesat-taps-blue-origin- et al.. BILSAT: Advancing smallsat
to-launch-broadband-satellite-fleet/ capabilities. In: AIAA/USU Conference
on Small Satellites; Logan, UT; 11-14
[9] Triharjanto R. Strategic environment August 2003
and implementation of satellite
technology acquisition programs. In: [17] Bradford A, Yuksel G, Gomes LM,
ASEAN, Proceeding Seminar Nasional Ozkaptan C, Sweeting M, Orlu U.
Kajian Kebijakan Penerbangan dan BILSAT-1: A low-cost, agile, earth
Antariksa; Surabaya, Indonesia; 2018 observation microsatellite for

14
System Designs of Microsatellites: A Review of Two Schools of Thoughts
DOI: http://dx.doi.org/10.5772/intechopen.92659

Turkey. In: Proceedings of 53rd IAC; observation. In: Proceeding 5th

Houston, TX; 10-19 October 2002. International Symposium Small Satellite
IAC-02-IAA.11.3.02 Systems and Services; La Baule, France;
2000
[18] Leloglu UM, Yuksel G,
Sweeting MN. BILSAT-1: A case study [25] Renner U, Bleif J, Roemer S. The
for the Surrey Satellite Technology TUBSAT attitude control system:
Ltd Know-How Transfer and Training flight experience with DLR-TUBSAT
Program. In: Proceedings of 53rd IAC; and MAROC-TUBSAT. In: 4th IAA
Houston, TX; 10-19 October 2002. Symposium; Berlin, Germany; 7-11
IAC-02-P.1.01 April 2003. IAA-B4-1201

[19] Yuksel G, Belce O, Urhan H, [26] Roemer S, Renner U. Flight

Gomez L, Bradford A, Bean N, et al. experience with the microsatellite
BILSAT-1: First year in orbit-operations MAROC-TUBSAT. In: 54th IAF
and lessons learned. In: Proceedings Congress; Bremen, Germany;
of AIAA/USU Conference on Small September 29–October 3, 2003;
Satellites; Logan, UT; 9-12 August 2004. IAC-03-IAA.11.2.07
SSC04-IX-3
[27] Triharjanto RH, Hasbi W,
[20] Kim EE, Choi YW, Choi W, Widipaminto A, Mukhayadi M,
Kim HG, Park SJ, Yun JH, et al. Renner U. LAPAN-TUBSAT: Micro-
Integration and testing of optical satellite platform for surveillance &
imaging system for RASAT. In: remote sensing. In: Proceedings of the
Proceedings of Recent Advances in 4S Symposium: Small Satellites, Systems
Space Technologies (RAST); Istanbul, and Services; La Rochelle, France; 2004
Turkey; 2007
[28] Saifudin MA, Mukhayadi M.
[21] Ontaç S, Dağ S, Gökler MI. Structural LAPAN-A2 attitude control strategy
finite element analysis of stiffened and for equatorial surveillance mission.
honeycomb panels of the RASAT satellite. In: Proceedings of the 9th IAA
In: Proceedings of the 3rd International Symposium on Small Satellites for Earth
Conference on Recent Advances in Space Observation; Berlin, Germany; 8-12
Technologies (RAST 2007); Istanbul, April 2013. Paper: IAA-B9-1202
Turkey; 14-16 June 2007
[29] Triharjanto RH, Hakim PR. Review
[22] Yüksel G, Okan A, Leloğlu UM. First of satellite technology development in
LEO satellite built in Turkey: RASAT. In: Indonesian Space Agency based on its
Proceedings of the 3rd International technical publications in 2012-2016. In:
Conference on Recent Advances in Proceedings of the 68th International
Space Technologies (RAST); Istanbul, Astronautical Congress; Australia; 2017
Turkey; 14-16 June 2007
[30] Triharjanto RH. The comparison
[23] İsmailoğlu N, Vargün B, Sever R, of LAPAN-A1 and LAPAN-A2 system
Okcan B, Karakuş K, Kapucu K, et al. design: Weight ratios and moment
GEZGN-2: Image processing subsystem inertias analysis. In: Proceedings 2nd
of RASAT. In: Proceeding 5th International Seminar of Aerospace
International Conference on Recent Science and Technology; Jakarta,
Advances in Space Technologies Indonesia; 2014
(RAST); 2011
[31] Utama S, Saifudin MA,
[24] Schulz S, Renner U. DLR-TUBSAT, Mukhayadi M. Momentum biased
A microsatellite for interactive earth performance of LAPAN-A3 satellite

15
Satellite Systems - Design, Modeling, Simulation and Analysis

for multispectral pushbroom imager

operation. IOP Conference Series
Earth and Environmental Science.
2018;149(1):012062

[32] Mukhayadi M, Madina R,

Renner U. Attitude control of bias
momentum micro satellite using
magnetic and gravity gradient torque.
In: Proceedings of the 2014 IEEE
International Conference on Aerospace
Electronics and Remote Sensing
Technology; Yogyakarta, Indonesia;
2014. pp. 132-136

16
Chapter

Design of Intelligent and Open

Avionics System Onboard
Changqing Wu, Xiaodong Han and Yakun Wang

Abstract

The continuous development of space missions has put forward requirements

for high performance, high reliability, intelligence, effective integration, min-
iaturization, and quick turn around productization of the electronic system of
satellites. The complexity of satellites has continued to increase, and the focus of
satellite competition has shifted from the launch of success shifts to communica-
tion capacity, performance indicators, degree of flexibility, and continuous service
capabilities. So, the importance of onboard avionics system is becoming increas-
ingly prominent. In the future, the advanced avionics system integrates most of the
platform’s electronic equipment. The design level of the system largely determines
the performance of the satellite platform. This chapter focuses on the applica-
tion requirements of the new generation of intelligent avionics system for future
communication satellites and adopts an “open” architecture of “centralized manage-
ment, distributed measurement and drive, and software and hardware ‘modular’
design” to build a universal, standardized, and scalable intelligent avionics system.

Keywords: satellite, avionics system, intelligent, open architecture, modular design,

centralized management, reliability

1. Introduction

With the continuous advancement of electronics and computer technology, the

functions and performance of spacecraft avionics system have also continuously
improved, covering functions such as spacecraft remote measurement and remote
management, energy management, thermal management, health management,
payload information processing, and mission task management. Avionics system
plays a core role in the realization of information sharing and comprehensive
utilization, function integration, resource reorganization and optimization, and
information processing and transmission [1]. It is the foundation for spacecraft to
implement autonomous management and control and is also a bridge for communi-
cation management from a spacecraft to other spacecrafts and ground station [2].
The traditional spacecraft electronic system uses a layered centralized manage-
ment control mode similar to a pyramid. It not only needs a large amount of data
interaction between the management unit and the interface unit but also requires
the management unit to process a large amount of underlying data, which makes
the management unit overwhelmed. It severely limits the processing and support of
high-level tasks by electronic systems. Moreover, the management unit is at the top
of the “pyramid” of centralized management, which requires high reliability. Once

1
Satellite Systems - Design, Modeling, Simulation and Analysis

a failure occurs, the entire electronic system will fail. Thus, the centralized manage-
ment method is no longer suitable for the needs of spacecraft development.
The satellite intelligent avionics system is an information processing and trans-
mission system that uses computer network technology to interconnect satellite-
borne electronic equipment on the satellite to achieve internal information sharing
and comprehensive utilization, function integration, and resource reorganization
and optimization. Utilizing onboard computers to complete satellite data manage-
ment, control management, communication management, time management,
energy management, and job management functions through unified scheduling of
satellite missions. Its essence is the generation, identification, processing, analysis,
transmission, and distribution of information process. The integrated satellite
electronic system integrates the functions of the satellite platform electronic
equipment, and its design level directly determines the performance of the satellite
platform [3–5].
At present, satellite sub-systems mostly adopt independent design schemes,
which decentralize satellite attitude control, propulsion control, thermal control,
satellite-ground link communication, and power control functions. The onboard
computer is responsible for tasks such as remote control, telemetry, program-
controlled operation, thermal control, and time management. The attitude and
orbit control computer are responsible for attitude and orbit (including propulsion
control) control. Each sub-system such as power supply, thermal control, and
digital transmission is equipped with corresponding lower-level computers respon-
sible for telemetry acquisition and remote control of the respective sub-system.
However, the satellite system designed using this approach is usually resulting in
heavy weight, high power consumption, large volume (aka high size, weight, and
power (SWAP)), complex interface relationships, weak system reconfiguration
capabilities, and low functional density. In order to overcome the abovementioned
shortcomings and make the satellite avionics system better meet the SWAP and
flexible system configuration requirements of future missions, it is necessary to
improve its design technology, that is, from the current independent design of each
sub-system to the open and modular design of the entire satellite. Based on the prin-
ciple of unified application, deployment and operation of hardware resources, and
the full use of the various functions of the software, the information sharing of the
entire satellite, simple system configuration, and overall performance optimization
are realized.
This chapter focuses on the application requirements of the new generation of
intelligent avionics system for future communication satellites, and adopts an open
architecture of “centralized management, distributed measurement and drive, and
software and hardware modular design.” The universal, standardized, and scalable
intelligent avionics system is built based on the basic modular elements of open
hardware modules, open software components, and industry standardized internal
and external busses.

2. System structure

This section introduces the intelligent open system architecture, including

Sections 2.1, 2.2, 2.3, and 2.4. Section 2.1 introduces the overall architecture
design; the system adopts the distributed design mode and completes the intel-
ligent management of onboard tasks through the menu hardware architecture
and open interface protocol. Section 2.2 discusses the hardware architecture of
high-performance computing and introduces the onboard high-performance
computing and the corresponding storage capacity from the main functions,

2
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

processing, storage, and radiation resistance. Section 2.3 describes the dynamic
state reconfigurable task scheduling that improves the fault tolerance ability of the
satellite network in view of the typical scenarios of the satellite integrated elec-
tronic system in the operation process. Section 2.4 discusses the design of software
partition protection mechanism related to the next-generation avionics system
and analyzes the requirements, design, and functions of partition protection,
aiming to improve the robustness of the software system.

2.1 Overall architecture design

The architecture of the newly proposed next generation of intelligent com-

munication satellite avionics system is shown in Figure 1. The avionics system
architecture (ASA) is designed as a data bus (DB)-based real-time distributed
computer system. ASA consists of one Satellite Management Unit (SMU), one
Platform Integrated Services Unit (PFISU), one Payload Integrated Services Unit
(PLISU), and a set of DB and auxiliary software. The SMU is the core of the avion-
ics system. ASA controls the PFISU and PLISU by DB and connects with Telemetry
and Telecommand Unit (TTU) to receive commands and send the telemetry data.
PFISU and PLISU are the execution parts of the avionics system. PFISU and PLISU
are used to command driver, signal sample, power distribution, heater control,
pyrotechnic management, and interface management. To improve the reliability of
avionics system, the SMU, PFISU, and PLISU will have built in redundancies. This
avionics system supports the functions of satellite on-orbit dynamic registration,
spatial data interaction, and routing and can solve the problem of user-oriented and
task-oriented opening of satellite system.

2.2 High-performance computing hardware architecture

As the amount of data generated by satellite electronic equipment continues

to increase, a large amount of data processing requirements place higher require-
ments on satellite information processing capabilities. The avionics system is the

Figure 1.
Avionics system architecture diagram for satellite communication system.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

information core of the whole satellite, especially for the requirements of intelligent
satellite systems. Research on the realization of high-performance computing of
avionics systems is an inevitable requirement [6, 7]. In order to improve the com-
puting capacity of the avionics system, a high-performance onboard processor is
utilized. The following introduces onboard high-performance computer from four
aspects: main functions, processing, storage, and anti-irradiation.

i. Main functions:

• Uses redundant onboard computer supporting on-orbit reconstruction

and reconfiguration for highly reliable avionics system.

• Supports Consultative Committee for Space Data Systems (CCSDS)

telemetry and telecommand with optional radio-frequency channels. This
feature allows the proposed intelligent avionics system design to be open
interface using widely acceptable industry standards.

• Uses interface with external unit. Provide a brief description why is this
important function.

• Provides secondary power distribution and discrete instructions to exter-

nal units. Provide a brief description why is this important function.

ii. High-performance processors:

• 215 Dhrystone Million Instructions executed Per Second (DMIPS) and

floating-point arithmetic unit

• L1 instruction cache and L1 data cache with Error Correcting Code (ECC)
function

• Internal Random Access Memory (RAM), FLASH, and Electrically

Erasable Programmable Read-Only Memory (EEPROM), with ECC
function

• Contains basic software: BIOS and startup software

iii. High-performance memory:

• Volatile: 192 MB SDRM CPU, with error detection function

• Volatile: 64 MB SDRM IO, with ECC function

• Nonvolatile: 4GB FLASH, with ECC function

iv. Radiation resistance:

• Spaceborne components will not be locked due to space radiation.

• Dual-core processor lockstep technology is used for error detection.

• All memories have ECC function (RS code or EDAC).

4
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

Among them, the “lockstep” technology is a fault-tolerant computing technol-

ogy. This technology uses the same, redundant hardware components and processes
the same instructions at the same time. The core idea is to keep multiple central
processing units (CPUs) and memories executing the same instructions accurately
and synchronously by running synchronous comparisons in operation to improve
the fault-tolerant computing capability of the avionics system.

2.3 Dynamically reconfigurable task scheduling

The two typical scenarios usually encountered by satellite avionics systems during
operation are (a) a node fails or requires functional reorganization so that some tasks
on this node need to be migrated to other nodes through the network and (b) the
resource occupancy rate of a node is too high so that some tasks on this node will be
migrated to other relatively idle nodes for execution. The avionics system is designed
with networked real-time multitasking distributed system software, which can also
implement dynamic reconfiguration of functions and task scheduling. The embed-
ded system software running on each node in the network supports not only the local
real-time multitasking scheduling but also the network operation capability. The avi-
onics system supports function modification and function migration between nodes,
which realizes the transformation to software-defined satellite functions, reduces the
differences in hardware products, improves the fault tolerance of the intra-satellite
network, and also meets the growing needs of intra-satellite networking [8].
The avionics system networked real-time multitasking distributed system
software is shown in Figure 2 and has the following characteristics:

• Application tasks are directly oriented to users. In order to complete a top-level

function in a specific domain, the tasks are decomposed into functions of
appropriate granularity. The software functions that multiple tasks will use are
called domain public services. It is called public services in multiple fields, has
a clear interface definition, can complete certain functions relatively indepen-
dently, and adds service registration, management, control, and governance
to provide strong support for space application tasks. The user’s service

Figure 2.
Networked real-time multitasking distributed system software.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

composition capability, that is, when the business needs of the day change and
the service call is adjusted, can support the user to quickly combine services
and form a new business process [9].

• The middleware layer serves as a transition layer between the above and the
next [10]. By combining the characteristics of the tasks of each layer, the
corresponding theme is designed, and the publish-subscribe technology is
used to provide the entire application layer with access to various resources in
the basic resource layer. The software layer implements operating system and
communication protocol level shielding through the packaging of thread tasks,
synchronization resources, memory access, IO operations, Ethernet, shared
memory, and fiber-optic communications, providing access to the hardware
device layer. By virtualizing the calculation, storage, and network resources
of the basic resources, the basic resources as a service are realized, and the
availability and scalability of the hardware resources are guaranteed. At the
same time, the software is dynamically deployed for the hardware of the basic
resources, including the automation of basic software and application software
installation settings, maintenance and upgrades, etc., and provides the system
with general basic services such as system reconstruction, software fault toler-
ance, data management, subscription release, etc.

2.4 Software partition protection mechanism design

i. Partition protection requirements

Avionics systems can implement multiple functions to share resources. The

functional entities (which can be software modules, hardware modules) that
share resources are called partitions. The partitions of the original avionics system
shared resources, but sharing would bring potential problems described below:

• Multi-partition shared memory and IO: if one partition accidentally or

maliciously rewrites the memory and IO of other partitions, it will cause
the rewritten partition to fail.

• Multi-partition shared processor time: if a partition maliciously takes up

processor time due to a failure, the related partition will crash.

• Multiple partitions share the same communication link: if a partition

occupies channels too much, it will affect the bandwidth and real-time
performance of other partitions.

Therefore, the design of system software architecture of the avionics system

should meet the reliable partition protection to avoid the above problems.

ii. Partition protection design

The partition protection design of onboard system software in avionics

systems includes the following three aspects:

• Space protection. For processors with Memory Management Unit (MMU)

support, such as X86 processors, it is stipulated that the partition itself
cannot directly access physical memory, and only virtual memory can be
accessed through the MMU memory mapping table configured for each

6
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

partition by system software. For processors without MMU support, tak-

ing the Scalable Processor Architecture (SPARC) processor as an example,
during partition initialization, the system software protects the SRAM by
setting the privileged register to achieve partition space isolation.

• Inter-partition communication. Inter-division communication is another

important content of partition space protection. Inappropriate inter-
partition communication mechanisms can cause mutual interaction
between partitions. ARINC-653 specification provides a communication
mechanism that does not affect each other. Inter-partition communica-
tion within a single processor includes a memory buffer mechanism and
a blackboard mechanism. The buffer mechanism mainly provides data
communication between partitions, and the blackboard mechanism
mainly provides sampling services between partitions. Taking the buffer
mechanism as an example, the protection mechanism for inter-partition
communication is the following: (1) The core system software allocates an
inter-partition communication buffer (size, permission, and connection
relationship between buffers) for each partition according to the blueprint
information. Only the own buffer can be accessed. (2) The core system
software manages the resources for communication between partitions in
the privileged state and copies the information in the source buffer to the
destination buffer according to the blueprint information.

• Partition time protection. Its strategies include the following: (1) The basic
unit of scheduling is partitioning, and partition scheduling has no priority.
The Main Time Frame (MTF) is used to statically define the scheduling
order of each partition and the proportion window size. MTF is one of the
blueprint contents. The core system software is configured according to
the MTF, and multiple partitions are scheduled in a cyclic manner. (2) The
resources (such as timers, stacks, and memory) and blueprints required
for partition scheduling are in a privileged state, and the user partition
cannot overwrite the partition scheduling resources.

iii. Advantages of partition protection

• Blueprint only registers the interfaces on the modules and then mounts
them on the app as a whole. The purpose of blueprint itself is to organize
the parallel coexistence of multiple modules and avoid registering modules
directly on the app. In fact, it is more convenient for development and code
maintenance, because ultimately all interfaces on views are still directly
mounted on the app, which corresponds to the entire application; there is
no obvious difference [11].

• Blueprint is not a pluggable application, because it is not a real application,

but a set of operations that can be registered in the application and can be
registered multiple times.

• At the same time, we cannot use multiple objects to manage and register,
because this will cause each object to have its own configuration, which is
not easy to manage.

• With blueprint, the application will be managed in the flask layer, share
the configuration, and change the application object on demand through

7
Satellite Systems - Design, Modeling, Simulation and Analysis

registration. The disadvantage of blueprint is that once an application is cre-

ated, it can only be unregistered by destroying the entire application object.

3. Menu system composition

Based on the menu design idea, the avionics system can realize the sharing
of hardware modules and resources, task migration, and system reconstruction,
enhance the tolerance and processing ability of the avionics system for faults,
significantly improve the development efficiency and productization degree of
the integrated electronic system, and make the integrated electronic system highly
reliable. The intelligent spacecraft provides the necessary technical support, mainly
including the contents that will be presented in the subsequent sections [12].

3.1 Satellite management unit

The Satellite Management Unit is an improved satellite research equipment. The

design fully draws on the advantages of the previous satellite platform and has been
optimized and expanded. The main completed functions are as follows:

i. Telecommand function: The SMU receives the telecommand instructions

from the TTU and completes the distribution of the instructions.

ii. Telemetry function: The SMU collects its own telemetry and receive the indi-
rect telemetry parameters collected by the bus terminal equipment through
the 1553B bus. The SMU complete the framing processing according to the
CCSDS standard and transmit it to the TTU through the serial port.

iii. Satellite autonomous management function: Using application software

running on the SMU, functions such as energy management, thermal
management, bus management, payload management, and pyrotechnics
management are realized.

iv. On-orbit maintenance function: For the temporary adjustment of control

parameters during the execution of on-orbit tasks, the SMU is able to modify
the control parameters of the software. It also supports onboard software
maintenance function of onboard software, which can realize the update and
recovery of software modules.

v. Important data saving function: The SMU can periodically save and maintain
important data of satellite sub-systems. When the internal configuration of
the relevant sub-system changes or the corresponding module restarts, the
SMU can send the important data stored internally to the corresponding
module. Besides, when the SMU is reset or switched off, the SMU can restore
the current working state through the important data.

vi. Fault Detection Isolation and Recovery (FDIR) function: The SMU provides
the operating platform for satellite FDIR [13]. When the SMU is healthy, the
SMU monitor key information such as the entire satellite’s energy and thermal
control. The SMU detects various failure conditions in real time and performs
troubleshooting through the direct remote command interface or through the
1553B bus. The FDIR of the control sub-system is completed by the attitude
control computer. Note that the most advanced 1553B bus can handle data rate

8
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

up to 10 Mbps. For data rate higher than 10 Mbps, industry trend is moving
toward SpaceWire data bus that can handle data rate up to 400 Mbps.

The SMU adopts the dual-machine cold backup working mode. In the case of the
autonomous switching enabled, when the on-duty machine fails, the failure-tolerant
module automatically completes the switch of SMU and power off the faulty
machine. In the case of the autonomous switching not enabled, when the on-duty
machine fails, the power failure and machine switch is conducted by ground station.

3.2 Integrated service unit

The Integrated Service Unit (ISU) adopts modular hardware design [14, 15].
Each functional module is connected to the “bus interface management module”
through an internal bus, which uses widely acceptable data bus such as 1553B bus.
The ISU is mainly composed of a bus interface management module, a matrix
instruction and matrix telemetry module, and an analog quantity acquisition and a
discrete instruction output module.
The menu module is mainly composed of a bus interface management module,
a matrix acquisition and command module, and an information acquisition and
command module [16]. The module menu is shown in Table 1 below.
The functions of each module are also modularized. The capability of each
module is shown in Table 2.
After the module design is completed, the number of modules is determined
according to the task requirements. Then, complete the assembly according to the
standard interface. The number of modules menu is shown in Table 3.

3.3 Data bus network

The data bus network is the information transmission hub of the avionics
system. Through the data bus network, distributed data acquisition, and instruction
output, centralized operation and control are implemented, thereby improving the
efficiency of system processing. The avionics system first-level bus is 1553B bus [17].
Note that for data rate higher than 10 Mbps, SpaceWire data bus is recommended.
Data exchange between SMU and ISU and other equipment realized through
1553B bus. The master–slave response mode of the 1553B bus is used to transmit
platform command data and telemetry acquisition data. In the data exchange
process of the first-level bus, the SMU always acts as the controller of the 1553B bus

Functions Module menu

Telemetry/telecommand Acquisition and Command-A module (AA)

Acquisition and Command -B module (AB)

Matrix acquisition and command module (AC)
Temperature acquisition High-voltage heater control module (HH)

Low-voltage heater control module (LH)

Heater and distribution module (HD)
Pyrotechnic management Pyrotechnic management module (CA)
Bus data transmission 1553B bus

Table 1.
Function module list.

9
Satellite Systems - Design, Modeling, Simulation and Analysis

Module Module capabilities

AA Analog/temperature measurement acquisition channel

Small current discrete command output

High current command drive circuit
AB Analog/temperature measurement acquisition channel

Small current discrete instructions

Bi-level quantity acquisition
AC Matrix instructions
Switch status acquisition
HH High-pressure heater power distribution
Temperature measurement collection
LH Low-voltage heater power distribution
Temperature measurement
HD High-voltage instrumentation and power distribution

High-pressure heater power distribution

Temperature measurement
CA Pyrotechnic management

Table 2.
Modularized functions.

Demand Quantification AA AB AC CA
Pyrotechnics management 80 2
High current instruction 30 1
Low current instruction 100 1
Analog acquisition 220 2 2
Matrix acquisition 550 2
Total (take the maximum) 2 2 2 2

Table 3.
Menu-style design.

and initiates the communication. ISU and other equipment, as the RT end, receive
instructions and send collected telemetry data to the SMU.

4. Failure detection isolation and recovery (FDIR) design

In order to achieve autonomous and healthy operation of the satellite, the intel-
ligent satellite system uses the FDIR software to monitor the status of the satellite in
real time and diagnose and predict its working status and performance trends [18].
When a failure occurs, the FDIR software can locate the failure in time and deter-
mine which components are not working normally or the performance is degraded.

4.1 FDIR design goals and principles

Design goals:

i. Satellites can survive if any failure occurs.

10
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

ii. When a failure occurs, try to extend the mission time of the satellite and
reduce the loss of mission interruption.

iii. The life of the satellite should be guaranteed: optimize fuel consumption and
minimize system configuration and component losses.

The above three principles apply to the launch phase, the transfer orbit phase,
and the on-orbit phase.
FDIR is an important component of the onboard software, which can perform
on-orbit processing of failures, thereby reducing the impact of failures. However,
not all on-orbit failures can be detected and processed. FDIR design should follow
the following principles:

i. FDIR processing follows the single failure principle, that is, only one failure
is processed at a time.

ii. Failures are divided into 0 to 4 levels according to their impact on satellites.

iii. The higher the failure level, the higher the processing priority. During a
failure processing, if a higher-level failure occurs, the higher-level failure is
processed first.

iv. The failures of the same level are processed in the order of occurrence.

v. All FDIR processing requires failure recovery instructions and failure

processing records.

4.2 Failure levels

According to the impact of the failure on the satellite operation, the failure is
categorized as follows:

i. System-level failure: failures that damage the functions and performance of

the satellite system.

ii. Sub-system-level failure: the functions of the sub-system cannot be or are

partially completed, or the main performance indicators and parameter val-
ues of the sub-system exceed the range required by the sub-system design.
But it does not affect the main functions and performance of the system.

iii. Equipment-level failure: equipment functions cannot complete the main

performance indicators, or parameter values exceed the range of equipment
design requirements. But it does not affect the main functions and perfor-
mance of the system.

iv. Module-level failure: a failure in which the module function cannot be

completed, or the main performance indicators and parameter values exceed
the range required by the component design. But it does not affect the main
function and performance of the equipment.

According to the possible impact on components, functions, and systems,

FDIR is designed for five failure levels from levels 0 to 4, according to the different
sub-systems that each failure belongs to, including measurement and control FDIR,

11
Satellite Systems - Design, Modeling, Simulation and Analysis

avionics FDIR, and power supply and distribution FDIR as illustrated in Figure 3.
The larger the number, the higher the fault level, and vice versa.

• Level 0 failure: a level 0 failure refers to a failure that occurs inside an equip-
ment and can be recovered autonomously by the hot backup method inside the
equipment without affecting other components of the system.

• Level 1 failure: a level 1 failure refers to the failure of a single equipment or

module of each sub-system. After a level 1 failure occurs, the system will per-
form autonomous failure isolation and recovery according to the FDIR policy.
If the failure isolation and recovery is successful, it has no impact on system
tasks. The detection, isolation, and recovery of failures are implemented by
application software.

• Level 2 failure: a level 2 failure refers to the functional level abnormality of the
satellite sub-system. Under such failures, the system performance cannot meet
the design requirements. For level 2 failures, the recovery strategies need to be
implemented and related components need to be enabled or restarted. Level 2
failures can cause system performance degradation or temporary interruption
of system tasks. Its failure detection, isolation, and recovery are performed by
application software.

• Level 3 failure: a level 3 failure refers to the failure of the CPU hardware, which
is detected by the hardware. After the failure occurs, it is switched to the
backup CPU according to the failure handling strategy.

• Level 4 failure: a level 4 failure refers to the failure of the satellite to maintain
the pointing to the ground in the on-orbit phase and requires sun capture
processing.

Figure 3.
Schematic diagram of satellites in orbit during their lifetime.

12
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

4.3 FDIR scheme

The software autonomously isolates the failure and rebuilds the system at the
appropriate time according to the following FDIR scheme:

• High-level failure detection such as level 4 failure has priority over low-level
failures such as level 3. When two or more failures are detected at the same
time, the recovery sequence of high-level failures is performed preferentially.
Once a high-level failure occurs, all detection of the same-level and low-level
failures are suspended before the recovery sequence is completed.

• Only one failure recovery sequence is performed on the satellite at the same
time, that is, all FDIR failure recovery is shielded during the execution of any
failure recovery strategy. The sufficiency and necessity of the failure recovery
sequence should be effectively verified by ground testing to minimize the
interpretation during the sequence execution.

• After the execution of the failure recovery sequence is completed, the failure
detection of the flight can be continued, but the failure recovery enable status
should be set to disable. At the same time, the detection of other similar and
low-level failures should be enabled. After confirming the working status
of the products on the ground, reset the backup status and enable the FDIR
recovery function.

• FDIR only detects the status of the on-duty module. When the FDIR enable flag
is “disabled,” the status of the module is not detected. When the FDIR enable flag
is “enabled,” the status of the on-duty module is detected. If the failure detec-
tion condition is met on the on-duty module and the FDIR recovery enable flag
is “disabled,” the health status of the module is set to unhealthy. If the failure
detection condition is met on the on-duty module and the FDIR recovery enable
flag is “enabled,” it is determined whether the status of the on-duty module is
the same as the backup module. If they are the same, do not perform the recov-
ery operation and set the on-duty module as unhealthy. If they are different,
perform the recovery operation and set the backup module to the on-duty status.

• For autonomous maintenance on the satellite, the “health status” of each

module can only be changed from “healthy” to “unhealthy.”

• For dual-machine hot standby equipment or modules, only the health status of
non-duty module is detected, and no recovery is performed.

• Use its own fault-tolerant RAM and lower computer to save important data in
time for state recovery after failure.

4.4 FDIR processing requirements for satellites in orbit

The FDIR requirements for each phase of the satellite are as follows:

i. Launch phase: allows failure detection and recovery of level 0 and level 1
failures.

ii. Transfer orbit phase: allows failure detection and recovery of level 0 to level 3
failures.

13
Satellite Systems - Design, Modeling, Simulation and Analysis

iii. On-orbit phase: allows failure detection and recovery of level 0 to level 4
failures.

4.5 FDIR processing

The processing flow of FDIR mainly includes four parts:

i. Judgment of processing conditions: First, determine the scope of failure

detection according to the requirements of the satellite in orbit and ground
control. Then, according to the validity of the telemetry data and the situa-
tion of the modules on duty, determine the FDIR project that can be used for
failure detection.

ii. Fault detection: Determine whether a failure occurs based on the recognition
characteristics.

iii. Comprehensive information processing: After a failure occurs, it is deter-

mined that whether the current situation of the satellite meets the recovery
conditions. At the same time, in the case of multiple failures, priority judg-
ment is required. Finally, determine the failures that can be recovered and
the order of recovery.

iv. Failure recovery: According to the engineering and testing experience,

perform corresponding recovery operations.

5. Impacts on next-generation avionics system

The intelligent avionics system adopts a system engineering method using

modular and open design to uniformly design the information processing, control
and management processes, hardware, and software, which is to realize the opti-
mization of information and resource sharing. Based on the onboard computer and
high-speed bus such as SpaceWire data bus, a set of information fusion systems
and mechanisms is established. The system is a menu-style, modular, and exten-
sible open service platform, which achieves a high degree of integration of various
onboard software and hardware resources and can meet the requirements for differ-
ent tasks.
The intelligent avionics system adopts the design concept of a modular menu
system architecture to meet the needs of real-time, reconfigurable, autonomous
planning, and intelligence of the system. With the SMU as the 1553B and SpaceWire
bus controller for data rate less than 10 Mbps/for data rate more than 10 Mbps,
respectively, and the ISU as the remote terminal, a distributed, master-slave, and
menu-based satellite networks are constructed.
Satellites are designed with a network layout, which can design different menu
network nodes on the bus network. After the payload capacity is strengthened, the
network node can increase the corresponding payload processing unit. The SMU
is used as the main processing computer to perform the main control of satellite
services to form a master-slave network structure. The high-performance onboard
processor enables the intelligent avionics systems with high-performance comput-
ing capabilities, which not only meets the data processing requirements but also
lays the foundation for satellite intelligence. The intelligent avionics system adopts
partition protection measures. Through the design of space protection, time protec-
tion, and partition communication, it provides reliable functional entities (such as

14
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

software modules or hardware modules) that share resources. Partition protection

avoids the impact of other partitions under abnormal conditions such as single par-
tition failure or malicious access. At the same time, the intelligent avionics system
is equipped with real-time multitask distributed system software, which can realize
the dynamic reconstruction of functions and tasks. For example, when a node fails
or needs a functional reorganization, some tasks on that node will be migrated to
other nodes. Or, the resource occupation rate of a node is too high, and some tasks
on this node will be migrated to other relatively idle nodes for execution. This
design can improve the failure tolerance of the intra-satellite network and achieve
efficient resource allocation and scheduling. All information is collected into the
SMU for comprehensive analysis and processing through the 1553B/SpaceWire
bus network. For example, in the process of autonomous energy management, it
is found that the battery discharge depth reaches 80%. If the control sub-system is
still in the mode of pointing to the ground, it will seriously affect the safety of the
satellite. At this time, the instructions should be sent in time to orient the satellite
to the sun to ensure the safety of the satellite. The SMU can be fully applied to the
satellite’s autonomous information fusion processing, ensuring that the satellite can
still guarantee normal communication services in the event of a major failure, and
energy security in emergency situations.

6. Conclusion

The intelligent avionics system design is the key technology for future advanced
satellites. The system design has adopted modular and open system architecture
approach using an efficient computing hardware system to maintain multiple
central processing units and memory executing instructions accurately and syn-
chronously with high cost-performance and cost-efficiency ratio. This approach
improves the failure tolerance of the next-generation avionics systems. The func-
tion modification capabilities and function migration between modules realize the
transition to software-defined satellite. Furthermore, this approach also reduces the
differences in hardware products and improves the failure tolerance of the satellite’s
internal network, which meets the ever-increasing networking requirements of the
satellite.

Acknowledgements

This work was supported in part by the National Natural Science Foundation of
China (No. 61972398).

15
Satellite Systems - Design, Modeling, Simulation and Analysis

Author details

Changqing Wu, Xiaodong Han* and Yakun Wang

Institute of Telecommunication Satellite, China Academy of Space Technology,
Beijing, China

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

16
Design of Intelligent and Open Avionics System Onboard
DOI: http://dx.doi.org/10.5772/intechopen.93141

References

[1] Sherwood R, Chien S, Tran D, et al. [9] Heymans P, Boucher Q , Classen A.

The EO-1 autonomous sciencecraft. In: A code tagging approach to software
Small Satellite Conference. Logan, UT: product line development. An
AIAA/USU; 2007 application to satellite communication
libraries. International Journal on
[2] Rabideau G, Tran D, Chien S, et al. Software Tools for Technology Transfer.
Mission operations of earth observing-1 2012;14(5):553-566
with onboard autonomy. In: IEEE
International Conference on Space [10] Boreli R, Ge Y, Iyer T, et al.
Mission Challenges for Information Intelligent middleware for high speed
Technology. Pasadena, CA: IEEE; 2006 maritime mesh networks with satellite
communications. In: 9th International
[3] Sherwood R, Chien S, Tran D, et al. Conference on Intelligent Transport
Enhancing science and automating Systems Telecommunications (ITST).
operations using onboard autonomy. Harbin, China: IEEE; 2009
In: International Conference on Space
Operations (SpaceOps 2006). Rome, [11] Zhu H. Design of on-board
Italy: AIAA; 2006 software architecture based on design
pattern. Computer Application.
[4] Sherwood R, Chien S, Tran D. 2008;25(12):180-181
Autonomous science agents and sensor
webs: EO-1 and beyond. In: IEEE [12] Yu Z, Yang-Ming Z, Zheng-Liang H,
Aerospace Conference (IAC 2006). et al. Fault-tolerant design of memory
Big Sky, MT: IEEE; 2006 module for pica-satellite on-board
computer. Journal of Astronautics.
[5] Chien S, Sherwood R, Tran D, 2008;29(6):2057-2061
et al. Using autonomy flight software
to improve science return on earth [13] Guiotto A, Martelli A,
observing one. Journal of Aerospace Paccagnini C. SMART-FDIR: Use
Computing, Information, and of artificial intelligence in the
Communication. 2005;2(4):196-216 implementation of a satellite FDIR. In:
DASIA 2003. Prague, Czech Republic:
[6] Chien S, Sherwood R, Tran D, et al. DASIA; 2003
Lessons learned from autonomous
sciencecraft experiment. In: [14] Herpel HJ, Schuettauf A, Willich G.
Autonomous Agents and Multi-Agent Open modular computing platforms
Systems Conference (AAMAS 2005). in space — Learning from other
Utrecht, Netherlands: AAMAS; 2005 industrial domains. In: IEEE Aerospace
Conference. MT, USA: IEEE; 2016
[7] Rabideau G, Chien S, Sherwood R,
et al. Mission operations with autonomy: [15] Mark H, Helene DM, Alexandre C.
A preliminary report for earth Requirements baseline for integrated
observing-1. In: International Workshop modular avionics for space separation
on Planning and Scheduling for Space kernel qualification. In: Dasia-Data
(IWPSS 2004). Darmstadt, Germany: Systems in Aerospace. Barcelona, Spain:
IWPSS; 2004 DASIA; 2015

[8] Marie J. Spacebus 4000 avionics: Key [16] Panpan Z, Tingyuan G, Jianjun G,
features and first flight return. In: 24th et al. Plug-and-play on-board computer
AIAA International Communications system design based on BM3803
Satellite System Conference (ICSSC). processor. Journal of Aerospace
2006 Engineering. 2013;22(06):92-96

17
Satellite Systems - Design, Modeling, Simulation and Analysis

[17] Kan W, Jingfei J, Mianjiang H,

et al. Intelligent fault-tolerant 1553B
bus system based on adaptive learning.
In: IEEE International Conference
on Communication Software and
Networks. Xi an, China: IEEE; 2011.
pp. 378-381

[18] SalarKaleji F, Dayyani A. A survey

on fault detection, isolation and
recovery (FDIR) module in satellite
onboard software. In: International
Conference on Recent Advances in
Space Technologies (RAST). Istanbul,
Turkey: IEEE; 2013

18
Chapter

Dynamic Link from Liftoff

to Final Orbital Insertion
for a MEO Space Vehicle
Jack K. Kreng and Gleason Q. Chen

Abstract

During the entire launch sequences from liftoff to final orbital insertion of a
space vehicle (SV), adequate link requirements are to be maintained for telemetry,
tracking, and command (TT&C) for uplink and downlink services, from launch
vehicle (LV) and SV to ground stations (GS). A successful space vehicle launch
required adequate link coverage with good radio frequency (RF) performance. The
chapter is an extension of the IEEE/Aerospace Conference 2019 paper entitled
Dynamic Link Analysis and Application for a MEO Space Vehicle published by the
authors. The emphasis in this chapter is on the addition of the three distinctively
different tracking waveforms and their associated links, used from liftoff to final
orbital insertion. This chapter will describe the three required dynamic link analyses
(DLA) to cover (a) the LV link from liftoff to its end of line of sight (LOS), (b) the
LV link from LOS to Tracking and Data Relay Satellite System (TDRSS) at beyond
line of sight (BLOS), and (c) the final tracking link using Space-to-Ground Link
Subsystem (SGLS) or non-SGLS (NSGLS) link for the earliest or best separation
time of the SV from the LV. The chapter discusses the concept of the dynamic link
analysis, SV antenna switching schedule, recommended SV separation time, as well
as the performance for different launch scenarios within the 24-h launch window.
Topics include antenna patterns, launch trajectories, elevation angle and clock and
cone angle geometry, and dynamic link budget.

Keywords: wide band communications, satellite tracking

1. Introduction

Recently, there was an interest to extend the present dynamic link analyses
(DLA) beyond the early launch period to cover the period after the space vehicle
(SV) separation from launch vehicle (LV), which includes both booster and second
stage engine. The dynamic link from liftoff to final orbital insertion considers both
geometric (visibility coverage) and radiometric (link margins for all downlink and
uplink services) adequacy in the three launch stages. The purpose of the dynamic
link study for the launch is to provide the earliest and accurate time for final SV
separation and orbital insertion as compared to previous method which only relied
on visibility tracking coverage to the end of line of sight (LOS).
The present DLA typically covers only two stages of LV tracking, including
(a) liftoff to the end of LOS link and (b) the end of LOS to a period before SV

1
Satellite Systems - Design, Modeling, Simulation and Analysis

payload (PL) separation from LV, using LV, to Tracking and Data Relay Satellite
System (TDRSS) [1] satellite link, which is also called beyond line of sight (BLOS)
link. A third SV tracking, after SV payload separation from LV, is a tracking link
between SV and a ground station (GS). This third SV tracking is now added in this
chapter.
The tracking link used from liftoff to the end of LOS uses a UHF noncoherent
FSK signal for command and a digital FM or BPSK for tracking telemetry link as
described in detail in [2]. From the end of LOS to BLOS, the tracking telemetry link
is usually a BPSK or QPSK signal, using a NASA Tracking Data Relay Satellite
System to relay tracking data from the LV to White Sands or Goddard ground
station (WSGT/GRGT) and finally routing it to other user ground stations. After SV
payload separation and orbital insertion, the SV tracking link to an Air Force
Satellite Control Network (AFSCN) ground station [3] will use Space-to-Ground
Link Subsystem (SGLS) or a non-SGLS (NSGLS) waveform described in [3, 4] and
in Section 3 for tracking signals along the trajectory. In the following pages,
supporting link analyses for the two LV and one SV tracking stages will be
presented.

2. Geometry and coordinate definition for dynamic link analysis

2.1 Antenna coordinate system space vectors of interest

Figure 1 illustrates the antenna coordinate system space vectors of

interest [5].

2.2 Antenna coordinate system definition

Figure 2 defines the antenna coordinate system used in this chapter. The
azimuth (AZ or Ф) or clock angle is used in the antenna cut configuration. The
elevation angle (EL or ϴ) also called as cone angle is also used in the antenna
gain data file. The antenna gain is a changing variable as a function of
mission elapsed time (MET). The antenna gains are used in the following
dynamic links.

Figure 1.
Space vehicle and ground station vector definition.

2
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

Figure 2.
Antenna coordinate definition.

3. Dynamic link calculation

3.1 Dynamic link formulas of interest

This section provides a summary of the dynamic link model of interest [6]. More
detailed derivation of other variables, especially UVZBD, UVYBD, and UVXBD, can
be found in [5]. For station elevation angle, either LV or SV elevation or cone angle
(EL or theta or ϴ), and LV or SV clock or azimuth angle (AZ or Phi or Ф), we have1

1800


UVZBD
ð SRÞ
1
Cone Angle ¼ cos in Deg (1)
π jSRj

1800


0 1 GS
SR
EL ¼ 90 cos in Deg (2)
π jGSj ∗ jSRj

1800


1 UVXBD
ð SRÞ UVYBD
ð SRÞ
φ¼ tan 2 , in Deg (3)
π jSRj jSRj

Clock Angle ¼ Ф Offset

3.2 Tracking signals and link analyses along the trajectory

As mentioned in the introduction, there are three separate tracking signals along
the flight trajectory that we need to analyze ensuring that they have adequate link
margins of three dB or more. Most of the present DLA covers only two stages of
launch coverage and neglecting the third stage coverage. The requirement for third
stage tracking is explained below:

1
The authors would like to thank Dr. James Yoh for his derivation of these formulas.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

1. From liftoff to the end of LOS, the waveform for this link is generally a digital
FM or BPSK for telemetry downlink as defined in the Range Commander
Council (RCC) 119-88 [2]. The liftoff to LOS 5 link margin plot is shown in the
left half of Figure 3 for five different ground stations.

Figure 3.
TLM dynamic link margin from TEL4 to TDRSS versus mission elapsed time.

Table 1.
LV to TDRSS range TLM link (based on NASA source).

4
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

2. From the end of LOS to NASA TDRSS at geosynchronous orbit, the telemetry
link is a BPSK or QPSK (telemetry + data) which is sent from LV to TDRSS to
be relayed to White Sands or Goddard ground station (WSGT/GRGT), as
shown in the second right half of the Figure 3 and in more link details in
Table 1.

3. For the third link after SV payload separation, when the satellite or SV starts its
transfer orbit, the tracking link from the SV payload (or bus) to an AFSCN
ground station will be in SGLS, Unified S-Band (USB), or a NSGLS waveform
as described in more detail in [3, 4]. For a more secure tracking, SV normally
will be using SGLS link for tracking as described in [4], with a MEO satellite in
Table 2 as an example. A commercial and less secure SV launch may use USB

Table 2.
Link budgets for SGLS TT&C uplink and downlink services [4].

5
Satellite Systems - Design, Modeling, Simulation and Analysis

Table 3.
Typical C/No for uplink and downlink budgets.

or NSGLS for SV tracking instead of using SGLS waveform. Table 3 shows link
budget for uplink and downlink C/No example for tracking links 1 and 2.
Table 2 shows SGLS telemetry, tracking, and command (TT&C) link budget
for tracking link 3 for a MEO satellite. If the SV is using a USB or a NSGLS [3],
the tracking waveform can be an AQPSK signal with telemetry on the I
channel and ranging on the Q channel.

4. Basic link parameters and formulas

This section describes the basic link parameters including LV or SV transmitter

power amplifier gain (Pt), transmitter antenna gain (Gt), space loss (Ls), received
isotropic power (RIP), and received (C/No = SNR).
A modulation signal or information data is generated at a ground station, in an
LV or in an SV. This modulation signal will be used to modulate onto the radio
frequency (RF) carrier to become a modulated transmit signal. This transmit sys-
tem will be consisting of a high-power amplifier (HPA) which amplifies the signal
to generate an output power expressed in dBW (conversion from Watts to dBW is
simply dBW = 10*log10(Watts)); some cables and circuits with a loss and an
antenna with a gain are added together as shown below. The output from the
transmit system is therefore an effective isotropic radiated power or EIRP, which
can be found in either the uplink or the downlink of an LV or an SV tracking
system.

EIRP ¼ PT þ LC þ Gt in dBW (4)

where Gt = transmit antenna gain, in dBi; LC = transmit circuit loss, in negative

dB; and PT = HPA output power, in dBW.

6
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

For the uplink, where the transmit antenna is located on the ground, the antenna
can be easily directed to an LV or SV. This transmit antenna is likely to be a
directional high gain dish antenna for connectivity with an LV or SV located possi-
bly far away. In Table 3, line 9, a typical ground station has a large parabolic
antenna dish with a gain Gt = 43.94 dBi using the following formula.


π ∗ fC ∗ D
Gt ¼ 10 ∗ log 10 η ∗ in dBi (5)
c

where η = average antenna efficiency (0.70 in the calculation in Table 3);

fC = uplink frequency, in Hz; D = antenna diameter, in m; and c = speed of light, in
m/s.
For the downlink, the transmit antenna is typically an omnidirectional antenna
that covers a larger portion of the sky or the Earth, in which case the antenna gain is
small (e.g., 2 dBi) and is either specified as in line 9 of Table 3 or can be interpo-
lated from values extracted from a table of antenna gain pattern with a specific AZ,
EL, and MET, using a mission specific launch trajectory.
In general, there are terms that may be added to Tables 2 and 3. For example,
the uplink transmit antenna in Table 3 may have two more loss terms, namely,
radome loss to account for any loss for a radome and pointing loss to account for
any pointing error in directing the boresight of the antenna. For Table 3 they are
both negligible and ignored except for the polarization losses. The transmit and
receive polarization losses (lines 14 and 22, respectively, in Table 3) can be
accounted for as one-single combined receive polarization loss.

4.1 Transmission medium and losses

The signal path traverses through the transmission medium, in between transmit
and receive systems. When the distance between transmit and receive systems
increases, the signal beam has an angular spread which decreases the signal power
collected by a receiving antenna. We know that the portion of the transmission
medium near the ground station depends on the Earth’s atmosphere which attenuates
the signal to different degrees, dependents on the frequency, the altitude of the GS,
and the angle of the signal path through the atmospheric (GS elevation angle).
Beyond the Earth’s atmosphere, the signal path traverses through the space with little
atmospheric attenuation, only with free space loss to account for. Therefore, there are
essentially two losses through the transmission medium, namely, the space loss to
account for the spreading of the signal beam and the additional atmospheric loss [7].
"
2 #
c
LS ¼ 10 ∗ log 10 in dB (6)
4π ∗ f C ∗ SR

where fC = carrier frequency, in Hz; c = velocity of light, in m/s; and SR = slant

range between GS and SV, in m.
At L and S bands, the atmospheric loss is very small, at less than 0.001 dB/Km or
0.1 dB/100Km for a link availability of better than 98% [8].

4.2 Received isotropic power

The transmit signal, after accounting for the space and atmospheric losses, and
its signal path terminates at the antenna of either the ground station, the SV, or the
LV receiving system. Before considering the characteristics of the receive antenna

7
Satellite Systems - Design, Modeling, Simulation and Analysis

and the receiver, a good indication of the signal strength is given by the received
isotropic power. RIP is simply the transmitter EIRP after subtracting off the losses
of the transmission medium, i.e.

RIP ¼ EIRP þ LS þ LA in dBW (7)

where LA is the atmospheric loss extracted from tables or curves, in negative dB

(very small at 0.02 dB/Km per Datron chart in L and S bands). Ls can also be
obtained from the Datron calculator [8].

4.3 Receive system and performance

The last portion of Table 3 addresses the receiving system and assesses how well
it performs. This section involves with the calculation of the signal strength and the
noise strength, resulting in the ratio of signal power over noise power density
(C/N0). In general, we first address the signal power and then the noise power
density. Line 21 provides for the receive antenna size consistent with the receive
antenna gain to be calculated later. The next parameter in Table 3 is the polariza-
tion loss, which accounts for the mismatching between the polarization axial ratios
of the received signal and the receiving system. The axial ratio is the ratio of the
major axis of an ellipse to its minor axis. For circularly polarized signal, the ratio
should be 0 dB. Any deviation from 0 dB results in a polarization loss. Line 23 shows
the values of receive antenna gain. For downlink in which the receive antenna is a
dish antenna located at GS, Gr is calculated using the standard dish antenna equa-
tion (similar to Eq. (6) for Gt).


π ∗ fC ∗ D
Gr ¼ 10 ∗ log 10 η ∗ in dBi (8)
c

where η = average antenna efficiency (assumed to be 0.6 in the calculation in

Table 3); fC = downlink frequency, in Hz; D = antenna diameter, in m; and
c = velocity of light, in m/s.
At the end, the received power at the antenna feed is just the sum of RIP, minus
the polarization loss, plus the receive antenna gain, i.e.

C ¼ RIP þ LP þ Gr in dBW (9)

where LP is the polarization loss, in negative dB.

For the downlink transmit antenna on the SV, as in the case of uplink receive
antenna, the SV antenna is a broad beam Earth coverage (EC) omnidirectional type
of SV antenna, with a gain of 2 dBi (see line 23 of Table 3).
For the noise power density (N0), we need to calculate the system temperature
(TS) measured at the antenna feed. The system temperature is the sum of antenna
sky temperature (TA) and the composite temperature from antenna line loss (LL)
and low noise amplifier noise figure (NF) which are referred to the antenna feed. In
linear quantity, TS is given by [1].
 
ðLL þNFÞ=10
T S ¼ T A þ 10 1 ∗ 290 in Deg K (10)

where NF = low noise amplifier noise factor, in dB and LL = line loss, in dB.
The noise density (N0) is given by

8
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

 þ 10 ∗ log ðT S Þ in dBW=Hz
N 0 ¼ kdB (11)

where k = Boltzmann’s constant in dB = 228.6 dBW/KHz.

Using Eqs. (9) and (11), the ratio C/N0 at the receiver input is obtained in
Table 3, line 31. It represents the final product before going into specific service(s)
such as telemetry and ranging to evaluate their performance.

5. Space vehicle link services

For many SVs, we are interested in their uplink and downlink services. Table 2
shows an example, taken from an IEEE paper [4]. This is the standard link budget,
where the ground station is an AFSCN [3] remote tracking station (RTS) using
SGLS waveform [3, 4]. The waveform is described in an AFSCN interface control
document (ICD) [3] and is implemented in DLA, although other waveforms can be
readily incorporated. The uplink has two services of interest—carrier and command
—while the downlink has three services of interest: carrier, ranging, and telemetry.
In general service margins are calculated for these five services. For the SGLS
waveform, command is coupled with ranging and modulated on the uplink carrier;
therefore command is also turned around at the SV along with ranging. This SGLS
turnaround process explains the reason that Table 2 shows a power allocation for
command in the downlink and no calculation for its margin. As a result, downlink
power allocated to command is essentially wasted while robbing power from other
downlink services. The requirements and service margins for command and telem-
etry are expressed in Eb/N0, since it is the bit error rate (BER) that counts for both
cases. The carrier and ranging are expressed in C/No given by a specific station. For
ranging, it is the autocorrelation value between the decoded ranging code and the
transmitted ranging code that needs to be maximized in order to successfully
perform accurate ranging.
Table 2 represents uplink and downlink budgets for SGLS TT&C. Let us address
the important aspects of the calculation of uplink and downlink services in the next
few subsections. The role of modulation indices is to divide up the power for
allocation to services. The modulation index is expressed in radians so that it can go
right in as an argument in a sinusoidal or Bessel expression. If the modulation
indices of all services are zero radians, no power is allocated to the services, and the
carrier retains all the link power calculated in Section 4. If the modulation indices of
services are not zero, portions of the power are taken from the carrier and allocated
to the services. The remaining power stays with the carrier as the “residual carrier
power.”

5.1 SNR and link margin calculation

After SV separation, we are dealing with the SV uplink and downlink using
SGLS or NASA Unified S-Band waveforms as described in [3, 4]. For telemetry
service, the requirement is SNR = Pservice/NoB = Eb Rb/NoB = Eb/No in dB. For
carrier and ranging, the requirements are stated in terms of C/No as mentioned
before. For acquisition, the uplink carrier loop bandwidth could be as high as +/
100 KHz, while its tracking bandwidth could be as small as a few Hz. For the station
the carrier tracking loop bandwidth is about 20–50 Hz, as in Table 2 in line 36. For
ranging, the bandwidth of 10 Hz represents ranging tracking loop bandwidth
(Table 2, line 56), which corresponds to the sampling rate of the autocorrelation

9
Satellite Systems - Design, Modeling, Simulation and Analysis

Uplink service modulation losses for SGLS and NASA USB

No. Uplink modulation Carrier Ranging Command

1 AM-3FSK/PRN RNG/PM (SGLS unfiltered J 2o ðβ1 Þ: cos 2 ðβ2 Þ J 2o ðβ1 Þ: sin 2 ðβ2 Þ 2J 21 ðβ1 Þ: cos 2 ðβ2 Þ
uplink) [3, 4]

2 BPSK/PRN RNG/PM (USB filtered case) J02(β1)cos2(β2) J02(β1) sin2(β2) 2J12(β1) cos2(β2)
(Eqs. (1-22) and (1-23)) [9]

3 BPSK/Tone RNG/PM (USB unfiltered J02(β1) J02(β2) 2J02(β1)J12(β2) 2J02(β2) J12(β1)

case) +… … + … ..
(Eqs. (1-18) and (1-21)) [9]

Downlink modulation losses for SGLS and NASA USB

No. Downlink modulation Carrier Ranging Telemetry

4 PRN RNG/PSK TLM/PM (SGLS filtered J02(β01 )J02(β02 ) 2J02(β01 ) J12(β02 ) 2J02(β01 )J02(β02 )
case) J02(β3 ) J02(β3) J12(β3 )
(Eqs. (2-18) and (2-20)) [4]

5 Tone RNG/PSK TLM/PM (USB filtered J02(β3 )J02(β02 ) 2J02(β3 )J12(β02 ) 2J02(β02 )J12(β3)
case)
(Eqs. (2-18) and (2-21)) [9]

Table 4.
Uplink and downlink service modulation losses for SGLS and NASA USB.

value between the detected ranging code and the transmitted ranging code. For
command and telemetry, the requirements are expressed in terms of Eb/N0. The
command and telemetry data bit rates of 1000 bps each are representing the lower
end of their SGLS choices. As shown in Table 2, the results from the SNR calcula-
tion are the values of C/N and Eb/N0 for various received uplink services (lines 37
and 48 for command C/N or Eb/No) and for various downlink services (line 55 for
ranging service Prng/No and line 70 for telemetry service Eb/No). The uplink
service modulation losses for SGLS and NASA USB with subcarrier (S/C) [3, 4, 9]
are shown in Table 4. The downlink service modulation losses for SGLS and NASA
USB with subcarrier (S/C) [3, 4, 9] are shown in Table 4. Also note that β1, β2, and
β3 represent the modulation indices for command (CMD), ranging (RNG), and
telemetry (TLM), respectively, per [3, 4, 9]. These uplink and downlink modula-
tion losses are in lines 34, 44, and 54 in Table 2.
For NSGLS waveforms such as the direct mod BPSK, QPSK, and AQPSK, the
service mod losses are negligible. Finally, the calculated service SNR in Table 2 is
compared with the required SNR to obtain the link margin for each service. The
required SNR values capture all the performance requirements for the services,
such as ranging accuracy, tracking loop loss likelihood, bit error rate, and others.

6. Launch vehicle dynamic link

Before SV separation from the LV, we are also interested in the dynamic link
from a ground station to the LV, from liftoff to the SV after separation, along the
entire LV flight path using the tracking stations in the line of sight (TEL4, JDMTA,
ANT, DGS, TDRSS). The waveforms for this LV tracking are described in Range
Commander Council (RCC) handbook [2]. One must ensure that the downlink
telemetry link from the launch vehicle to these ground stations and TDRSS relay
satellite are positive as can be seen in Figure 3. The basic LV range modulations are
digital FM, BPSK, QPSK, AQPSK, etc. as discussed in RCC [2]. In Figure 4, dynamic
LV slant range “received TLM Eb/No” and “TLM link margin” for a specific mission

10
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

Figure 4.
LV slant range received TLM Eb/No and TLM link margin.

are displayed together. As an example a specific LV to TDRSS BPSK link using

NASA data is shown in Table 1.

7. Conclusion

This chapter discusses the required three DLAs and related tracking waveforms
to cover the three launch stages, namely, (a) the launch vehicle tracking link from
liftoff to its end LOS using the digital FM or BPSK signal, (b) the launch vehicle
tracking link from LOS to TDRSS at BLOS using NASA USB signal, and (c) the final
tracking link from SV to an AFSCN ground station using AFSCN SGLS, AFSCN
NSGLS, or NASA USB waveforms. In the third tracking link case, BPSK, QPSK, or
AQPSK waveforms were used, in which for QPSK and AQPSK, the telemetry data is
put on the I channel and the ranging signal is on the Q channel.
The chapter shows that good telemetry link margins from LV to tracking sta-
tions such as TEL4, JDMTA, and ANT or to a NASA TDRSS relay satellite can be
achieved using digital FM, BPSK, QPSK, or AQPSK signals, after SV separation. The
chapter also shows that good tracking link margins can be achieved from SV to
AFSCN ground stations, including IOS or DGS as the first contact station.

11
Satellite Systems - Design, Modeling, Simulation and Analysis

Author details

Jack K. Kreng* and Gleason Q. Chen

Aerospace Corporation, El Segundo, CA, USA

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

12
Dynamic Link from Liftoff to Final Orbital Insertion for a MEO Space Vehicle
DOI: http://dx.doi.org/10.5772/intechopen.92462

References

[1] Sklar B. Digital Communications

Fundamentals and Applications. New
Jersey: Prentice Hall; 1988. p. 221

[2] RCC, Range Commander Council,

Telemetry Group. Document 119-88;
1988. (approved for public release)

[3] Air Force Satellite Control Network

(AFSCN) Interface Control Document:
Range Segment to Space Vehicle, ICD-
000502A, prepared by The Aerospace
Corporation for Space and Missile
Systems Center; 2011

[4] Kreng JK, Ardeshiri MM. Effects of

turnaround command on SGLS C/No
and SNR performance. In: 2018 IEEE
Aerospace Conference, Paper #2513;
2018

[5] Krikorian YY et al. Dynamic link

analysis tool for a telemetry downlink
system. In: 2004 IEEE Aerospace
Conference Proceedings; 2004.
pp. 3994-4001

[6] Spiegel MR, Lipschutz S, Liu J.

Mathematical Handbook of Formulas
and Tables, Third Edition, Section 10.
New York: McGraw Hill; 2009

[7] Kreng J, Do S, Mathur A. USB

Command Link Via TDRSS to Satellites
and Possibly Space Launch Vehicles
(LV). AIAA; 2008

[8] Datron/Transco Inc., USA. Antenna,

Space and Atmospheric Calculator

[9] Kreng J, Krikorian Y, Raghavan S.

Telemetry, tracking, and command link
performance using USB/STDN
waveform. IEEE/Aerospace Journal.
2007

13
Chapter

Effective Algorithms for Detection

Outliers and Cycle Slip Repair in
GNSS Data Measurements
Igor V. Bezmenov

Abstract

The chapter describes effective algorithms that are often used in processing data
measurements in Global Navigation Satellite Systems (GNSSs). Existing effective
algorithm was developed for detection and elimination of outliers from GNSS data
measurements. It is based on searching for a so-called optimal solution for which
standard deviation and maximum absolute deviation of the measured data from
mean values do not exceed specified threshold values, and the number of the
detected outliers is minimal. A modification of this algorithm with complexity of
N log 2 N is discussed. Generalization of the existing algorithm to the case when data
series included some unknown trend will be presented. The processing trend is
assumed to be described by an unknown function of time. The generalized algo-
rithm includes the outlier detection algorithm and trend searching algorithm that
has been tested using simulated data. A new algorithm will be presented for cycle
slip repair using Melbourne-Wübbena linear combination formed from GNSS data
measurements on two carrier frequencies. Test results for repair data in the case of
multiple (cascade) cycle slips in actual observation data will also be presented in
this chapter.

Keywords: global navigational satellite systems (GNSSs), GNSS measurements,

outliers, data screening, optimal solution, trend function, Melbourne-Wübbena
combination, cycle slips

1. Introduction

At present, five constellations of GNSS satellites are involved in the formation of

observational data, which serve as a source for many applications related to navi-
gation, geodesy, geodynamics, and in the performance of solving of many funda-
mental problems. These are American Global Positioning System (GPS), Russian
Global Navigation Satellite System (GLONASS), European Galileo, Chinese
BeiDou, and Japanese Quasi-Zenith Satellite System (QZSS). The satellites of each
of the operating systems transmit signals, as a rule, on two L-band carriers, which
are received by GNSS receivers. A large number of stations equipped with GNSS
receivers and located around the World are part of the International GNSS Service
(IGS) network. These stations generate observation data files and transmit them to
international databases in real time [1, 2], after which these data become available
for use by many institutions and laboratories over the World. When solving

1
Satellite Systems - Design, Modeling, Simulation and Analysis

applications, the measurement data go through various processing steps. Significant

element of the data processing is the detection of rough measurements and removal
them from the further processing. Despite the fact that many of the laboratories use
a high-end application of the software regarding accuracy, reliability, and robust-
ness, the presence of rough measurements in the observational data excludes the
possibility of obtaining an accurate final result. In order to obtain results of unprec-
edented accuracy, the measurement data must be cleared of coarse measurements
or outliers. It should be noted that the concept of outliers is key in the measurement
processing theory [3], and there is no general definition for it. In order to distin-
guish outliers from the rest of the measured data, in some cases, the deviation of the
data series values from some average value of the data is considered. If the deviation
from the average is exceeded by a predetermined threshold value, the measured
value is considered as an outlier. Such an approach has a significant disadvantage
that the exact mean is generally unknown, and the estimate obtained by averaging a
series may be very inaccurate due to outliers. Existing iterative procedures are also
based on the idea of calculating deviation from the average and often result in the
unjustified rejection of many observations. Reducing the data involved in
processing may, in turn, result in a loss of accuracy of the final result.
This chapter describes the outliers cleaning algorithm for GNSS data. The pro-
posed algorithms are based on the search for the so-called optimal solution with the
minimum amount of invalidly rejected data. The algorithm for accelerated detection
of outliers in a large amount of measurements has been developed, as well as an
algorithm for detecting outliers in data containing an unknown trend. In conclusion,
the algorithm of jump detection in the Melbourne-Wübbena combination [3–5],
including the developed procedure of cleaning data from outliers, is considered.
In Section 2, the problem of searching for the so-called optimal solution is
formulated. Section 3 provides a search algorithm, with a common number of
arithmetic operations not exceeding N2. Section 4 presents the test results for
actual measurements in global navigational satellite systems at two carrier frequen-
cies. The searching of outliers was performed in the Melbourne-Wübbena combi-
nation. In Section 5, the assertions that are the mathematical prerequisites for
justifying a fast outlier search algorithm are proved. In Section 6, the fast outlier
detection algorithm with the number of arithmetic operations of Nlog2N is pro-
posed. Section 7 describes the case of data with unknown trend. Iterative procedure
of outlier search is proposed based on the finding of suitable trend approximation in
polynomials class. The idea of excluding coarse measurements is based on finding a
so-called minimizing set of measurement data of a given length. This distinguishes
the proposed algorithm from known similar procedures in which outliers are
detected by exceeding a preset threshold. The test results with simulated data are
given. Sections 8 and 9 discuss the problem of detecting jumps in the Melbourne-
Wübbena combination. An algorithm is proposed that includes the outlier cleaning
procedure based on the search for the optimal solution. Section 10 shows the
numerical calculations with real data for algorithms presented in Sections 3–9.
Section 11 concludes the chapter.

2. Statement of the problem: a Melbourne-Wübbena combination

Often, measurements yj taken at moments of time j can be presented in the

form:

yj ¼ f j þ z þ ξj ; j ¼ 1 … N, (1)

2
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

where f j is a trend function, depending on physical process and as a rule is

unknown, the sum z þ ξ j is unknown random variable imposed on the trend with
unknown constant z and a centered random variable ξ j .
Detection of outliers in data series expressed in Eq. (1) with unknown trend is
uncertain since the concept of measurement or outliers itself is uncertain. In many
cases, however, the trend function is known a priori.
For example, many data processing programs often use different linear combi-
nations formed of code and phase measurement data to eliminate unknown param-
eters. One such combination is the Melbourne-Wübbena combination composed of
both, carrier phase and code observables as described by Melbourne [4] and
Wübbena [5]. This combination eliminates the effect of the ionosphere, the geom-
etry, the clocks, and the troposphere [3], and it is often used to detect loss of carrier
phase capture in the preprocessing stages. The Melbourne-Wübbena combination
generated for a specific satellite-receiver pair can be presented in the form of the
sum of three terms [6]. One of the terms includes the integer wide-lane ambiguity
for the two carrier frequencies [3]; the second component accounts for the satellite
and receiver instrumental delays; and the third component is the measurement
noise, including carrier phase and code multipath. Thus, during a time interval
where the integer wide-lane ambiguity does not change, the Melbourne-Wübbena
combination can be written as formula (1) with f j ¼ const.
Another example is satellite clock correction values derived from navigation
message data, which can also be represented as in Eq. (1) with f j ¼ dj þ a, where d
and a are the drift and offset parameters known from navigation data, respectively.
In the case where the trend is known, the measurement data after the trend
subtraction can be represented as, assuming N observations:

yj ¼ z þ ξj ; j ¼ 1 … N, (2)

with an unknown constant z, which we cannot be determined in advance,

because the random value ξ j is not known and may contain outliers.
In Sections 2–6, we consider the case where the trend is known a priori, that is,
the data can be presented as Eq. (2). A problem with an unknown trend will be
discussed in Section 7. In principle, the outlier detection procedure described below
is not affected by the measurement format expressed in Eq. (1) or (2); it can be
applied to any set of data measurements yj .
The preliminary processing task includes rejection of rough measurements or
outliers from data series (2). In other words, it is necessary to find a set YL ¼
n o
yj , … , yj of L elements, where L is the length for which the following conditions
1 L

are satisfied:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2
X
σYL ¼ ðL 1Þ 1 j ∈ j ,::,j yj z ≤ σmax , (3)
f 1 Lg
  n o
yj z ≤ 3
σmax ; yj ∈ yj , … , yj ,
 
(4)
1 L

L ≥ MINOBS, (5)

where σYL and σmax are the standard deviation (SD) and their associated
specified threshold values; MINOBS is a parameter limiting from below the length
of the required set of measured values (e.g., 10), and we will assume hereafter that
MINOBS < N.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

The values yj that are not included in the set YL are treated as coarse measure-
ments and removed from further processing. Typically, expressions in Eqs. (3)–(5)
are the only conditions considered in processing programs when screening out
rough measurements. Below we will formulate the problem of searching for a
solution, complementing the conditions expressed in Eqs. (3)–(5) with two extreme
conditions [7].

1. First, we will require that the length of the set sought be the maximum, that is,
the number of measurements deemed to be coarse is the minimum:

L ! max : (6)

Note that for the predetermined values yj , the problem solution satisfying
conditions in Eqs. (3)–(5) may not exist (e.g., when yj includes a trend, in
particular when yj is an arithmetic progression with a step greater than σmax ). In
the case when the solution does exist, we will denote the value L at which the
maximum of Eq. (6) is reached as Lmax . Note that the condition expressed in
Eq. (6) does not ensure the uniqueness of the set because several sets of length
Lmax can be found that satisfy the conditions in Eqs. (3)–(5).

2. From all possible sets that satisfy conditions expressed in Eqs. (3)–(5) and (6),
we will select the one for which the variable σYL receives the smallest value:
σYL !
n o min : (7)
YL ¼ yj , … , yj ; L¼Lmax
1 L

Let us define Yopt as follow:

Definition 1. For a given sequence of values y j , j = 1, 2, … N, the set of values:
 
Yopt ¼ yj , … , yj , (8)
1 L max

satisfying conditions in Eqs. (3)–(7), we refer to as the optimal solution of the problem
expressed in Eqs. (3)–(7). The corresponding SD value is denoted by σopt .
Thus, the problem consists in the creation of a search algorithm for the optimal
solution of the problem shown in Eqs. (3)–(7).
In a practical situation, the precise value z, given conditions in Eqs. (3) and (4),
is not known. We will estimate the values using the following formula:
X
z¼L 1 yj : (9)
j ∈ fj1 , … , jL g

Note that the value z depends on the required solution, which will complicate its
search.
Usually, iterative methods are used to find a solution to problem expressed in
Eqs. (3)–(5). For example, the algorithm implemented in the observation data
smoothing program (see [3]) is designed to find a set Y satisfying the conditions
given by Eqs. (3)–(5). The proposed step-by-step algorithm is based on iterations
(the index number of iteration is designated by the upper index in parentheses):
Step 1: Initialization: Y ð0Þ ¼ y1 , … , yN ; Levelð0Þ ¼ 1020 ; k = 0.
Step 2: Checking the length of the set Y ðkÞ ; if it is less than MINOBS, then the
process comes to an end and a solution is not found.

4
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

Step 3: Calculation of the values zðkÞ and σðkÞ on the available set Y ðkÞ using
formulas (3) and (9).
Step 4: Checking the fulfillment of the inequality σðkÞ ≤ σmax. If it is satisfied, the
set Y ðkÞ is accepted as the solution, and the search process comes to an end.
Otherwise, there is a transition to Step 5.
Step 5: Definition of Levelðkþ1Þ for outlier detection:

Levelðkþ1Þ ¼ 3
σðkÞ ;

In order to prevent an infinite loop of iterations, a required verification is carried

out:

Levelðkþ1Þ < LevelðkÞ :

If this inequality is not satisfied, then the following is assumed:

Levelðkþ1Þ ¼ LevelðkÞ =2;

Step 6: Definition of a new set Y ðkþ1Þ to include those and only those y j for which

 
z  ≤ Levelðkþ1Þ :
ðkÞ 

y j

Step 7: Increasing k by 1: k++. Transition to Step 2.

Note that the optimal solution cannot be found in such a manner, as confirmed
by numerical calculations (see Section 4).

3. Algorithm for solving the problem

Let us formulate a statement that is the key to the creation of an effective search
algorithm for the optimal solution(Eqs. (3)–(7)).

Assertion 1. Let the set Y opt ¼ y j , …,y j be optimal for a specified sequence
1 L max
n o
of values y j and

   
ymin ¼ min y j , … , y jL , ymax ¼ max y j , … , y jL ,
1 max 1 max

then the interval ymin , ymax does not contain values y j that are not in the set Y opt .
Proof. In fact, let us assume the opposite: Let y j ∉ Y opt , ymin < y j < ymax and yk
and yl be values from the set Yopt for which yk = ymin and yl = ymax. One of these cases
is possible:
   2
2
a. z < y j and, therefore 0 < y j z < yl z => y j z < yl z ,

   2
2
b. z ≥ y j and, therefore 0 ≤ z yj < z yk => y j z < yk z .

5
Satellite Systems - Design, Modeling, Simulation and Analysis

In the first case, Case (a), we replace the value yl in the set Yopt with y j . In the
second case, Case (b), we replace yk with y j . In any of the cases, we will have
another set of the same length Lmax for which conditions expressed in Eqs. (3) and
(4) are satisfied, but the SD, as follows from Eq. (3), is less than σopt . Consequently,
the set Yopt is not optimal since requirement expressed in Eq. (7) is not satisfied.
The contradiction that is reached proves Assertion
 1.
Further, note that if Yopt ¼ yj , … , yj is the optimal solution of the prob-
1 L max

lem given by Eqs. (3)–(7), then an arbitrary permutation of the numbers

yj , … , yj will also be the optimal solution of the problem described by
1 L max

Eqs. (3)–(7). Thus, the optimal solution does not depend on the arrangement of
given numbers yj . This allows us to arrange the numbers yj in the order most
suitable for searching the optimal solution.
n o Taking advantage of this important
circumstance, we arrange the values yj in the ascending order and we will look
for the optimal solutionnin o
the ordered sequence. For brevity, the ordered sequence
will also be denoted by yj . Thus, y1 ≤ y2 ≤ … ≤ yN . Moreover, for simplicity of
logic, we will assume that all yj are different, that is,

y1 < y2 < … < yN (10)

Note that due to the formulated Assertion 1, if Yopt is the optimal set and y j and
1
yj are its smallest and greatest values, respectively, then all values of yj from the
L max

interval y j , y j belong to Yopt . Consequently, considering Eq. (10), we have
1 L max

yj ¼ yj þL 1 and
L max 1 max

n o
Yopt ¼ yj , yj þ1 , … , yj þLmax 1
1 1 1

Thus, the optimal solution should be sought in the ascending sequence yj among
all possible sets yk , … , ykþL 1 of length L with k and L satisfying the following
conditions:
MINOBS ≤ L ≤ N, (11)
1≤k≤N L þ 1: (12)

Hence, instead of searching for all possible sets of various length, numbering 2N ,
for the solution of the problem described by Eqs. (3)–(7), it is sufficient to vary just
the two parameters, k and L, associated with the conditions expressed in Eqs. (11)
and (12). The number of pairs of integer numbers k and L subject to condition in
Eqs. (11) and (12) is equal to:

ðN MINOBSÞðN MINOBS þ 1Þ=2:

Let L ¼ l k þ 1 be the length of an arbitrary set yk , … , yl . We introduce the

designations:

1 kþL
X1
zðk; LÞ ¼ y, (13)
L j¼k j

6
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

kþL
X1 
1 2
σ2 ðk; LÞ ¼ yj zðk; LÞ : (14)
L 1 j¼k

We rewrite conditions given by Eqs. (3) and (4) in the new designations:

σ2 ðk; LÞ ≤ σ2max , (15)


ykþL 1 zðk; LÞ ≤ 3
σmax
: (16)
zðk; LÞ yk ≤ 3
σmax

Note that the two last inequalities directly follow from Eq. (4), monotony of yj,
and obvious inequalities: yk ≤ zðk; LÞ ≤ ykþL 1 .
Remark. In the conditions expressed in Eqs. (15) and (16), L means the length of the
set under checking, and k is the index of the smallest number included in the set. Although
“k” and “L” are also encountered as indexes in the sets we use below for monotonically
increasing sequences, we hope nevertheless that this will not lead to confusion.
The following recursive relationships are available, making it possible to find
zðk; LÞ and σ2 ðk; LÞ through the calculated values zðk; L þ 1Þ и σ2 ðk; L þ 1Þ for seven
arithmetic operations:

zðk; LÞ ¼ zðk; L þ 1Þ þ ALþ1
zðk; L þ 1Þ ykþL , (17)

2
σ2 ðk; LÞ ¼ BLþ1
σ2 ðk; L þ 1Þ CLþ1
ykþL zðk; L þ 1Þ , (18)
1 L 1 L
AL ¼ ; BL ¼ ; CL ¼ : (19)
L 1 L 2 ðL 1ÞðL 2Þ

The values of the fractions may be computed in advance as elements of a one-

dimensional array. Analogously, the following formulas can make it possible to
express zðk þ 1; LÞ and σ2 ðk þ 1; LÞ through zðk; LÞ and σ2 ðk; LÞ:

zðk þ 1; LÞ ¼ zðk; LÞ þ ALþ1
ykþL yk , (20)

σ2 ðk þ 1; LÞ ¼ σ2 ðk; LÞ þ ALþ1
ykþL yk ykþL zðk; LÞ þ DLþ1 yk zðk; LÞ , (21)
L
DL ¼ : (22)
L 2

The algorithm described below is based on the search for all possible pairs (k, L),
where L denotes the length of the set to be checked and k is the index of the smallest
of the values included in the set. At that, k and L must satisfy conditions Eqs. (11)
and (12). The set yk , … , ykþL 1 corresponding to each such pair must be checked
for fulfillment conditions (15) and (16).
We organize this search according to the algorithm described below, at each step
of which we check the validation of Eqs. (15) and (16) for all possible sets of a
certain length. We start the examine process with Step 1, where we check the set of
maximum length N. Further, with each next step, we will reduce the length of the
sets to be checked by 1.
Step 1: We consider the set of length N. There is only one such set: y1 , … , yN .
We check it for fulfillment conditions expressed in Eqs. (15) and (16). If they are
satisfied, this set is selected as a solution, and further search stops. Otherwise,
transit to Step 2.
Step 2: We consider the sets of length N 1. There are two sets of length N 1.

7
Satellite Systems - Design, Modeling, Simulation and Analysis

y1 , … , yN 1 and y2 , … , yN :

We test each of these sets for compliance with the conditions specified by
Eqs. (15) and (16). If none of them satisfies conditions (15) and (16), then we
transit to the next step. Otherwise, the following options are available:

• Option 1: if only one set from them is found that satisfies conditions (15) and
(16), then it will also be the solution of the stated problem; the search process
stops here, where Lmax = N 1.

• Option 2: if both sets simultaneously satisfy conditions (15) and (16), we will
select the set corresponding to the smallest of two values σ2 ð1, N 1Þ or
σ2 ð2, N 1Þ, and the search process stops here, where Lmax = N 1.

Step N L + 1: We consider the sets of length L. If L < MINOBS, then the search
process stops, and a solution is not found. For L ≥ MINOBS, we examine N L + 1
sets of length L:

y1 , … , yL , y2 , … , yLþ1 , … , yN Lþ1 , … , yN : (23)

We check each of these sets for fulfillment of conditions (15) and (16). If any of
them does not satisfy these conditions, then we transit to the next step where we
consider the sets of length (L 1). Otherwise, two options are possible:

• Option 1: if only one set from (23) is found that satisfies the conditions of (15)
and (16), then it will also be the solution of the stated problem with Lmax = L,
and the search process stops here.

• Option 2: if several sets simultaneously satisfy conditions (15) and (16), we

chose the set for which the value σ2 ðk, LÞ appears smallest as the solution, and
the search process stops here, where Lmax = L.

In order to calculate the values zðk, LÞ and σ2 ðk, LÞ, we use the recursive formu-
las (17)–(22) in accordance with a search scheme shown in Figure 1 where only the
zðk, LÞ is involved.
In accordance with the proposed arrangement, we calculate the values zð1; NÞ
and σ2 ð1; NÞ in the first step of the algorithm using formulas expressed in Eqs. (13)
and (14), and 4 N arithmetic operations are required for this arrangement. In order
to find zðk, LÞ and σ2 ðk, LÞ on all subsequent steps, we apply the recursive formulas
(17)–(22), making it possible to calculate the values of the specified pairs of

Figure 1.
Scheme of calculations when finding the optimal solution.

8
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

variables, each for the seven arithmetic operations, based on the results of the
calculations of the preceding step. So, for example, on the second step, proceeding
from the known values zð1; NÞ and σ2 ð1; NÞ, we find the values zð1; N 1Þ and
σ2 ð1; N 1Þ (vertical arrow on the diagram) by using formulas in Eqs. (17)–(19)
and the values σ2 ð2; NÞ and σ2 ð2; NÞ (horizontal arrow on the diagram) by using
Eqs. (20)–(22). In the general case, the transition to the following step in the
direction of the vertical arrows (see diagram) is carried out according to formulas
(17)–(19), and in the direction of horizontal arrows, according to formulas
(20)–(22). The number of arithmetic operations required to find the solution should
not exceed:

4N þ 9ððN Lmax þ 2ÞðN Lmax þ 1Þ=2 1Þ (24)

In the above number of computations, the computational costs of verifying the

satisfaction of inequalities (15) and (16) are also considered, which comprise from 0
to 2 arithmetic operations.

4. Results of calculations using algorithm presented in Section 3

We test the algorithms discussed above on the real data obtained by the ONSA
station that is a part of the IGS network [2]. These data are included in the distri-
bution kit of the installation software package [3] and available for usage. We
consider measurement data received from global positioning system (GPS) satellite
with system number PRN = 12 for 2010, day 207 to check the efficiency of the
proposed algorithm described in Section 3. Figure 2 plots the values of the
Melbourne-Wübbena combination over a time interval of 89.5 min (N = 180). The
index numbers j of time epochs counting from the beginning of a 24-h period with a
30-second interval are plotted on the horizontal axis. The values y j of the combina-
tion are plotted on the vertical axis and are expressed in cycles with wavelength
λ5 ≈ 0:86 [3]. Figure 3 shows the values of deviations from the mean of the data
cleared of outliers using the algorithms described in Sections 2 and 3. In both cases,
σmax = 0.6 and MINOBS = 10. The values y j z are plotted on the vertical axes in
cycles with wavelength λ5 and the index numbers j of epochs on the horizontal axis.

Figure 2.
Melbourne-Wübbena combination for ONSA station (GPS satellite, PRN = 12 for 2010, day 207).

9
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 3.
(a) Deviations of values of the Melbourne-Wübbena combination from the mean value after data cleaning
from outliers using the algorithm described in Section 2. (b) Deviations of values of the Melbourne-
Wübbena combination from the mean value after data cleaning from outliers using the developed algorithm
(see Section 3).

Epochs in which the data were rejected are designated by white circles. In the first
case (see Figure 3a), 47 data of the measurements were rejected, which are 26.1% of
the total amount of data. In the second case (see Figure 3b), 14 of these measure-
ments were rejected (7.8%), which are almost 18% less than in the previous
calculation.
We also provide similar results for data obtained by TLSE station, which is also
included in the IGS network. We consider measurement data from GLONASS,
Russia satellite with system number PRN = 1 for 2010, day 207. Figure 4 shows the
values of the Melbourne-Wübbena combination over a time interval of 65.5 min
(N = 132). Figure 5 plots the values of deviations from the mean value of the data
cleared of outliers using the algorithms described in Sections 2 and 3, respectively.
Parameters σmax and MINOBS are the same as in the previous calculation example.
In the first case (see Figure 5a), 41 data of the measurements were discarded, which
are 31% of the total amount data. In the second case (see Figure 5b), 8 of these
measurements were rejected (6%), which are 25% less than in the previous
calculation.

10
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

Figure 4.
Melbourne-Wübbena combination for TLSE station (GLONASS satellite, PRN = 1 for 2010, day 207).

Figure 5.
(a) Deviations of values of the Melbourne-Wübbena combination from the mean value after data cleaning
from outliers using the algorithm described in Section 2. (b) Deviations of values of the Melbourne-Wübbena
combination from the mean value after data cleaning from outliers using the developed algorithm (see
Section 3).

11
Satellite Systems - Design, Modeling, Simulation and Analysis

5. Mathematical prerequisites for modifying of existing algorithm

Note that the number of arithmetic operations required to find the optimal
solution according to the algorithm described in Section 3 depends on the Lmax ,
which is the length of the solution. As can be seen from Eq. (24), the smaller the
length of the found solution (i.e., the larger the number of detected outliers), the
more arithmetic operations are required to find it. This number of arithmetic
operations is estimated to be of order N þ ðN Lmax Þ2 . Note that the expression in
the parentheses herein is equal to the number of outliers detected. Thus, if the
number of outliers in the original data series is comparable to N, it will take N2
arithmetic operations to find the optimal solution. In particular, to make certain
that there is no solution (e.g., in the case where the data contain a trend), N2
arithmetic operations will also be required. In Section 6, we will modify the existing
algorithm and describe fast outlier search algorithm that requires Nlog2 N
arithmetic operations.
The necessary preparations are given in this section. Note that in this and the
n oN
next sections we are dealing with the sequence yj arranged in the ascending
j¼1
order.
n oN
Assertion 2. Let yj be monotonically increasing sequence. The following
j¼1
inequality is true:

σ2 ðk; L þ 1Þ ≥ min σ2 ðk; LÞ, σ2 ðk þ 1; LÞ : (25)

Proof. From the monotonicity of the sequence yj and the definition zðk, L þ 1Þ
[see Eq. (13)],

yk ≤ zðk, L þ 1Þ ≤ ykþL : (26)

One of two cases is possible:

a. 2zðk, L þ 1Þ ≤ ykþL þ yk , ) zðk, L þ 1Þ yk ≤ ykþL zðk, L þ 1Þ,

b. 2zðk, L þ 1Þ > ykþL þ yk , ) zðk, L þ 1Þ yk > ykþL zðk, L þ 1Þ.

Suppose, for example, the case (a) holds. Let us show that in this case

σ2 ðk; L þ 1Þ ≥ σ2 ðk; LÞ: (27)

At first, we will show that:

∣yj zðk, L þ 1Þ∣ ≤ ykþL zðk, L þ 1Þ; j ¼ k, … , k þ L: (28)

Truly, inequalities:

yk ≤ yj ≤ ykþL ; j ¼ … , k þ L,

and the above inequality derived in Case (a) implies:

zðk, L þ 1Þ yj ≤ zðk, L þ 1Þ yk ≤ ykþL zðk, L þ 1Þ, (29)

12
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

yj zðk, L þ 1Þ ≤ ykþL zðk, L þ 1Þ: (30)

These inequalities, in turn, imply Eq. (28). Next let us prove (27). This inequal-
ity is expanded as follows:

kþL  kþL
X1 
1X 2 1 2
y zðk; L þ 1Þ ≥ yj zðk; LÞ :
L j¼k j L 1 j¼k

Substituting here of the expression (17) in place of zðk; LÞ and writing for
brevity, z instead of zðk; L þ 1Þ, we get inequality:

kþL  kþL
X1 
X 2 z ykþL 2
ðL 1Þ yj z ≥L yj z : (31)
j¼k j¼k
L

Transform the right-hand side of this inequality:

kþL 
X ykþL z2 ðL þ 1Þ2 2
RHSð31Þ ¼ L yj zþ ykþL z
j¼k
L L
kþL
X 2 ðL þ 1Þ
2 ðL þ 1Þ2 2
¼L yj z þ ykþL z ykþL z :
j¼k
L L

PkþL  
Here we take into account the equality: j¼k yj z ¼ 0. Next, we have:

kþL 
X 2
2
RHSð31Þ ¼ L yj z ðL þ 1Þ ykþL z :
j¼k

Substituting this expression in Eq. (31), we get inequality

kþL 
X 2
2
yj z ≤ ðL þ 1Þ ykþL z ,
j¼k

that is true due to Eq. (28). Thus, Eq. (27) is proved for case (a). Analogously,
case (b) is considered.
We introduce the notation:

σ2min ðLÞ ¼ min σ2 ðk, LÞ: (32)

1 ≤ k ≤ N Lþ1

Assertion 3. The following inequalities hold:

σ2min ðNÞ ≥ σ2min ðN 1Þ ≥ … ≥ σ2min ðMINOBSÞ: (33)

That is, the sequence σ2min ðLÞ monotonically decreases when L decreases from N to
MINOBS.
Proof. Assertion 2 and definition of σ2min ðLÞ expressed in Eq. (32) imply:

σ2 ðk, L þ 1Þ ≥ min σ2 ðk, LÞ, σ2 ðk þ 1, LÞ ≥ σ2min ðLÞ:

13
Satellite Systems - Design, Modeling, Simulation and Analysis

Since k is chosen arbitrarily, then for all L = MINOBS, … , N 1 the following

inequalities hold:

σ2min ðL þ 1Þ ≥ σ2min ðLÞ,

which proves Assertion 3.

Assertion 3 implies the following corollary.
Corollary 1. If the inequality
σ2min ðL0 Þ > σ2max : (34)

holds for some L0 , then for existence of the solution YL ¼ yk , ykþ1 , … , ykþL 1 for
the problem expressed in Eqs. (3)–(7), it is necessary that the length L of the set Y L
satisfies the condition L < L0 .
Proof. Let us assume that L ≥ L0. Assertion 3 on account of monotony of σ2min ð
Þ,
expressed in Eq. (33) implies the following inequalities σ2min ðLÞ ≥ σ2min ðL0 Þ > σ2max for
all L≥L0 . From this, it follows that for any set YL ¼ yk , ykþ1 , … , ykþL 1 we will
have σ2 ðk, LÞ ≥ σ2min ðLÞ > σ2max . Thus, any of sets YL of length L ≥L0 does not satisfy
the condition in Eq. (15) and, therefore, cannot be a solution of the problem,
expressed in Eqs. (3)–(7).
In particular, we have come to the next important result. If, for example, the
inequalities σ2min ðMINOBSÞ > σ2max are fulfilled, then the solution for the problem
described in Eqs. (3)–(7) does not exist.
In the above-described procedure for solving problem (3)–(7), it takes N2
arithmetic operations to make certain that the solution not exists. Taking into
account Assertion 3 and Corollary 1, the search procedure may begin by checking
the conditions.

σ2min ðNÞ ¼ σ2 ð1, NÞ ≤ σ2max and σ2min ðMINOBSÞ ≤ σ2max :

This will require approximately N arithmetic operations. If none of these

conditions are fulfilled, the solution search must stop because the solution does not
exist. As a result, only N arithmetic operations are required to ensure that there is
no solution.

6. Fast outlier search algorithm

The above proposed search procedure consists in the calculating values of zðk, LÞ
and σ2 ðk, LÞ using recurrent formulas (17)–(22) and checking at every k and L the
fulfillment of the inequalities (11) and (12). The algorithm complexity is estimated
by value of  N þ N2Outlier , where NOutlier is the number of outliers found. If it is
known a priori that there are few outliers in the measurement data, then the search
algorithm for the optimal solution that described in Section 3 can be applied. If the
measurement data contain an N-comparable number of outliers, the complexity of
such an algorithm will be estimated by about N2. It is namely for such type of data we
below describe a modified outlier search algorithm with complexity of about Nlog2 N.
First of all, note one property that is the key to the construction of a fast outlier
search algorithm. Note that if the inequality (15) holds for some set of length L + 1,
then there exists a set of length L for which the inequality (15) is valid too. Truly, let
assume for some k the inequality σ2 ðk, L þ 1Þ ≤ σ2max holds. This inequality and
Eq. (25) imply

14
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

min σ2 ðk; LÞ, σ2 ðk þ 1; LÞ ≤ σ2max :

From this, it follows that

σ2 ðk; LÞ ≤ σ2max and=or σ2 ðk þ 1; LÞ ≤ σ2max : (35)

This means that at least one of these sets yk , … , ykþL 1 and ykþ1 , … , ykþL
with length of L satisfies conditions expressed in Eq. (15).
However, this property is not true when checking the conditions expressed in
Eq. (16). In other words, if these conditions are fulfilled for any set of length L + 1,
it might happen that none of the sets of length L may satisfy them. This fact is a
significant obstacle to increasing the rate of outlier detection that is necessary
when processing a large amount of data with a large number of rough measure-
ments. To overcome this obstacle, we will make the condition expressed in
Eq. (16) weaker.
First of all, note that if for some set yk , … , ykþL 1 , the both conditions
expressed in Eq. (16) are fulfilled, then the following inequality will hold:

ykþL 1 yk ≤ 6σmax : (36)

Consider a problem with condition expressed in Eq. (36) instead of conditions

expressed in Eq. (16).
Remark. Recall that in this condition L means the length of the set under checking,
and k is the index of the smallest number included in the set. Although “k” and “L” are
also encountered as indexes in the sets we use hereinafter, we hope nevertheless that this
will not lead to confusion.
It is easily seen that condition expressed in Eq. (36) for an arbitrary set
yk , … , ykþL of length L + 1 implies the fulfillment this condition for each of the
sets yk , … , ykþL 1 and ykþ1 , … , ykþL of length L. In fact, the fulfillment condi-
tion (36) for the set yk , … , ykþL means the fulfillment of inequality ykþL
yk ≤ 6σmax from which due to monotony of yj , the inequalities imply ykþL
ykþ1 ≤ 6σmax and ykþL 1 yk ≤ 6σmax .
Thus, we have established the validity of the following assertion.
Assertion 4. If the set yk , … , ykþL of length L + 1 satisfies conditions (15) and (36),
then at least one of the two sets yk , … , ykþL 1 or ykþ1 , … , ykþL of length L also
satisfies conditions (15) and (36).
Based on this statement, we can formulate the following:
Assertion 5. Solution for the problem (15) + (36) can be found for N log 2 N
arithmetic operations.
Proof. Let us consider the sequence
h of steps.
i
ð 0Þ ð0Þ ð0Þ
Step 0: Consider the segment N Left , N Right of numerical axis, where N Left =
ð0Þ
MINOBS, N Right = N. In this step, there is one set y1 , … , yN of length N. We check
it for fulfillment of the conditions expressed in Eqs. (15) and (36) with k = 1, L = N.
If they are satisfied, this set is the solution, and our search stops. Otherwise, we pass
to considering of ðN MINOBS 1Þ sets of length MINOBS. We check each of
these sets for fulfillment of conditions expressed in Eqs. (15) and (36). If none of
them satisfy these conditions, then we stop the search process and conclude that the
solution does not exist. Otherwise, once we find the set of length MINOBS satisfy-
ing conditions (15) and (36), we transit to Step 1.

15
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 6.
Three possible cases for the proposed search. In case (a) we go to the right-hand side range (range for length of
sets) to find a solution, in case (c) we go to the left-hand side range to find a solution, in case (b) we look for a
ðk 1Þ
solution with length L = N Mid and the search ends.

Step 1: Step 1 is the same as Step k described below for k = 1

… h i
ðk 1Þ ðk 1Þ
Step k: On the kth step, where (k ≥ 1), we consider a segment N Left , N Right ,
h i
ðk 1Þ ðk 1Þ
which we halve, as a result we receive two segments N Left , N Mid and
h i h  i
ðk 1Þ ðk 1Þ ðk 1Þ ðk 1Þ ðk 1Þ ðk 1Þ
N Mid þ 1, N Right , where N Mid = N Left + N Right N Left =2 , [
] designate
ðk 1Þ ðk 1Þ
integral part of a number. Next, we check the sets of length N Mid and N Mid + 1 for
fulfillment of conditions (15) and (36). The following three cases are possible,
schematically shown in Figure 6. The sign “ ” above the point means that for none
of the sets of the corresponding length the conditions (15) + (36) are satisfied, the
sign “+” vice versa, that is, there is at least one set of the corresponding length
ðkÞ ðk 1Þ ðkÞ ðk 1Þ
satisfying (15) + (36). In the case of (a), we set N Left = N Mid + 1, N Right = N Right and
ðkÞ ðk 1Þ ðkÞ ðk 1Þ
transit to the (k + 1)-th step; in the case of (c), we set N Left = N Left , N Right = N Mid
and transit to the (k + 1)-th step; in the case of (b), we search the solution
ðk 1Þ
(15) + (36) with minimal value of σ2 ðk, LÞ with L = N Mid ; the algorithm is
terminated.
The search
h processi will continue until either case (b) or until the length of the
ðkÞ ðkÞ
segment N Left , N Right is less than or equal to 1. In either case, the total number of
steps will not exceed the number of log 2 ðN MINOBSÞ. Since N operations are
performed in each step, the search process is guaranteed to be terminated in
N log 2 N arithmetic operations.

7. Search an unknown trend in the power polynomial class

The need to process GNSS measurements including a trend on which noise and
outliers are superimposed arises at different processing stages of the application
process. As already stated above, satellite clock corrections contain a linear trend. In
some cases, it may not be known, and then, one has to search for it, for example, by
the least square method. The presence of outliers in the measurement data is a
significant obstacle to accurate determination of drift and offset parameters of
satellite clocks. Other examples are linear combinations of code and phase data on
two carriers [3]. To obtain high accuracy results, it is necessary to detect outliers
against an unknown trend and remove them from further processing. This is the
subject of this section.

16
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

7.1 Statement of the problem

Consider the problem of outlier detecting in data presented in the form of

Eq. (1), recall that:

yj ¼ f j þ z þ ξj ; j ¼ 1 … N:

The procedure described above for finding the optimal solution in an ordered
series of numbers may not produce an adequate result if applied to data containing
an unknown trend. For example, there may be no solution, and all data will be
defined as outliers. In order to detect outliers in a series of numbers with a trend
using the algorithm described above, it is necessary to find a suitable approximation
of an unknown function f j and subtract it from the data set. Searching for this
approximation is usually done by selecting functions from some functional class.
The choice of the functional class depends on the task. In some cases, these may be
power polynomial, in other cases, trigonometric polynomial, etc. The presence of
outliers in the data measurements makes it much more difficult to find such an
approximation. In this section, we will describe the general approach to solving the
problem and look for suitable approximations of trend f j in the class of power
polynomial with real coefficients:
 
!
Pn,j a ¼ an ðj=NÞn þ an 1 ðj=NÞn 1
þ … þ a0 , (37)

!
where n is the polynomial degree, and a ¼ fa0 , … , an g is vector of coefficients.
Thus, the problem consists in the creation of an algorithm for searching the
trend in the class of power polynomial and detecting outliers in specified data series
yj represented in Eq. (1).

7.2 Minimizing set of specified length L

Before we turn to the trend search algorithm construction, we will define the so-
called minimizing set of given length L, which plays an essential role in the trend
search. In addition, in Section 7.3, we will describe a search algorithm for such set
based on the nrecurrent formulas
o (17)–(19) and (20)–(22).
Let YL ¼ yj , … , yj be an arbitrary set of length L, composed of the values of
1 L
n oN
a numerical series yj , the monotony is not supposed. The mean and the SD
j¼1
values for it are denoted by zY L and σYL . These values are calculated by standard
formulas:
X
zYL ¼ L 1 yj , (38)
j ∈ fj1 , … , jL g

X  2
σ2YL ¼ ðL 1Þ 1 yj zYL : (39)
j ∈ fj1 , ::, jL g

n oN
Definition 2. Given L for a specified sequence of values yj the set of values:
j¼1

n o
YL, min ¼ yj , … , yj ,
1 L

17
Satellite Systems - Design, Modeling, Simulation and Analysis

at which the minimum value of σ2YL defined in Eq. (39) is reached will be called the
minimizing set of length L. The corresponding mean and SD values are denoted by zYL, min
and σYL, min .
According to this definition, we have
X
zYL, min ¼ L 1 yj , (40)
j ∈ fj1 , … , jL g

X  2 n o
2 1
σYL, min ¼ ðL 1Þ yj zYL, min ¼ min σ2YL : (41)
YL
j ∈ fj1 , … , jL g

Minimum in Eq. (41) is searched by all kinds of sets of length L composed of

n oN
numbers of series yj .
j¼1
Note that the numbers yj are not supposed to be in the ascending order.
Next, we will formulate and prove a statement similar to Assertion 1, which will
allow us, when searching for a minimizing set, to proceed from the original series to
its ordered permutation. n o
Assertion 6. Let YL, min ¼ yj , … , yj be a minimizing set of length L for a
1 L
n oN
given sequence of values yj and
j¼1
n o n o
ymin ¼ min yj , … , yj , ymax ¼ max yj , … , yj ,
1 L 1 L

then the interval ymin , ymax does not contain values yj that are not in the set YL, min .
Proof. Let us assume the opposite: Let yj ∉ YL, min , ymin < yj < ymax . Let
yk , … , ykþL 1 is a permutation of the numbers yj , … , yj in the ascending order;
1 L
then yk ¼ ymin and ykþL 1 ¼ ymax . One of these cases is possible:

a. yj < zYL, min and therefore subject to inequality yk < yj , we have

   2
2
zYL, min yk > zYL, min yj ) zYL, min yk > zYL, min yj
 2
2
) zYL, min yj zYL, min yk <0 (42)

b. yj ≥ zYL, min and therefore subject to inequality ykþL 1 > yj , we have

   2
2
ykþL 1 zYL, min > yj zYL, min ) yj zYL, min ykþL 1 zYL, min <0
(43)

In the first case, Case (a), we replace the value yk in the set yk , … , ykþL 1 with
yj . In the second case, Case (b), we replace the value ykþL 1 with yj . We want to
show that doing such replacement, the value σ2YL, min expressed in Eqs. (40) and (41)
will decrease. This will mean that YL, min is not a minimizing set.
Suppose Case (a). For brevity, we will write below z instead of zYL, min and σ2
~2 , the similar values obtained after replacement yk
instead of σ2YL, min . Denote ~z and σ
with yj . We have:
1
z¼L yk þ … þ ykþL 1 , (44)

18
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658
 
1 2 2
σ 2 ¼ ðL 1Þ yk z þ … þ ykþL 1 z , (45)

and
 
1
~z ¼ L ykþ1 þ … þ ykþL 1 þ yj , (46)
 2 
2 1 2 2
~ ¼ ðL
σ 1Þ ykþ1 ~z þ … þ ykþL 1 ~z þ yj ~z : (47)

We want to show that

~2 < σ2 :
σ (48)

Eqs. (44) and (46) imply:

 
1
~z ¼ z þ L yj yk : (49)

~2 expressed in Eq. (47) taking account of Eq. (49):

Modify σ
  2
1
~ 2 ¼ ðL
σ 1Þ ykþ1 z L 1
yj yk þ…
  2   2 
1 1
þ ykþL 1 z L yj yk þ yj z L yj yk :

After simplification with taking into account Eq. (45), we get from here:
 2  2 
2 2 1 2 1
~ ¼ σ þ ðL
σ 1Þ yj z yk z L yj yk :

From (42), it follows (recall that we write z instead of zYL, min ) that the expression
in the square brackets is strictly less than zero. Thus, inequality (48) is proven.
From this follows that the set YL, min is not minimizing one because the condition
(41) is not met. Thus, we have arrived at a contradiction that proves the validity of
the formulated Assertion 6. Case (b) is considered similarly.

7.3 YL, min search algorithm

Assertion 6 is much like Assertion 1, which made it possible to go from an

arbitrary numerical series to an ordered one for the optimal solution search (see
Section 3). Similarly, Assertion 6 makes it possible to go to an ordered number
series to find the minimizing set of numbers of a given length. In our case, consid-
erations similar to those presented innSection 3 may
o be made when searching for a
minimizing set. Namely, if YL, min ¼ yj , … , yj is the minimizing set of length L,
1 L

then an arbitrary permutation of the numbers yj , … , yj will also be the minimizing

1 L
set. Thus, the process for searching of YL, min does not depend on the arrangement
of given numbers yj . This allows us to arrange them in the order most suitable for
n o
searching the minimizing set. We arrange the values yj in the ascending order,
and we will look for the minimizing set in the ordered n
sequence.
o As in Section 3, for
brevity, the ordered sequence will also be denoted by yj . For simplicity of logic,

19
Satellite Systems - Design, Modeling, Simulation and Analysis

we will assume that all yj are different, that is, Eq. (10) holds. Rewrite it for
convenience:
y1 < y2 < … < yN

Note that due to Assertion 6, if YL, min is the minimizing set and yj and yj are its
1 L
smallest and greatest values, respectively, then all values of yj from the interval
 
yj , yj belong to YL, min . Consequently, considering Eq. (10), we have yj ¼
1 L L

yj þL 1 and therefore
1
n o
YL, min ¼ yj , yj þ1 , … , yj þL 1 :
1 1 1

Thus, in order to find a minimizing set of length L, it is sufficient for us to check

N L + 1 sets
y1 , … , yL , y2 , … , yLþ1 , … , yN Lþ1 , … , yN :

and choose from them the one that has minimal SD value. In the minimizing set
searching procedure, we use the appropriate designations zðk, LÞ and σ2 ðk, LÞ
expressed by Eqs. (13) and (14) to denote mean and square SD in the case of sets
composed of the ascending ordered values.
The calculation of zðk, LÞ and σ2 ðk, LÞ when searching YL, min can be carried out
in the manner similar to that of the optimal solution according to the schematic, in
which only the σ2 ðk, LÞ is shown:

Figure 7.
Scheme of calculations when finding the minimizing set of length L.

At first, we carry out the transitions in the direction of the vertical arrows
(see diagram in Figure 7) and calculate sequential values
σ2 ð1, NÞ, σ2 ð1, N 1Þ, … , σ2 ð1, LÞ using formula (17)–(19). Finally, we go
in the direction of the horizontal arrows and calculate values of
σ2 ð1, LÞ, σ2 ð2, LÞ, … , σ2 ðN L þ 1, LÞ using formula (20)–(22) and choose from
them the minimal one.

7.4 Trend search algorithm

In order to correctly detect outliers in measurement data that include an

unknown trend, it is necessary to find and remove trend from the original data. The
problem in determining of unknown trend is to find a suitable profile for the
measurement data by adjusting the fitting parameters. This implies, in nturn, the o
needs to select from the specified series y j the “right” reference values yj , … , yj
1 L

used for fitting. If the data do not contain coarse measurements, all N values of the
original data can be treated as reference ones and used for fitting. If the data contain

20
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

rough measurements, the participation all or some of those values in the fit will
result in a fatal distortion of the trend function and, as a result, an incorrect
determination of the outliers. Thus, when determining the trend, it is necessary to
exclude rough measurements from the number of reference values used for fitting
and by which the trend approximations are built. We have the vicious circle. To
determine outliers, it is necessary to find a proper trend approximation, and to find
an exact trend approximation, we need to find all outliers and remove them from
the values used for fitting. Another vicious circle is obtained when trying to choose
the number of values for fitting. If we take it too small to minimize the possibility
for outliers to be included into the subset of values for fitting, the data fit may not
be accurate enough for the rest values of the data outside of the subset. If, on the
contrary, we take rather large number of points for fitting, the outliers may be
among them, and we will also get a bad approximation of the trend.
We describe herein a strategy for finding an unknown trend and detecting outliers.
This strategy assumes that the number of outliers in the series presented by Eq. (1)
does not exceed a certain value Nmaxout known a priori. Thus, we can suppose that the
number of “right” values suitable for fitting in the series y j is not less than N Nmaxout.
Below L is supposed to be fixed and associated with the number of the reference
values of the series (1) used for fitting, and L ≤ N Nmaxout .
Let us consider the following algorithm. It contains internal iterations, which we
will denote with the upper index “s” in parentheses.
Step 0: n = 0. We set some L, satisfying the condition L ≤ N Nmaxout .
ð 0Þ
Step 1: n++; s = 0; flagj ¼ 1.
!ðsÞ
Step 2: s++. We fit polynomial to the data set and find fitting coefficients a ¼
n o
ðsÞ ð sÞ
a0 , … , an :
 
!ðsÞ 1Þ !
a ¼ arg min Φðs a , (50)
a0 , … , an 1 , an

  XN   2
1Þ ! ! ðs 1Þ
Φðs a ≜ yj Pn,j a
flagj : (51)
j¼1

ðsÞ
^j obtained after subtraction from yj the polynomial
Step 3: Consider the values y
values with coefficients found in Step 2:
 ð sÞ 
ðsÞ !
y
^j ¼ yj Pn,j a : (52)
n oN
ðsÞ
Using the algorithm described in Section 7.3, we find from the numbers y ^j
j¼1
 
ðsÞ ðsÞ ðsÞ
defined in Eq. (52) a minimizing set YL, min ¼ y ^ ðsÞ , … , y
^ ðsÞ of length L. For this
j1 jL
set, we calculate the corresponding values of the mean zYðsÞ and SD σYðsÞ
L, min L, min

according to the formulas (40) and (41) in which j1 , … , jL is replaced by

n o
ðsÞ ðsÞ
j1 , … , jL . We redefine the reference points for the next fit process (Step 2) by
ðsÞ
marking them with flagj :
( n o
ð sÞ ðsÞ
ðsÞ 1, if j ∈ j 1 , … , jL
flagj ¼ : (53)
0, otherwise

21
Satellite Systems - Design, Modeling, Simulation and Analysis

We transit to Step 2 and do so until the convergence of the σYðsÞ , s = 1,2, … , is

L, min

achieved. Note that we will consider the issue of convergence further in Section 7.5.
After reaching convergence of values σYðsÞ , we transit to Step 4 for outlier detection.
L, min
 ðsÞ 
ðsÞ !
Step 4: Searching the optimal solution for data set y
^j ¼ yj Pn,j a , j = 1, … , N.
ðsÞ
We find the optimal solution for values y
^j using the algorithm of Section 3 and
determine the number Nout of outliers detected. If it turns out that Nout ≤Nmaxout ,
 ðsÞ 
!
then solution is considered found; searching process stops: f j ¼ Pn,j a . Other-
wise, if Nout >Nmaxout , then we transit to Step 1.
We do this until we find a solution or until n reaches some preset value Nmax
(e.g., 10). In this case, probably, we may need to select a different functional class
to search for a trend.
Example. Let us consider the data simulated in accordance with formula:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
y j = 10 j þ 10 þ 2random j , where j ¼ 1 … N, random j —pseudorandom numbers,
evenly distributed in the segment [0,1]. Let us set N = 150 and model 10 outliers at
points j ¼ 6:::10 and j ¼ 140:::144 (see Figure 8). Further, let us set σmax = 0.6,
MINOBS = 10 and search for outliers as described above with L = 130. If n = 1, we
get N out = 88; if n = 2, we get N out = 33; if n = 3, we get N out = 17; if n = 4, we get N out
= 10. Thus, a suitable trend approximation turned out to be a fourth degree poly-
nomial. Figure 9 shows values for fourth degree polynomial fitted to the data set on

Figure 8. pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Data simulated using: y j = 10 j þ 10 þ 2random j and fourth degree polynomial after the first iteration.

Figure 9.
Simulated data and fourth degree polynomial after the eighth iteration.

22
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

Figure 10.  
!
The differences ^y j ¼ y j P4,j a , approximation of unknown trend with fourth degree polynomial.

the eighth iteration after the convergence

 discussed above is reached. In Figure 10,
!
the differences ^y j ¼ y j P4,j a are plotted. Positions of detected outliers are
marked with white circles. Note that the trend was modeled by a function that does
not belong to the class of power polynomials.

7.5 Convergence of iterations in trend search algorithm

This section explains the convergence of the iterations described in the trend
search algorithm (see previous section).
Assertion 7. The SD sequence σYðsÞ of values calculated in the trend search algo-
L, min

rithm monotonically decreases at s = 1, 2, … :

σYðsÞ ≥ σYðsþ1Þ ≥ … (54)

L, min L, min

and therefore converged.

Proof. We start our consideration with Step 3 and sth iteration, s = 1, 2, … In
n oN
ð sÞ ðsÞ
Step 3, for the sequence y ^j , we find a minimizing set of length L: YL, min ¼
j¼1
 
ðsÞ ðsÞ
^ ðs Þ , … , y
y ^ ðsÞ . Write the expressions for the mean and the square of SD
j1 jL
corresponding to this set. We have due to Eqs. (40) and (41):
X ðsÞ
1
zYðsÞ ¼L y
^j , (55)
L, min
ðs Þ ðs Þ
j∈f j1 , … , jL g
X 2
1 ðsÞ
σ2YðsÞ ¼ ðL 1Þ y
^j zYðsÞ : (56)
L, min L, min
ðsÞ ðs Þ
j∈f j1 , … , jL g

Transform the expression on the right side of Eq. (56). Substitution here instead
ðsÞ
of y
^j expression in Eq. (52) gives:

X   2
1 !ðsÞ
σ2YðsÞ ¼ ðL 1Þ yj Pn,j a zYðsÞ :
L, min L, min
ðs Þ ðsÞ
j∈f j1 , … , jL g

23
Satellite Systems - Design, Modeling, Simulation and Analysis

ðsÞ
Using the flagj defined in Eq. (53), rewrite the last equality as follows:
XN  ðsÞ  2
2 1 ! ðsÞ
σYðsÞ ¼ ðL 1Þ yj Pn,j a zYðsÞ
flagj ¼
L, min L, min
j¼1
XN  !2 !
1 ðsÞ 1 ðsÞ
¼ ðL 1Þ yj Pn,j b
flagj ¼ ðL 1Þ Φ b : (57)
j¼1

 
! ðsÞ ðsÞ ðsÞ
Here we introduce the designation b ¼ þ zYðsÞ a0
and take into , a1 , … , an
L, min
 
!
account the definition of Pn,j a expressed by Eq. (37) and the definition of
 
!
functional ΦðsÞ a given in Eq. (51). From Eq. (57), we get:
!
σ2YðsÞ ¼ ðL 1Þ 1 ΦðsÞ b (58)
L, min

!ðsþ1Þ
We transit to Step 2; s is incremented by 1. Wefind
 a vector a of polynomial
ðsÞ !
coefficients, which minimizes the functional Φ a :
 
!ðsþ1Þ ðsÞ !
a ¼ arg min Φ a :
!
a

Thus,
 ðsþ1Þ  n  o
! ðsÞ !
ΦðsÞ a ¼ min
!
Φ a (59)
a

From the definition of the extremum of functional, it follows:

  !
ðsÞ !ðsþ1Þ ðsÞ
Φ a ≤Φ b : (60)

Taking into account Eq. (58), we have:

 ðsþ1Þ 
!
σ2YðsÞ ≥ ðL 1Þ 1 ΦðsÞ a : (61)
L, min

 
ðsÞ !
The extremum condition (one of n + 1) of functional Φ a is as follows:

∂ ðsÞ !
Φ a  ¼ 0:
∂a0 ! !ðsþ1Þ
a¼a

From here, we derive:

N 
X  ðsþ1Þ 
! ðsÞ
yj Pn,j a
flagj ¼ 0:
j¼1

ðsþ1Þ
Taking into account the designation for y
^j given by Eq. (52), we can write
this equality in the form.

N
X X
ðsþ1Þ ðsÞ ðsþ1Þ
y
^j
flagj ¼ 0 or y
^j ¼ 0: (62)
j¼1 j∈f
ðsÞ
j1 ,
ðs Þ
… , jL g

24
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658
 
ðsþ1Þ ðsþ1Þ ðsþ1Þ
Consider the set YL ¼ y
^ ðsÞ , … , y
^ ðsÞ . The mean and SD values for it are
j1 jL
calculated using formulas (38) and (39). Taking into account Eq. (62), we get:
X ðsþ1Þ
1
zYðsþ1Þ ¼ L y
^j ¼ 0,
L
ðsÞ ðs Þ
j ∈ fj 1 , … , j L g

and
X  2 X  2
1 ðsþ1Þ 1 ðsþ1Þ
σ2Yðsþ1Þ ¼ ðL 1Þ y
^j zYðsþ1Þ ¼ ðL 1Þ y^j ¼
L L
ðsÞ ðs Þ ðs Þ ðs Þ
j ∈ fj1 , … , jL g j ∈ fj1 , … , jL g
N 
X  2  
1 !ðsþ1Þ ðsÞ 1 ðsÞ !ðsþ1Þ
¼ ðL 1Þ yj Pn,j a
flagj ¼ ðL 1Þ Φ a :
j¼1

Thus,
 ðsþ1Þ 
!
σ2Yðsþ1Þ ¼ ðL 1Þ 1 ΦðsÞ a : (63)
L

n oN
ðsþ1Þ
We transit to Step 3. In this step, for sequence y ^j , we find a minimizing
j¼1
 
ðsþ1Þ ðsþ1Þ ðsþ1Þ
set of length L: YL, min ¼ y ^ ðsþ1Þ . The SD square σ2 ðsþ1Þ for this set does
^ ðsþ1Þ , … , y
j1 jL YL, min

ðsþ1Þ
not exceed the same magnitude for any other set, particularly for YL ¼
 
ðsþ1Þ ðsþ1Þ
y
^ ðs Þ , … , y
^ ðs Þ :
j1 jL

σ2Yðsþ1Þ ≥ σ2Yðsþ1Þ : (64)

L L, min

Finally, from Eqs. (61) and (63) and the last inequality (64), we get
 ðsþ1Þ 
!
σ2YðsÞ ≥ ðL 1Þ 1 ΦðsÞ a ¼ σ2Yðsþ1Þ ≥ σ2Yðsþ1Þ
L, min L L, min

that is, σ2 ðsÞ ≥ σ2 ðsþ1Þ , that proves Assertion 7.

YL, min YL, min

8. Detection and repair cycle slips in the Melbourne-Wübbena

combination algorithm

In this and the following section, we describe cycle slip repair algorithm for
observers represented in the form of Melbourne-Wübbena combination, which is
often used in modern GNSS measurement data processing programs. Loss by the
receiver of the carrier phase capture results in jumps in the code and phase mea-
surement data. In the absence of jumps, as we already discussed in Section 2, the
values of the combination consist of measurement noise superimposed on an
unknown constant value dependent on a specified satellite-receiver pair.
In case of temporary loss of carrier phase capture by the receiver, jumps occur in
a series of values representing the Melbourne-Wübbena combination. The proce-
dure of detecting jumps and eliminating them from the values of the combination,
25
Satellite Systems - Design, Modeling, Simulation and Analysis

called cycle slip repair, is one of the most important steps of preprocessing GNSS
data. The main difficulty in detecting jumps is that neither the exact size of jumps
nor their epochs are known. A number of algorithms, descriptions of which can be
found, for example, in Refs. [3, 6, 8] are proposed for the detection of jumps.
Although differing in detail, they are based on a common idea, that is, comparison
of the SDs of the time series of measurement data obtained from one of the bounds
of the time interval to an arbitrary moment, an epoch. If the differences of the SDs
corresponding to two adjacent epochs exceed a predetermined threshold value, then
a jump is declared in one of these two epochs. A drawback of similar algorithms is
the frequent false detection of jumps during epochs containing rough measure-
ments (outliers) since the values of outliers can exceed the sizes of a jump itself. On
the other hand, an attempt to increase the threshold value leads to the opposite
effect, an inability to recognize jumps that are small in magnitude.
Below, we propose a robust cycle slip repair algorithm that allows, more reliably
than similar known algorithms, to detect jumps and determine their sizes. The
proposed algorithm is based on search for so-called clusters consisting of epochs, in
which the values of the combination are grouped about corresponding predefined
values. Besides, this algorithm implements the above-described method of cleaning
data from outliers based on the search for the optimal solution. This method,
combined with Springer’s algorithm used in Ref. [3], allows for the reliable deter-
mination of multiple (cascade) jumps in the Melbourne-Wübbena combination.
The Melbourne-Wübbena combination LM W can be presented in the form of
the total of three terms [6]:

LM W ¼ λ5 n5 þ B þ ν, (65)

where λ5 is the formal wavelength (for all GPS satellites λ5 = 86.16 cm and for
GLONASS satellites λ5 = 84.0 ÷ 84.36 cm); n5 is an unknown integer, the so-called
wide lane ambiguity [3]; B is the residual systematic error caused by instrumental
delays relative to the specific receiver-satellite pair and, as assumed, not time-
dependent; and ν is a random component. Both parts of the equality in (65) are
expressed in meters. At the same time, the LM W combination is often expressed in
cycles of wavelength λ5 . Let us divide both parts of Eq. (65) by λ5 . We introduce the
designations y ¼ LM W =λ5 , β ¼ B=λ5 , and ξ ¼ ν=λ5 and derive for each of the epochs
associated with indexes j ¼ 1, … , N:

y j ¼ n5j þ β þ ξ j : (66)

where n5j is an unknown piecewise-constant function of a discrete argument,

regarding which it is only known that it takes integer values and can undergo
integer jumps; β is an unknown constant value; and ξ j is a random variable. The
problem consists in the development of an algorithm for automatic processing of
the informational data y j of form of Eq. (66), making it possible to effectively
determine integer-valued jumps against the noise component and outliers.

9. Description of the proposed algorithm

Let us present the proposed algorithm as the following sequence of steps.

Step 0: We introduce parameter Δ by specifying its value as Δ ¼ 2
σ max . We
mark the indexes of epochs with function flagj as nulls: flagj ¼ 0; j = 1, … , N.
Arrange the array y j in the ascending order and renumber it by placing the
indexes in the ascending order. Denote the resulting array as Y j . For simplicity of
26
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

Figure 11.
Searching for the maximum of the density of values of the array Y j .

logic, we suppose that all Y j are different. Therefore, we have: Y 1 < Y 2 < … < Y N .
Denote the corresponding permutation of the values of function flagj as FLAGj .
Step 1: On this step, we are looking for the maximum density of array values y j .
 
We consider the segments Y j , Y j þ Δ of length Δ and look for the one that
contains the largest number of Y i values. Here, only those indexes i and j from the
range of 1 ≤ i, j ≤ N are considered for which FLAGi ¼ FLAG j = 0. The purpose of
the ordering of the original array consists in improving the effectiveness of the
maximum density search procedure. It can be shown that the number of compari-
sons required when searching for the above segment does not exceed 2 N in the case
of the ordered array. For an unordered array, the number of comparisons is evalu-
ated as N 2 .  
Step 2: Let Y J max , Y J max þ Δ be the segment found in the previous step, containing
N max values of Y j (see Figure 11). We calculate the mean m of these numbers Y j .

Xmax 1
J max þN
1
m¼ N max
Y j, (67)
j¼J max
 
Y j ∈ Y Jmax , Y J max þ Δ : (68)

Step 3: Cluster searching. The values y j that are clustered about the mean m
found in the previous step can pertain to the epochs scattered along the time axis
in chaotic order. The task of this step is to define an accumulation of such points,
a so-called cluster, which we define as follows:
Definition 3. We designate as an ðm, ΔÞ cluster the set of points (epochs) of the time
axis, the indexes j of which belong to the segment [k,l] (1 ≤ k < l ≤ N) and for which the
following requirements are satisfied:

a. all points of the segment [k, l] are marked by the same value: flagk = … = flagl .

b. at the points j= k, l of the left and right boundaries of the segment, this inequality
is satisfied:
 
y j m ≤ Δ:
 
(69)

c. the amount of points at which (69) is satisfied is no less than the present value of
MINOBS;

d. the number of consecutive points in which (69) is not satisfied does not exceed the
predefined value MAXGAP (e.g., 5);
27
Satellite Systems - Design, Modeling, Simulation and Analysis

e. the value of k cannot be reduced while maintaining the requirements of

a–d; and

f. the value of l cannot be incremented while maintaining the requirements

of a–d.

This definition illustrates in Figure 12.

It is understood that for specified values of m and Δ, there might be one, several,
or no (m, Δ) clusters. Searching for clusters is performed by sequential checking of
the satisfaction of inequality (69) for all j = 1, … , N for which flagj = 0. Note that
inequality (69) is satisfied for at least N max values of j. In fact, from expressions (67)
and (68), we derive the following equations:

Y J max ≤ m ≤ Y J max þ Δ, (70)

and since the arrays y j and Y j differ only by permutation, it follows from
Eq. (68) that the inequalities
Y J max ≤ y j ≤ Y J max þ Δ: (71)

are fulfilled for N max index j. Inequality (69) is satisfied also for all these indexes,
as follows from Eqs. (70) and (71).
If cluster was found, then we mark all points of it as 1, and then, we repeat the
cluster search procedure. If even just one cluster has been found at this step, we
transfer to Step 1. If not even one cluster has been found, then the search for
clusters is complete and we transfer to Step 4.
Step 4: If even just one cluster has been found, we transfer to Step 5, and
otherwise: (a) all points of the segment ½1; N Š are marked as outliers and removed
from further processing and (b) the operation of the algorithm terminates.
Step 5: Search for individual jumps in clusters. Let us assume that n ≥ 1 clusters
have been found:

1: ½k1 , l1 Š is the ðm1 , ΔÞ cluster

…
n: ½kn , ln Š is the ðmn , ΔÞ cluster:

In each of the clusters that are found, outliers and 1-size jumps are possible.
This follows immediately from the inequalities in (69) and the preestablished value
Δ ¼ 1.2.

Figure 12.
Epochs with indexes k ≤ j ≤ l form ðm, ΔÞ cluster.

28
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

Substep 5.1: In detecting a 1-size jump in a cluster, we use modified Springer

algorithm (see Refs. [9, 10]) combined with the proposed in Section 3 algorithm that
executes a search for the optimal solution with a minimum quantity of defective data.
Substep 5.2: We find all epochs J p and values Δn5,J p ≜ n5,J p n5,J p 1 (p = 1, … , n)
of the jumps inside the clusters. The possible values for Δn5,J p are 0,  1.
Substep 5.3: We repair the data by the value of each found jump, using the formula
8
< yð p 1 Þ ; j < J p
ð pÞ j
yj ¼ ; p ¼ 1, ::, n, (72)
: y 1Þ Δn ; J ≤ j < N
ð p
j 5,J p p

ð0Þ
where y j ¼ y j .
ðnÞ
Substep 5.4: We rename: y j ¼ y j .
Step 6: Marking of points outside clusters. All points outside of the found clus-
ters (if any) are marked as outliers.
Step 7: Ordering of clusters. We renumber the clusters so that they are placed
left to right on the time axis. For the ordered set of clusters [kp, lp], p = 1, … , n, the
conditions 1 ≤ k1 < l1 < k2 < l2 < … < kn < ln ≤ N are satisfied.
Step 8: Data screening within clusters and improving the mean values of y j .
Substep 8.1: In accordance with the algorithm proposed in Section 3, we perform
screening from outliers in each of the n clusters.
Substep 8.2: For each of the clusters cleaned of outliers, we determine the
modified mean values mp∗ .
Step 9: Jumps between clusters. It follows from the description presented above
that the remaining jumps in the data y j are on boundaries between clusters. If only
one cluster is found, the algorithm is terminated: no additional jumps are detected,
and the data are cleaned of outliers. If more than one cluster is found (n > 1), then
the epochs j1, … , jn–1 of the jumps will be the coordinates of the left-hand bound-
aries of clusters, beginning from the second: jl = k2, … , jn–1 = kn. The values of jumps
Δn5, jp are found in this case as rounded to the nearest integer of the difference of
mean values of adjacent clusters:
 
∗
Δn5, jp ¼ NINT mpþ1 mp∗ , p ¼ 1, … , n–1: (73)

Step 10: Repair data. We delete the jumps between clusters using formula anal-
ogous to (72):
8
< yðjp 1Þ
ðpÞ
; j < jp
yj ¼ 1Þ
; p ¼ 1, … , n 1, (74)
: yðp Δn5,J p ; jp ≤ j < N
j

ð0Þ
where y j ¼ y j .
ðn 1Þ
We rename: y j ¼ y j . The algorithm is terminated.

10. Numerical calculations using algorithms presented

in Sections 8 and 9

We present here the results of testing the proposed algorithm using real data
obtained by the JOZ2 station, which is part of the IGS network [2]. These data are
29
Satellite Systems - Design, Modeling, Simulation and Analysis

included in the distribution set of the installation software package [3]. Testing was
carried out for data obtained from GPS satellite with number PRN = 13 for 2010,
day 207. Figure 13 shows the Melbourne-Wübbena combination values in the

Figure 13.
Values of the Melbourne-Wübbena combination for the JOZ2 station (PRN = 13 for 2010, day of year = 207).

Figure 14.
(a) Deviations of the values of the Melbourne-Wübbena combination from the mean after detection and
elimination of jumps from the data using the algorithm from Ref. [3]. (b) Deviations of the values of the
Melbourne-Wübbena combination from the mean after detection and elimination of jumps from the data using
the proposed algorithm in Section 9.

30
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

107 min time interval (N = 215). The number j of time epochs counted from the
beginning of 24-h day with interval of 30 seconds is plotted along the horizontal
axis. The combination values y j expressed in cycles with wavelength λ5 are plotted
along the vertical axis.
Figure 14a and b presents the values of the deviations from the mean value z of
data after cycle slip repair procedure, by using the algorithm applied in Ref. [3] (see
Figure 14a) and the proposed algorithm (Figure 14b). The values y j –z in cycles of
λ5 are plotted along the vertical axes, and the number j of epochs is plotted along the
horizontal axis. Epochs in which the measurement data were rejected are marked by
light circles. In the first case (see Figure 14a), 111 of the conducted measurements
or 51% of the total number of data was discarded. In the second case (see
Figure 14b), 29 of these measurements were rejected (13%), which are almost 38%
less than in the previous computation. The epochs of detected jumps are marked by
daggers.

11. Conclusion

This chapter presents several effective and stable algorithms for processing data
received from GNSS receivers. These data form the basis of almost all engineering
applications in the field of computational geo-dynamics and navigation and cadas-
tral survey and in numerous fundamental research works as well. The accuracy of
the results obtained is significantly influenced by the quality of the data used in the
calculations. In particular, the presence of rough measurements (outliers) in the
observation data can significantly reduce the accuracy of the calculations carried
out. One of the tasks at the preliminary stage of data processing is reliable detection
and removal of rough measurements from the series of measured data with mini-
mum amount of rejected data. The so-called optimal solution, introduced in the
chapter, made it possible to detect and eliminate outliers from observed data min-
imizing the number of rejected measurements. In addition, it is assumed that the
data may contain a trend as an unknown function of time. The strategy for deter-
mining of the trend is depending on the physical process in question under an
assumption that the trend is a continuous function of time. The efficiency of the
search is definitely influenced by the choice of the function class from which the
trend is searched. In this chapter, we considered the class of power polynomials, but
in some cases, this choice may not lead to the expected result. It may require, for
example, a class of trigonometric functions to find a suitable trend. The automatic
search for the best functional class, together with the strategy of effectively finding
an unknown trend, against the background of random noise and outliers, is a
complex task for future research.

31
Satellite Systems - Design, Modeling, Simulation and Analysis

Author details

Igor V. Bezmenov
“VNIIFTRI”, Mendeleevo, Moscow region, Russian Federation

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

32
Effective Algorithms for Detection Outliers and Cycle Slip Repair in GNSS Data Measurements
DOI: http://dx.doi.org/10.5772/intechopen.92658

References

[1] Perov AI, Kharisov VN. GLONASS: satellite systems. Measurement

Principles of Construction and Techniques. 2018;61:878-884. DOI:
Operation. 4th ed. Moscow: 10.1007/s11018-018-1518-y
Radiotekhnika; 2010
[8] Blewitt G. An automatic editing
[2] International GNSS Service. algorithms for GPS data. Geophysical
Available from: wwv.igs.org/network Research Letters. 1990;17(3):199-202
[Accessed: 4 April 2018]
[9] Springer TA. Modeling and
[3] Dach R, Andritsch F, Arnold D, et al. validating orbits and clocks using the
In: Dach R et al, editors. Bernese GPS global positioning system. In:
Software, Ver. 5.2. Astronomical Geodätisch-Geophysikalische Arbeiten
Institute, University of Bern; 2015. DOI: in der Schweiz. Vol. 60. Zurich:
10.7892/boris.72297. Available from: Institute. Geodesy and
www.bernese.unibe.ch [Accessed: 20 Photogrammetry, ETH Zurich; 2000
May 2020]
[10] Bezmenov IV, Blinov IY,
[4] Melbourne WG. The case for ranging Naumov AV, et al. An algorithm for
in GPS based geodetic systems. In: cycle-slip detection in a Melbourne–
Goad C, editor. Proceedings of the 1st Wübbena combination formed of code
International Symposium on Precise and carrier phase GNSS measurements.
Positioning with the Global Positioning Measurement Techniques. 2019;62:
System. Rockville, Maryland: US 415-421. DOI: 10.1007/s11018-019-
Department of Commerce; 1985. 01639-5
pp. 373-386

[5] Wubbena G. Software developments

for geodetic positioning with GPS using
TI 4100 code and carrier measurements.
In: Goad C, editor. Proceedings First
International Symposium on Precise
Positioning with the Global Positioning
System. Rockville, Maryland: US
Department of Commerce; 1985.
pp. 403-412

[6] Sanz Subirana J, Juan Zornoza M,

Hernández-Pajares M. Detector Based in
Code and Carrier Phase Data: The
Melburne–Wübbena Combination. ESA
Navipedia. Available from: https://gssc.
esa.int/navipedia/index.php/Detec
tor_based_in_code_and_carrier_pha
se_data:_The_Melboume-Wübbe
na_combination [Accessed: 24 January
2019]

[7] Bezmenov IV, Naumov AV,

Pasynok SL. An effective algorithm for
elimination of outliers from data
measurements of global navigation

33
Chapter

Analysis of Spatiotemporal
Variability of Surface Temperature
of Okhotsk Sea and Adjacent
Waters Using Satellite Data
Dmitry Lozhkin

Abstract

The chapter is divided into 5 parts. The first part describes what the satellite data
is and describes the levels of its processing. In the second part, attention is paid to
sea surface temperature anomalies, the conditions for the appearance of the most
significant anomalies and their influence on the behavior and survival of aquatic
organisms are described. The third part is devoted to calculating linear trends in
ocean surface temperature from a 20-year series of satellite data. It is shown that
the heat content of the surface layer of Okhotsk Sea decreases, most significantly in
its northern and western parts. This trend is especially pronounced in the spring,
which may be due to a decrease in ice cover and a more significant cooling of the
waters due to winter convection. In the fourth part, periodic fluctuations in the
temperature of the surface of the ocean are considered. It is demonstrated how,
using the calculated trend and several basic harmonics, one can try to predict the
temperature next year. And the last part concludes the chapter.

Keywords: satellite data, sea surface temperature, Tsushima current, Okhotsk Sea,
regional effects of global warming, SST anomalies, trends, harmonic, cyclicity

1. Introduction

The study of the thermal regime of various water areas is one of the most
important oceanological problems, since the spatio-temporal variability of water
temperature reflects complex processes of formation, transformation and dynam-
ics of water masses. In addition, temperature is one of the key parameters that
determine the conditions for the existence and development of most species of
aquatic organisms; therefore, the study of this problem is also of key importance for
hydrobiology. The zoning of water areas by the nature of temperature conditions,
as well as their forecasting, taking into account the peculiarities of seasonal and
interannual variability, is an important scientific task, which also has a pronounced
applied aspect associated with the fact that accumulations of some species of com-
mercial fish are confined to the zones of separation of water masses with different
characteristics.
Direct measurements are a traditional source of water temperature data.
Their accuracy depends only on the accuracy of the device, and the discreteness

1
Satellite Systems - Design, Modeling, Simulation and Analysis

depends on the specific task (in some cases it can reach a fraction of a second).
They can also be used to construct vertical profiles with high vertical resolu-
tion. However, recently the number of expeditions has been steadily declining,
and the data obtained through direct measurements are point and irregular and
are not suitable for studying large-scale phenomena or for obtaining data from
hard-to-reach places.
Satellite data, on the other hand, are regular and allow covering the entire
water area of the studied basin. Therefore, they are very good sources for studying
seasonal and interannual variations in the temperature of the surface layer (the
thickness of which ranges from 1 to 10 meters). The disadvantage of these data is
the strong dependence of accuracy on cloudiness (especially in the infrared and vis-
ible ranges). Ideally, they should be regularly compared with direct measurements
in order to identify errors in data interpretation.
For scientific purposes, data are usually used from satellites located either in a
geosynchronous (in a particular case, geostationary orbit) or in a heliosynchronous
orbit. The advantage of the geostationary orbit is obvious: you can get a picture of
the same area with discreteness of up to half an hour, and after pointing to the satel-
lite, a constant correction of the antenna position is not required. It is very useful
for telecommunication systems as well as for obtaining meteorological data. The
disadvantage is the lack of coverage at polar latitudes.
This disadvantage is easily eliminated by satellites in a polar sun-synchronous
orbit. A sun-synchronous orbit is important to science because it keeps the angle
of incidence of sunlight on the Earth’s surface more or less constant, although the
angle will change with the change of seasons. Without a sun-synchronous orbit, you
would have to account for changes in shadow and lighting angles, making it difficult
to track changes over time. It would simply be impossible to collect the information
needed to study climate change.
In 1997, a TeraScan satellite receiving station was installed at SakhNIRO, with
the help of which data are received from the NOAA, Metop, Aqua and Terra satel-
lites in a polar sun-synchronous orbit (Figure 1 [1]). At the moment, a 21-year
series of satellite data on the temperature of the surface of the Sea of Okhotsk
and adjacent waters has been accumulated, which makes it possible to analyze the
interannual variations of this parameter.

Figure 1.
Example of a sun-synchronous orbit for a NOAA satellite [1].

2
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

This chapter is organized as follows:

• Section 2 discusses satellite data

• Section 3 discusses Sea Surface Temperature (SST) anomalies

• Section 4 analyzes long-term data series using linear regression

• Section 5 presents results on periodic SST fluctuations

• Section 6 concludes the chapter.

2. Satellite data

What is satellite data? An electromagnetic (EM) signal with certain charac-

teristics leaves the sea and contains information about the primary observable
quantities, which are light, radiation temperature, roughness and sea level. This
signal travels through the atmosphere, where it can be distorted, and where noise is
added to the signal before it is received by a sensor that registers certain properties
of the radiation and converts each measurement into a digital signal to be encoded
and sent back to earth. The sensor geometry limits each individual observation to a
specific instantaneous field of view.
In order to convert a digital signal received at a ground station into a useful one
for scientific measurements with a certain accuracy and quality, the process of
obtaining data by the sensor should be numerically transformed using knowledge
about signal processing technology, physical and technical properties of the sensor,
the physics of the atmosphere, and the interaction of EM radiation with the ocean
and near-surface processes in the ocean. The process of extracting data from a
unique observation location in space is at the heart of the satellite data feature.
Table 1 summarizes 5 levels of satellite data processing.
Using the TeraScan software, raw data received from satellites is automatically
calibrated and atmospheric correction, as well as georeferencing of data and projec-
tion onto a standard map (selection of parameters, projections, etc., is performed

Level Product description

0 Raw satellite data in binary format.

1 Image in measurement coordinates. Satellite data of individual calibrated channels.
1,5 (or 1А) In special cases, a Level 1 product with atmospheric correction applied.

2 Atmospheric corrected and calibrated product containing upstream values or directly

variable ocean response with reference to geographic coordinates, but most often not
mapped.

3 Composite images of the obtained ocean variables transferred to a standard map, obtained
by averaging over space and time of several flights in the form of a level 2 product. This
product contains data from only one sensor and may have gaps.
4 An image representing the ocean variable averaged within each cell of the spatio-temporal
grid, for the creation of which the gaps in the level 3 product data are filled with data
analysis methods, incl. Interpolation. In the course of the analysis, it is possible to use several
level 2 or 3 products from different sensors, and it is also possible to use data from field
observations or model calculations.

Table 1.
A summary of various satellite data processing levels.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 2.
An example of a ten-day average image with a spatial resolution of 0.25 × 0.25 degrees (left) and 2 × 2 km
(right) [2].

once by the operator, after which it is performed in automatic mode). Then data
with different spatial resolution (the maximum resolution for a set of satellites is
2 km, images with a spatial resolution of 250 m can be obtained on separate chan-
nels) are entered into the database, where the operator manually corrects each
image (eliminates gross errors that most often appear in the case of the presence of
semi-transparent clouds). After that, daily data averaging is performed. As the data
accumulate, ten-day and monthly averaging is also performed (as shown in Figure 2
[2]), as well as the calculation of anomalies (deviations from the long-term average
value in a specified decade or month, obtained over a number of previous years).
The resulting data set should be carefully analyzed. Determination of the char-
acteristics of unidirectional trends, as well as the amplitudes and phases of cyclic
components and the possibility of using them to predict thermal conditions in the
next season constituting the content of this study.

3. SST anomalies

There are two types of SST anomalies as for any parameter distributed in space
and time. The first is the sharp deviation of the specified parameter from the mean
values in the neighboring areas. Anomalies of this kind can be regular and even
present on average long-term maps (Figure 3 [3]). Typical examples from the point
of view of SST are upwelling zones (quasi-stationary), mouths of large rivers, as
well as zones of influence of warm and cold currents. Some anomalies of this nature
may be unstable and dependent on weather conditions. Such anomalies determine
the behavior of aquatic organisms, since each species has its own conditions of
existence and development (some species need colder water, and some are more
comfortable in warmer water).
The second type of SST anomaly is a deviation from the multiyear average.
This type of anomaly depends entirely on the current conditions and reflects the
peculiarities of the distribution of the parameter in a given year. From the point
of view of SST, this may mean more or less intense heating or cooling than usual.
The appearance of such anomalies can lead to a change in the timing of the fish’s
approach to spawning (Figure 3 [3]) or to migration to places with more favor-
able conditions, and in the early stages of development it can lead to death from
unfavorable conditions.
Let us consider in more detail the extraordinary conditions that took place in the
spring of 2011, when the greatest delay in the release of juveniles from the Sakhalin
SFHs was noted. The spatial distributions of sea surface temperature anomalies

4
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

observed during the period under consideration, which is strongly characterized by

unusual conditions (Figure 4 [4]), are quite remarkable.
In the third decade of May (Figure 4A), significant negative deviations from the
norm were noted along the entire eastern coast of Sakhalin Island and in the Aniva
Gulf; they were the largest (up to 3°C) in the basin of the Mordvinova Gulf. Judging
by the zone of negative anomalies stretching from the Terpeniya Peninsula along
the line of the depth increase, which corresponds to the core of the East Sakhalin
Current [5], they were caused by the outflow of cold water and ice-cover remains
from the northeastern shelf. This opinion is confirmed by a satellite image on June 2,
2011 (Figure 3 [3]), which clearly shows the transport of drifting ice from the
north to the Terpeniya Peninsula and further towards the Mordvinova Gulf. Such a
situation is observed relatively rarely: usually, under the influence of the southern
winds characteristic of this period of the year, the ice that is transported from the
Sakhalin Gulf is blocked at the northeastern tip of the island and does not move
south of 52° N. A similar ice removal was observed in spring 2005 [6]. Then, due
to the unusual distribution of the surface atmospheric pressure, the northeastern

Figure 3.
An example of the long-term average distribution of SST, the actual temperature in the second decade of July
2018 and the 2018 anomaly relative to the multi-year average (these conditions led to a two-week delay in the
approach of pink salmon to spawning rivers) [3].

Figure 4.
Distribution of surface temperature anomalies (°С) in the third decade of may (A) and in the second decade of
June (B) 2011. Ice conditions off the east coast of Sakhalin Island (right) on June 2, 2011 (snapshot of the aqua
satellite). The strait of Tatary is closed by heavy clouds [4].

5
Satellite Systems - Design, Modeling, Simulation and Analysis

winds prevailed in the study region, which just contribute to the transport of
drifting ice to the southeastern coast of Sakhalin Island, as well as to the estuarine
regions of the Naiba and Ai rivers and to the Mordvinova Gulf.
In spring 2011, the distribution of surface atmospheric pressure also signifi-
cantly differed from the characteristic one in the given period of the year. For
example, in the third decade of May, the high-pressure region was located in
the southern part of the Sea of Okhotsk, above the Kuril Basin (Figure 5A [7]).
Usually this area is located in the eastern part of the sea, and air flows directed
to the north are formed above this basin. When the pressure distribution has this
structure, the atmospheric circulation typical for the summer monsoon is not
formed; in principle, it was weakly expressed, since the pressure gradients were
small and cold waters with low salinity moved freely to the south, transporting
drifting ice.
Subsequently, the high-pressure region shifted from the sea to the Pacific Ocean.
In particular, in the second decade of June, zonal, northeast-oriented air flows
prevailed, which is not typical for the warm season (Figure 5B [7]).
However, it is unlikely that only the peculiarities of the synoptic conditions
can explain the significant negative temperature anomalies of the sea surface in
the northern part of the Sea of Japan, primarily in the main zone of the Tsushima
Current influence, off the west coast of Hokkaido Island (up to 5–6°C). Another
zone with approximately the same negative anomalies was noted in the zone of the
influence of the Amur River runoff in the region of its mouth, in the western part
of the Amur Estuary, and in a significant part of the Sakhalin Gulf, as well as on
the northeastern shelf of Sakhalin Island (up to about 52° N). Pronounced nega-
tive anomalies were observed both on the southwestern and southeastern coasts of
Sakhalin Island, as well as in the Aniva Gulf. It is difficult to explain the cause of
such a heat deficit in the surface waters of a very wide basin: one of them could be
a decrease in the intensity of the Tsushima Current, whose quasi-periodic fluctua-
tions (a cycle with a period of 5–6 years is especially distinguished) have a signifi-
cant influence on the climate of the Sakhalin-Kuril region [8]. It is also difficult to
assess how often such situations can be observed which have led to a delay in the
release of juveniles from Sakhalin SFHs lasting from 3 weeks to a month. Probably,
due to the low temperature of coastal waters observed for a long time, there could
be a mass death of juveniles of natural origin which rolled along the spawning rivers
of the island.

Figure 5.
Distribution of surface pressure (hPa) over the Sea of Okhotsk in the third decade of may (A) and the second
decade of June (B) 2011 [7].

6
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

4. Linear trends

When analyzing long-term data series, a question arises – are there unidirec-
tional tendencies of one or another parameter in the specified period of time?
This is especially important in the context of global warming, the most noticeable
results of which are an increase in winter air temperature in the Arctic latitudes and
a decrease in ice cover (area of the water area covered with ice) both in the Arctic
Ocean and in the Sea of Okhotsk.
In order to answer this question, in each spatial cell of the 1000 × 1022 matrix,
covering the investigated water area (42–60°N and 135–163°E), the linear trend
coefficients were calculated using the least squares method, meaning the tempera-
ture increase over the year … The calculation was carried out on a 20-year series of
data for each month separately, for annual averages, as well as for average tempera-
tures for the season. The seasons for the calculations are shifted by a month relative
to the calendar ones due to the specifics of the region: January–March (winter),
April–June (spring), July–September (summer) and October–December (autumn).
The linear trend coefficients calculated from the full annual series in the study
region are mostly negative, and there are no regions with positive values in the Sea
of Okhotsk (Figure 6 [9]). Significant cooling of the surface layer (about 1.5°C over
10 years) was observed in the northern and western parts of the sea; most clearly
this process is manifested in the northern part of the sea at some distance from
the coast. A less pronounced decrease (at a rate of about 0.5°C over 10 years) is
observed in the Kuril Islands and in the band along the western coast of Kamchatka
beyond the shelf zone, as well as in the influence zone of the Amur River outflow
(Amur Estuary, Sakhalin Bay, northern shelf of Sakhalin Island).
A significant decrease (more than 0.5°C over 10 years) was found in the north-
ern part of the Strait of Tatary; further southward, the cooling rate of surface
waters decreases and, south of 45° S, a slight increase in the temperature of the
surface waters in the Sea of Japan was recorded. The warming process is more
pronounced in the northwestern part of the Pacific Ocean, i.e., in the southeastern
part of the study region.
Let us now consider the results of calculating the parameters of the linear trend
in different seasons. In winter, there is not enough data for a reliable calculation
in the northwestern part of the sea (the “Sea of Okhotsk fridge”), on the northern
shelf of the sea, and on the northeastern shelf of Sakhalin. In many regions of the
basin, where the ice cover is significantly influenced, the calculation is not reliable
enough; nevertheless, some ideas of the trends in the thermal regime during the
cold period can be put forward.
In the entire Sea of Okhotsk, especially in the northern and western parts of the
sea, as well as in the Strait of Tatary of the Sea of Japan and in the regions of the
Pacific Ocean adjacent to the Kuril Islands, there is a tendency for the temperature
decrease in the surface layer. Signs of the opposite trend are recorded only in the
southeastern part of the study region, in the Pacific Ocean, in the northern part
of the Sea of Japan, and in the very southern part of the Strait of Tatary (south
of 47° N).
The most pronounced decrease in the surface layer temperature is in the
springtime. It covers the entire study region, including the northwestern part of the
Pacific Ocean, and it is particularly significant in the northern and western parts of
the Sea of Okhotsk (the rates of temperature decrease range from 1.0 to 1.5°C over
10 years and even higher in some regions).
The cooling of the surface layer in spring is the most evident consequence of
a decrease in ice cover, since in the absence of the ice cover the cooling processes

7
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 6.
Distribution of linear trend coefficients by number of average annual (above) and average seasonal values. The
scale corresponds to degrees celsius over 10 years [9].

cover a layer of water thicker than in the presence of the ice cover. Undoubtedly, this
has a significant impact on the climate of the region and, above all, on the weather
conditions of Sakhalin, which are the most affected by this process.
While calculating the trends in sea surface temperature in individual months, it
was found that the most intense cooling is observed in the month of May, especially
in the northwestern part of the Sea of Okhotsk and along the entire eastern coast
of Sakhalin Island from Cape Elizabeth in the north to Cape Aniva in the south.
Exceptions are the western part of the Amur Estuary, which accounts for the bulk
of the Amur flood, and the basin adjacent to Tauiskaya Bay. It is obvious that the
hydrological regime of coastal areas experiencing the influence of the river outflow
differs from that typical for the Sea of Okhotsk as a whole.
In summertime, the region is dominated by an increase in the sea surface tem-
perature most pronounced in the northwestern part of the Pacific Ocean. In the Sea
of Okhotsk, it is noted in its northwestern part, in the region of the Shantar Islands,

8
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

in Sakhalin Bay, and in the Amur Estuary, to the east of the northern tip of Sakhalin
Island, at the northwestern coast of the Kamchatka Peninsula, and near the Urup
and Simushir islands.
In the northern part of the sea, near the Kuril Islands, on the southeastern coast
of Sakhalin Island and along the Primorye coast in the Sea of Japan, there is a trend
towards a decrease in temperature, although it is more moderate than in springtime.
Moreover, in July, the tendency towards a decrease in the sea surface temperature
still prevails in the study region; the trend changes only in the northwestern part
of the Pacific Ocean. In August, in these regions, warming reaches the highest rates
(about 1.5°C over 10 years); the temperature increase was noted in the central part
of the Sea of Okhotsk, along the western coast of Kamchatka and off the southeast-
ern coast of Sakhalin, in the Strait of Tatary and the Amur Estuary. In the northern
part of the sea, on the northeastern shelf of Sakhalin, and in the Kuril region, the
trend towards a decrease in sea water temperature continues.
In September, on the contrary, the intensity of warming in the Pacific Ocean
decreases; the greatest warming rate was noted in the northern and northwest-
ern parts of the sea (especially in the shelf zone between Ayan and Okhotsk
settlements).
A diffuse pattern is observed in the fall. Relatively small negative trends prevail
in the greater part of the Sea of Okhotsk; positive trends (also insignificant) are
noted in the eastern part of the sea on the western shelf of Kamchatka and in a rela-
tively narrow band along the northern coast. Smaller areas were also found in the
Shantar Islands and Kashevarov Banks and in the southwestern part of the analyzed
basin of the Sea of Japan.
The highest rates of the temperature decrease in the surface layer are observed
in the Strait of Tatary, in its northern part, and along the western coast of Sakhalin,
as well as in the Sea of Okhotsk, in the northern part of it, beyond the shelf, off the
southeastern coast of Sakhalin, and also in the band between 145° and 150° E from
the Urup and Iturup islands to the Kashevarov Banks.
It was found in the calculations for individual months that a positive trend
prevailed in October, and the trend reversed in November and December.
In the work [10], it was found that, in the Sea of Okhotsk and in the region of
the Kuril Islands, negative trends were also noted in the spring months; the largest
positive trends were recorded in October. Similarity results become very interesting
if we take into account the differences in the studied time periods and spatial char-
acteristics of the regions in which the trends were calculated. In the Sea of Japan, no
negative trends were identified in this work.
As a result of the analysis of data on the surface temperature of the Sea of
Okhotsk over a 20-year period (1998–2017), it was found that the global climate
change in this basin caused a decrease in the ice cover and a decrease in the tem-
perature of the upper water layer during the winter–spring period. The negative
trends in temperature in the spring in the northern and western parts of the study
region, as well as in the Strait of Tatary of the Sea of Japan (from 0.5° to 1.5°C over
10 years), are especially large. This effect in the reduction of the ice cover, both
in time and in space, is the most evident and can be explained by an increase in
the depth of winter convection. The predominance of a decrease in sea surface
temperature, although less pronounced, was also noted in winter and autumn, and
generally throughout the year.
In summer, the region is dominated by an increase in sea surface temperature,
most pronounced in the northwestern part of the Pacific Ocean. In the Sea of
Okhotsk, it was noted in its northwestern part, east of the northern tip of Sakhalin
Island, off the northwestern coast of the Kamchatka Peninsula, as well as in some
other regions. Moreover, in July, the main role belongs to the processes of cooling

9
Satellite Systems - Design, Modeling, Simulation and Analysis

of the surface layer; the change in the trend occurs in August and continues in
September–October.
These processes play a significant role in the climate variations in the Sea of
Okhotsk; in particular, a decrease in the temperature of sea water is noted in the
coastal waters of Sakhalin Island. In addition to the weather conditions, the results
are important for studying the habitat conditions of commercial fish species and
invertebrates in the basin, which is of great fishery importance.

5. Periodic fluctuations

If you look closely at the graph of the course of the mean monthly SST (Figure 7
[11]), then in addition to seasonal fluctuations, you can notice interannual varia-
tions. They are especially pronounced in August, when the mean monthly water
temperature reaches its maximum value, and are expressed in the modulation of the
annual harmonic. Moreover, these oscillations are of a quasiperiodic nature, most
likely associated with certain phenomena in the atmosphere and hydrosphere.
Any periodic oscillations can be described by knowing their amplitude, phase
and period. Using the least squares method, you can find the corresponding ampli-
tude and phase for each selected period. Thus, it is possible to establish a kind of
“influence zones” of harmonics with a certain period, i.e. areas in which the ampli-
tude of one or another harmonic exceeds a certain threshold value. And in order
not to interfere in the calculations with long-term components (the period of which
exceeds half the length of the series), the trend obtained according to the method
described in the previous section was subtracted from the initial data.
The distribution of the amplitudes of temperature fluctuations in the studied
region is rather complex. Having determined in each spatial cell the period cor-
responding to the largest amplitude and displaying the obtained data on the screen,
we have established several main periods that play a significant role in interannual
SST variations in most cases. For a series of 21 years, it is not entirely correct to
calculate fluctuations with a period exceeding 11 years. Short-period fluctuations
are unstable and generally have little information. Thus, the spatial distributions of
harmonic amplitudes with a period from 3 to 11 years were considered in detail.
In most of the studied water area, the main role was played by variations with
a period of about 5.5–6 years (on the graphs for points 2, 5 and 8 in Figure 8,
you can see that the largest amplitude at these points corresponds to a period of
66–68 months), the spatial distribution of the amplitude with a period of 6 years is
shown in Figure 9 [12]. The zone of its influence is the most extensive and occupies

Figure 7.
Annual variation of average monthly temperature (left) and average temperature in august (right) [11].

10
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

the southern half of the Sea of Okhotsk and the northern part of the Sea of Japan,
up to the Amur estuary. The amplitude of this harmonic in the zone of its influence
ranges from 1 to 2°C. In the vicinity of the Kuril Islands, off the western coast of
about. Hokkaido, off the northeastern coast of about. Sakhalin, the amplitude is
slightly lower (from 0.5 to 1°C). Further to the north, starting from 52 N, its influ-
ence decreases and practically disappears.
In [8], based on the EOF decomposition of the sea surface temperature in the
North Pacific Ocean, it was shown that an oscillation with a similar period is charac-
teristic of the entire region influenced by the Kuroshio Current and its branch, the
warm Tsushima Current. This is evidenced by the large amplitudes in the zone of
the indicated currents and in Figure 9. Consequently, as a result of this study, it was
possible to estimate the boundaries of the influence of this component in the Sea of
Okhotsk, which runs parallel to the islands of the Kuril ridge and divides this basin
into two practically equal parts. The entire northern Sea of Japan is significantly
affected by this cycle.
As seen from Figure 8, at most points, one can also note peaks in the range of
periods from two to three years. The highest values of the amplitude of the 3-year
harmonic can be noted in the northwestern part of the Sea of Okhotsk, at a distance
from the coast, as well as near the northwestern coast of Kamchatka and in the strip
from 47 to 490 N. and from 147 to 1490 E. in the area of the Kuril deep-water basin
and in the northwestern part of the Pacific Ocean. However, the amplitudes of these
oscillations are somewhat lower than those of the six-year harmonic (from 1 to
1.5°C). At a distance from these regions, the amplitude gradually decreases to zero.
The spatial distribution of the amplitudes of the cyclic component with a period
of 5 years differs markedly from that considered above for a period of 6 years. The
zone of its influence is noticeably narrower, it is concentrated mainly on the north-
ern shelf of Hokkaido, in the region of the South Kuril Islands (vast waters both on

Figure 8.
Examples of graphs of the dependence of the amplitude of the harmonic (in °C) from its period (in months).
The location of the points is shown in Figure 9 [12].

11
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 9.
Distribution of amplitudes (in °C) of interannual fluctuations in the mean monthly sea surface temperature
(august) with a certain period (indicated in years in the upper left corner of the image) [12].

the Sea of Okhotsk and on the ocean sides), and, surprisingly, on the northeastern
shelf of Sakhalin, where the influence of a lower frequency component was not
noted. In the Tatar Strait, its role is also noticeable, but expressed to a lesser extent
than the 6-year harmonic.
In the western part of the Sea of Okhotsk and the northwestern part of the
Pacific Ocean, a cyclical component with a period of about 8 years is significantly
manifested. The zone of influence of the Amur river runoff in summer is clearly
distinguished in the spatial distribution - the Amur estuary, the southern and
eastern parts of the Sakhalin Bay, the area between the Schmidt Peninsula and the
Kashevarov Bank [13]. It is interesting that in the area of the Kuroshio Current
manifestation, this component has large amplitudes, in the Tsushima Current zone
in the Sea of Japan - insignificant, while on the northern shelf of Hokkaido and on
the Sea of Okhotsk side of the Southern Kuril Islands, where the warming effect of
the Soya Current affects, the amplitudes are significantly.
The lowest frequency of the considered harmonics with a period of 11 years
is manifested in the northern part of the Sea of Okhotsk; in other parts of the
study area, its role is insignificant. It is rather difficult to put forward a reasonable
hypothesis that could explain such significant differences in the very long-term
variations in SST in different parts of the same basin. It can only be assumed that

12
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

due to the comparative shallowness of the northern region, the effect of the winds
of the southern rumba and the greater number of sunny days than in the southern
part, due to the lesser influence of cloudiness, the influence of the solar cycle is
more noticeable here.
Attention is drawn to the fact how the zones of manifestation of harmonics shift
with a period of 5 to 11 years. If the zone of influence of the 5-year harmonic is
focused near the islands of Sakhalin, Hokkaido and the Southern Kuriles, then with
an increase in the period, the region with the highest amplitude shifts clockwise
(towards the northeastern coast of Sakhalin and further to the northern part of the
Sea of Okhotsk). This interesting fact is also difficult to give a reasonable explana-
tion, it requires additional study.
In work [8], a method was developed for predicting thermal conditions for a
year in advance in certain areas of the studied water area (this method was also
used to recover data gaps associated with the influence of cloudiness or technical
reasons), which consisted in calculating the temperature in a given square in time t
according to the formula:
N
T ( t ) = at + b + ∑ck cos (ωk t − ϕ k ) (1)
k =1

where a and b are the parameters of the linear trend, 𝑐𝑘 are the amplitudes, and
𝜑𝑘 are the phases of the cyclic components (harmonics) of sea surface temperature
variations. An essential feature of the method is the fact that the amplitudes and
phases of the main cyclical components are calculated by the least squares method,
with their periods ranging from 18 to 144 months with a step of 1 month. For each
cell, a set of 3–4 harmonics was determined, which make the largest contribution to
the interannual variations in SST. Since they are not orthogonal, for forecasting
using formula ( 1 ), it is necessary to subtract the calculated wave from the initial
series before determining the parameters of the next one in order to avoid double
inclusion of coherent components (in [14, 15]) such a technique was called
“Sequential spectra method”).
Based on the parameters of the obtained cyclic components, a retrospective
forecast of thermal conditions for the summer of 2018 was carried out (observa-
tional data for 1998–2017 were used to calculate the parameters of harmonics and a
linear trend). The calculation was carried out for each spatial cell according to
formula ( 1 ), taking into account the trend and four harmonic components with the
highest amplitudes. For the forecast, periods from 18 to 144 months were covered.
The calculation results are presented in Figure 8 in the form of graphs of forecast
curves and real variations in sea surface temperature, including the predicted values
that took place in the summer of 2018. Figure 10 shows the spatial distribution of
the difference between the predicted and actual temperatures for August 2017 and
2018. The forecast was built for a year ahead along the entire previous series.
The curves, which are the sum of the trend and the first four harmonics, gener-
ally repeat the actual interannual temperature fluctuations. The correlation coef-
ficient of the initial and predicted series at the selected points exceeds 90%. Note
that even the first two harmonics in many cases provide a correlation coefficient
of more than 70%. Despite the fact that 2018 was anomalous in terms of thermal
conditions (the Tsushima Current and its Okhotsk branch of the Soya Current were
weakened, a heat deficit was felt in the zone of influence of the Amur River runoff),
and in some other areas, even in such water areas, the forecast can be considered
acceptable. An example of a similar situation is given for the Tatar Strait, where
the predicted value was higher than the actual one, but the general course was
predicted correctly, and the error was not so great. For the northern part of the Sea
of Okhotsk, the northeastern shelf of Sakhalin Island, and a number of other areas,

13
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 10.
Examples of temperature forecast graphs (in °C) for the next year. The dashed line shows the prognostic curve.
The forecast is carried out for the period from 1998 to 2017. The actual temperature in 2018 is marked with a
cross [11].

good agreement was observed between the calculated and real values of the surface
layer temperature.
Let us consider some of the parameters of the graphs below. The standard devia-
tion of the initial and predicted series correspond to each other and range from
1.5°C (Tatar Strait) to 2°C (South Kuriles). The average displacement of the pre-
dicted series relative to the initial one ranges from 0.4 to 0.6°C. The forecast error is
1.7°C in the Tatar Strait, 1.3°C near the Southern Kuriles, 0.3–0.4°C in the northern
part of the studied water area.
More detailed studies devoted to predictability and the limits of applicability of
the approach used will be carried out later. However, we can already say that for a
significant part of the Sea of Okhotsk regions and adjacent water areas, the forecast
of the surface layer temperature with a one-year lead time is quite successful,
although the abnormally cold temperatures that took place in a number of areas in
2018 are rather difficult to predict.
Let us take a closer look at Figure 11. The forecast for August 2017 turned out to
be quite successful, the discrepancy between the actual and predicted temperatures
in most of the water area does not exceed ±2°C, with a standard deviation of SST
of about 1.5-2°C (only in the northwestern part of the Pacific Ocean is the standard
SST deviation is within 2-4°C). At the same time, the forecast for August 2018
contains a large area within which the temperature estimate was greatly overesti-
mated (over 4°C). The map of SST anomalies for August 2018 [16] also contains
areas of low temperatures (3-4°C lower than the average multiyear norm), which
coincide in space with areas of unsuccessful forecast. This area is located in the zone
of influence of the Tsushima Current, and the forecast inaccuracies are due to the
fact that the weakening of this current occurred two years earlier than the expected
date. Indeed, in Figure 8, we see that in the Tatar Strait and the Southern Kuriles,
the distance between two neighboring SST minimums decreased to 3–4 years, while
its quasiperiodic oscillations with a period of about 6 years are described in the
literature [17].
As a result of the analysis of the data set on the surface temperature of the Sea
of Okhotsk and adjacent waters, the main cyclical components responsible for
the interannual variations of this parameter and the “zones of influence” of each

14
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

Figure 11.
Difference between predicted and actual temperatures [11].

harmonic were determined. It is shown that the main contribution to these varia-
tions comes from components with a period of about 6 years, as well as 3, 5, 8, and
11 years.
The zone of influence of the fundamental harmonic is the most extensive and
occupies the southern half of the Sea of Okhotsk and the entire northern part of the
Sea of Japan; its amplitude is within 1–2°C. In the vicinity of the Kuril Islands, off
the western coast of Hokkaido Island, off the northeastern coast of Sakhalin Island,
the amplitude is slightly lower (0.5–1°C), and in the northern part of the Sea of
Okhotsk its influence is insignificant. Most likely, this component is associated with
fluctuations in the Kuroshio Current and its branch, the Tsushima Current [17].
The highest values of the amplitude of the 3-year harmonic (1–1.5°C) can be
noted in the northwestern part of the Sea of Okhotsk, at a distance from the coast,
as well as off the northwestern coast of Kamchatka, in the region of the Kuril deep-
water basin and in the northwestern parts of the Pacific Ocean.
The area of manifestation of the component with a period of 5 years is notice-
ably narrower, it is concentrated mainly on the northern shelf of Hokkaido, in the
region of the South Kuril Islands (both from the Sea of Okhotsk and the ocean
side), and on the northeastern shelf of Sakhalin. In the Tatar Strait, its role is also
noticeable, but expressed to a lesser extent than the 6-year harmonic.
In the western part of the Sea of Okhotsk, in the zone of influence of the Amur
River runoff, as well as in the northwestern part of the Pacific Ocean, a cyclical
component with a period of about 8 years is significantly manifested.
The lowest frequency of the considered harmonics with a period of 11 years is
manifested in the northern part of the Sea of Okhotsk; in other parts of the study
area, its role is insignificant.
Together with the parameters of the linear trend [9], the amplitudes and phases
of the main cyclical components (in each spatial cell, 4 harmonics with the high-
est amplitudes were used) can be used to predict thermal conditions for the next
summer. The retrospective calculation for 2018 gave generally satisfactory results,
despite the abnormally cold conditions of this year, noted in a number of areas of
the studied water area. The possibility of predicting thermal conditions is of practi-
cal importance, primarily for assessing the conditions for the approach of Pacific
salmon to spawning. And the results obtained show that in some areas of the water
area, one can count on a fairly accurate forecast even for such an unstable parameter
as the ocean surface temperature.
In general, the success of the forecast is influenced by how pronounced the
cyclical components with a certain period in a given area. Identification of the
zone of influence of various harmonics allows you to determine the boundaries of
the regions in which the applicability of this method can be expected. As for the

15
Satellite Systems - Design, Modeling, Simulation and Analysis

accuracy of the forecast, one should pay attention to the presence of large areas in
which the modulus of the difference between the predicted temperature and the
actual one was of the order of two standard deviations. This fact shows that this
method does not guarantee the success of the forecast in cases where strong tem-
perature anomalies are observed. It can only be used to obtain a primary estimate of
the ocean surface temperature (or another parameter that experiences quasiperiodic
oscillations) based on a sufficiently long series. The 21-year series is not enough to
estimate low-frequency components (with a period of 30–50 years), which could
also affect the quality of the forecast. For a better forecast, you can combine this
method with an assessment of the current conditions, adjusting the forecast in
the direction of increasing or decreasing temperature, depending on the current
meteorological conditions.

6. Conclusion

This chapter shows how satellite data, going through all levels of processing,
become useful products for both science and applied environmental prediction
problems. Assessment, and later predicting the dynamics of various environmental
parameters, will help reduce damage to nature and identify the degree of influence
of anthropogenic factors on the future of the Earth. Based on our analysis results, it
seems that the existence of global warming phenomenon has already been proven,
but its influence on different parts of the planet is heterogeneous and is not traced
in all seasons in the same way. In the spring, there is a tendency to a decrease in the
temperature of the surface layer of the Sea of Okhotsk, this tendency continues
until August, where it changes to warming condition.
Periodic fluctuations of physical parameters are also not homogeneous in space.
It is possible to identify areas where the main oscillation period is 3, 5, 6, 8 and
11 years. Using the sequential spectra method gives a relatively good estimate for
the next year’s ocean surface temperature. However, to improve the quality of the
forecast, it is necessary to improve the quality of the initial data, comparing data
from various sources and filling in the gaps associated with the presence of clouds
and ice cover, as well as improve the interpolation algorithms to avoid losing sight of
mesoscale phenomena.

Author details

Dmitry Lozhkin
Russian Federal Research Institute of Fisheries and Oceanography,
Sakhalin branch, Yuzhno-Sakhalinsk, Russian Federation

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

16
Analysis of Spatiotemporal Variability of Surface Temperature of Okhotsk Sea and Adjacent…
DOI: http://dx.doi.org/10.5772/intechopen.94918

References

[1] TeraScan System Training Guide [7] https://rda.ucar.edu/

(http://www.seaspace.com)
[8] Shershneva OV, Shevchenko GV. On
[2] Puzankov, K. L. Issledovaniye the prediction of thermal conditions
okhotomorskogo regiona pri in the Sakhalin–Kuril region
pomoshchi sputnikovoy sistemy according to satellite data. Izv. TINRO.
TeraScan [study of the Okhotsk region 2005;142:161-187
using the TeraScan satellite system]
// Sakhalinskaya molodezh’ i nauka: [9] Lozhkin DM, Shevchenko GV. Trends
Materialy 1-y mezhvuz. nauch.- in the sea surface temperature in the
prakt. konf. studentov i molodykh Sea of Okhotsk and adjacent basins
uchen. Sakh. region (March 11-12, based on the satellite data in 1998-2017.
1997). - 1997. – Issue 2, pp. 135-137. (In Izvestiya - Atmospheric and Ocean
Russian). Physics. 2019;55(9):1133-1137

[3] Tshay Zh. R., Shevchenko G. V. [10] Shatilina TA, Tsitsiashvili GS,
Otsenka ekstremal’nykh termicheskikh Radchenkova TV. Assessment of
usloviy v period neresta the statistical significance of water
tikhookeanskikh lososey u poberezh’ya temperature variability in the sea of
o-va Sakhalin [assessment of extreme Japan and the Northwest Pacific in
thermal conditions during the spawning 1982-2007, Resursy kolyuchego kraba,
period of Pacific salmon off the coast of perspektivy ispol’zovaniya i usloviya
Sakhalin Island] // Geodinamicheskiye obitaniya: Trudy Sakhalinskogo
protsessy i prirodnyye katastrofy, nauchno–issledovatel’skogo instituta
Yuzhno-Sakhalinsk, May 27-31, 2019, P. rybnogo khozyaistva i okeanografii
184. (In Russian). (Brown king crab resources, usage
prospects and habitat conditions:
[4] Lozhkin DM, Tshay ZR, Transactions of the Sakhalin
Shevchenko GV. Satellite monitoring of research Institute for Fisheries and
temperature conditions near the mouths Oceanography). Yuzhno-Sakhalinsk:
of spawning Rivers in the southern SakhNIRO. 2011;12:180-190
part of Sakhalin Island. Izvestiya -
Atmospheric and Ocean Physics. [11] Lozhkin D. M., Shevchenko G.
2019;55(9):1166-1171 V. Lineynyye trendy i tsiklicheskiye
variatsii temperatury poverkhnosti
[5] Vlasova GA, Vasil’ev a S, Okhotskogo morya i prilegayushchikh
Shevchenko GV. Prostranstvenno– akvatoriy po sputnikovym dannym
Vremennaya izmenchivost’ Struktury [linear trends and cyclical variations in
i Dinamiki Vod Okhotskogo Morya the surface temperature of the Sea of
(Spatial and Temporal Variability of the Okhotsk and adjacent waters according
Structure and Dynamics of Waters of to satellite data] // VII nauchno-
the Sea of Okhotsk). Moscow: Nauka; prakticheskaya konferentsiya molodykh
2008 uchonykh s mezhdunarodnym
uchastiyem «Sovremennyye
[6] Shershneva OV, Shevchenko GV, problemy i perspektivy razvitiya
Novinenko EG. Temperature conditions rybokhozyaystvennogo kompleksa». -
in the areas of juvenile release from 2019. – pp. 257-262. (In Russian).
salmon fish-breeding plants from
the mouth of rivers in Sakhalin [12] Lozhkin DM, Shevchenko GV. Cyclic
and the Iturup Island. Izv. TINRO. Sea surface temperature variations in
2007;150:217-225 the sea of Okhotsk and adjacent water

17
Satellite Systems - Design, Modeling, Simulation and Analysis

areas according to satellite data in 1998-

2018. Issledovaniye Zemli iz kosmosa.
2020;(1):44-51

[13] Rostov ID, Zhabin IA.

Gidrologicheskie osobennosti
priust’evoy oblasti reki Amur
[hydrological conditions of the
Amur River near-mouth area].
Meteorologiya i Gidrologiya.
1991;(7):94-99 (In Russian)

[14] Ivanov VV. Periodicheskie

kolebaniya pogody i klimata [periodic
weather and climate variations].
Uspekhi fiz. nauk. 2002;122:777-811
(In Russian)

[15] Ivanov VV. Issledovanie

variatsiy srednemesyachnoy
temperatury vozduha s pomosch’yu
posledovatel’nykh spektrov [study
of monthly mean air temperature
variations with consecutive spectra].
Meteorologiya i Gidrologiya.
2006;(5):39-45 (In Russian)

[16] http://www.sakhniro.ru/

[17] Shevchenko G., Tshay Z.,

Puzankov C. EOF and wavelet analysis
of satellite SST data in the northern
Pacific // Proceedings of the Third
PICES Workshop on the Okhotsk
Sea and Adjacent Areas. Sidney,
B.C., Canada, 2004. P. 13-18. (PICES
Scientific report; N 26, 2004).

18
Chapter

Communication Subsystems for

Satellite Design
Hung H. Nguyen and Peter S. Nguyen

Abstract

The objective of this chapter is to provide a comprehensive end-to-end over-

view of existing communication subsystems residing on both the satellite bus and
payloads. These subsystems include command and mission data handling, telemetry
and tracking, and the antenna payloads for both command, telemetry and mission
data. The function of each subsystem and the relationships to the others will be
described in detail. In addition, the recent application of software defined radio
(SDR) to advanced satellite communication system design will be looked at with
applications to satellite development, and the impacts on how SDR will affect future
satellite missions are briefly discussed.

Keywords: command and data handling subsystem (C&DHS), telemetry and

tracking, antenna payload, software defined radio (SDR)

1. Background and introduction

In the context of this chapter, a satellite is a spacecraft (SC) that orbits around a
celestial body such as the earth. A spacecraft has several design constraints placed
upon it before it can be placed in an orbit around the intended celestial body. First,
satellite designs are limited in their mass and volume to fit on the launch vehicle
that places them into orbit. Secondly, the mass and volume limits affect the size of
the power system on the spacecraft; therefore, the amount of power available to
the satellite is also limited. In addition, the space environment (thermal, radiation,
atomic oxygen, space debris, micrometeoroids, etc.) imposes constraints on the
design such as parts and material selection.
A spacecraft is consisted of two parts: the spacecraft bus and the payload (PL)
[1, 2]. The spacecraft bus provides control of the satellite and support services to the
mission payload, while the mission payload provides the mission part of the satel-
lite including payload control, mission data processing, and mission data downlink
dissemination. Examples of mission payloads (or payloads or PLs) are: scientific
instruments, remote sensing instruments, navigation service transmitters, or
communications equipment. A satellite may have one type of PL or a combination
of payload types to accomplish its mission such as navigation, remote sensing, and
communications. Shown below in Figure 1 is a typical imaging satellite used for the
remote sensing mission. Note the clear separation between the spacecraft bus that
provides solar power and maneuvering capability via thruster, while the payload
consisting of the camera and supporting communication devices such as antennas
and guidance devices such as star trackers.

1
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 1.
A typical satellite with bus and payload separation.

Figure 2.
The three main segments for satellite system.

Regardless of the mission type1 and the payload that a spacecraft carries, a sub-
system that must exist in all satellites is the communication subsystem that enables
the spacecraft to communicate with the ground stations that control the satellite and
to deliver the data that the mission requires. This chapter focuses on architecture and
functionalities of the communications subsystem that usually resides on the satellite.

1
Mission type has different meaning depending on context. For example, U.S. military satellite has
three basic mission types: imaging/sensing; communications; positioning, navigation and timing (PNT)
missions. For NASA space exploration, they are near-earth and deep-space missions.

2
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

There are three specific segments shown in Figure 2 below that must work
together for the larger overall system to provide communication, navigation, or any
other type of missions:

• The space segment consisting of all satellites and associated equipment

required for the mission applications and the launch vehicles used to deliver
those satellites to orbit.

• The satellite control (or control) segment consisting of all the personnel, facili-
ties, and equipment that are used to monitor and control all the assets in space.
Practically, the control segment is also referred to as satellite ground segment
because it is usually located on the ground.

• The user segment consisting of all the individuals and groups who use and
benefit from the data and services provided by the payloads of the satellite and
the equipment that allows this use.

2. Typical satellite major subsystems

In general, the space mission dictates the type of orbit2, satellite design and its
expected life cycle, and its operational scenarios. The PL design includes dimen-
sions, interfaces, weight, physical characteristics, and basic utility needs (e.g.,
power consumption), which usually influences spacecraft (SC) bus design. The PL
is often a unique and one-of-a-kind design tailored to meet specific mission require-
ments, frequently relying heavily on newer technology, while the satellite bus has
the supporting function, and as such relies largely on existing or modified hardware
such as batteries, inertial devices, and star trackers. Since PLs and their missions
vary widely, so is this satellite bus supporting role.
Traditionally, the PL is considered a subsystem of the satellite bus that is
designed to generally satisfy the corresponding mission requirements. The PL
operational requirements sometimes impose specific requirements on the satellite
bus that must be satisfied for the PL to accomplish its mission. This interdepen-
dence between satellite bus and PL subsystems has historically resulted in many
nonstandard interfaces developed and implemented by the incumbent spacecraft
builders. As a result, the aerospace industry has been moving toward a more
standardized and commodity satellite bus framework that can potentially result in a
tremendous cost saving approach.
As shown in Figure 3 below, a satellite bus typically consists of the following
subsystems: command and data handling subsystem (C&DHS); communications
subsystem (CS); electrical power subsystem (EPS); propulsion subsystem (PS);
thermal control subsystem (TCS); attitude control subsystem (ACS) also known as
guidance, navigation and control (GNC) subsystem; structures and mechanics sub-
system (S&MS); and life support subsystem for manned missions if required. The
C&DHS will be described in detail below. The CS provides the satellite bus with the
necessary communication functionalities to connect the user and ground segments
to different satellite subsystems. The EPS provides the electrical power generation
and distribution for various spacecraft subsystems. The PS provides maneuvers

2
There are three main types of satellite orbits: low earth orbit (LEO) of 2000 km in altitude or less;
geostationary (GEO) with altitude around 35,786 km; and medium earth orbit (MEO) with altitude
between LEO and GEO.

3
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 3.
A typical satellite bus and payload subsystem.

necessary for altitude, inclination adjustment, and momentum management

adjustments. The TCS provides active thermal control from electrical heaters and
actuators to control temperature ranges of equipment within specific ranges. The
ACS provides proper pointing directions for the satellite subsystems, such as sun
pointing for EPS to the solar arrays and earth pointing for CS. The S&MS provides
the necessary mechanical structure to withstand launch loads by the launch vehicle,
during orbital maneuvers, as well as loads imparted by entry into the atmosphere of
earth or another planetary body.
On the other hand, a PL is tailored to a specific mission type. For example, a
remote sensing satellite can have as its payload an electro-optical (EO) camera to
take day-time pictures of the earth and then convert them to electrical signals that
can be captured. Alternatively, the camera may also have infra-red (IR) sensors
that enable the PL to see the earth at night, or microwave sensors that will let the
PL “see” radio frequency (RF) signals from the earth at several radio frequencies

Figure 4.
A typical and generic sensor payload.

4
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

(RFs). These sensors can be classified as passive or active, and each of them can
be further classified as imaging or sounding3. Figure 4 below illustrates a generic
imaging PL that will convert the sensor analog data into electrical signals that can
be captured and transmitted to a ground station. Note the existence of a communi-
cation subsystem as part of this imaging payload.

3. Typical communications subsystems for satellite

In this section, the different typical modules of a satellite communication sub-

system are discussed. In addition, the command and data handling subsystem, and
command, telemetry and mission data processing subsystem will also be described
in detail.

3.1 Antenna and RF front-end/back-end subsystems

At the physical layer, the communications subsystem starts with an antenna and
the RF front-end transceiver. The antenna is the most important component of the
communications subsystem where the electromagnetic (EM) signals are originated
or received. The RF front-end/back-end is where the EM signal is being down/up-
converted to baseband/RF signal to be demodulated/modulated for baseband signal
recovery or downlink transmission, respectively. Figure 5 below depicts a typical
transmitter and receiver (transceiver) chain with the modulation and demodulation
(MODEM), followed by the RF front-end and the antennas. The baseband commu-
nications function is carried out by the MODEM, whereas the RF portion is handled
in the transceiver, RF front-end, and antenna sections.
Modulation is the name given to the process of impressing the wanted signal to
be transported onto a radio frequency (RF) carrier, which is then conveyed over the
satellite link and demodulated at the receiving terminal to extract the wanted signal
from the carrier. Thus, modulation translates a baseband spectrum (at zero fre-
quency) to a carrier spectrum (at RF range) and demodulation is the process of recov-
ering the data at the receiver end of the link. Thus, the process requires a modulator

Figure 5.
Typical RF front-end chain.

3
In this context, sounding means sending a radio signal and interpreting the results from the returned
signals. Examples are radar and ranging signals.

5
Satellite Systems - Design, Modeling, Simulation and Analysis

and a demodulator, collectively known as a MODEM. The input to the modulator may
require some initial processing such as filtering and amplitude limiting.
Before the RF signal is sent to the antenna, a traveling wave tube amplifier
(TWTA) or solid-state power amplifier (SSPA) is needed to amplify the RF signal
to a desired level for transmission. Conversely, after the RF signal is received by the
antenna, a low noise amplifier (LNA) is needed to ensure that the received signal is
brought up to the desired signal level with minimum noise before demodulation.
In addition to being lighter than TWTA, the achievable power efficiency for
SSPAs is a major factor to support transmit phased arrays. Currently, the tube-based
TWTA implementations are still the most cost-effective design, even though both
options might be viable for lower power systems.
In increasing technical maturation over the years, the following types of space-
craft antennas have been used for satellite communications:

1. Low-gain omni and squinted-beam antennas for large earth coverage.

2. Increased gain types of satellite antennas (horn type and helix antennas) for
medium earth coverage.

3. Parabolic reflectors, including multi-beam antennas with multiple feed sys-

tems for multiple user and small area coverage.

4. Deployable antennas, particularly to achieve more highly focused beams and

support much high-gain multi-beam antennas.

5. Phased array feed and phased array antennas for scanning and hopping beams.

6. Optical communications systems, which have been used for intersatellite links
and interplanetary communications, and increasingly being considered for
earth-to-space systems.

In general, there are many different types of antennas, but the one most com-
monly associated with satellite communications is the parabolic dish antenna. These
dish antennas have a narrow beam width, concentrating the energy of the radiated
main beam into a smaller solid angle. This means more of the radiated energy
reaches, or “illuminates,” the satellite when using a dish antenna as compared to an
omnidirectional, or “omni” for short, antenna. An example of dish antenna used on
satellite is shown below in Figure 6 for a Ku-band space to ground antenna (SGANT)
mounted on the external stowage platform of the International Space Station (ISS).
There are several factors driving the design and development of satellite antennas.
These include the need to reuse frequency bands because of limited spectrum alloca-
tions; the need to have antennas that can operate at higher frequencies with higher
bandwidth; and the desire to deploy higher gain antennas at the same time minimizing
the required size, weight, and power (SWAP) constrains. In practice, there are sub-
stantially more SWAP constrains for satellite antennas than on the ground stations, and
this results in several design trade-offs between the space and control/user segments.
For example, the GEO orbit allows a high gain antenna to be pointed at a satellite
with a minimum of tracking. Thus, a large dish can be used and remain virtually
stationary without tracking a satellite as it moves around in its orbit. On the other
hand, a low earth orbit (LEO) satellite that can cross from horizon to horizon in a
few seconds can result in ground antenna installations that can be quite complex
and expensive. Consequently, trade-offs need to be made to support the mission
parameters of the whole satellite network.

6
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

Figure 6.
Example of a satellite dish antenna.

3.2 Command and data handling subsystem

The term “command and data handling subsystem” (C&DHS) was referred to as
“On-board Computer” (OBC), which is a legacy of the past in which many satellite
functions were performed by analog circuits with the help of an OBC. With the
current shift toward the digital domain, the term OBC does not fully cover the topic
anymore thus C&DHS is being used instead. An appropriate analogy to describe the
C&DHS subsystem is to regard it as the brain and nervous system of the spacecraft.
The function of a C&DHS subsystem is to perform onboard processing and
operations and internal communication [3, 4]. The task of managing the opera-
tions of the spacecraft subsystems is nowadays performed mostly by software in
an autonomous manner and is generally categorized as onboard operations. The
software is also responsible for preparing the data to be downlinked and handling
any commands that are received from satellite operators on the ground. Lastly, the
C&DHS facilitates and controls all internal communications (consisting of com-
mands, telemetry, and tracking data) between the different satellite subsystems.
The basic functions of the C&DHS can be summarized below:

1. Receives commands from the command or user segment through the telem-
etry, tracking, and control (TT&C) subsystem.

2. Decodes, executes, and/or distributes those commands to/from the onboard

computer.

3. Collects and formats telemetry data from all space vehicle (SV) units.

4. Distributes telemetry for downlinking. Provides a platform for bus flight soft-
ware (FSW).

5. Additional functions include ranging processing for satellite tracking purpose,

satellite timekeeping, computer health monitoring (watchdog), and security
interfaces.

An overview of the architecture of C&DHS in a typical satellite is provided in

Figure 7 below. In this figure, all components are connected to each other via a

7
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 7.
Block diagram of a typical command and data handling subsystem.

common low-speed data bus in red color, typically compliant with MIL-STD 1553
or other standards. Also shown is the data connection in blue from the C&DHS to
other components, which is more customized and high-speed in nature depending
on the design.
The heart of the system is the C&DHS’ onboard computer (or OBC) that runs
the software responsible for managing the onboard operations. The OBC is tightly
linked to the electrical power subsystem (EPS). The main reason is the importance
of the available and consumed power for managing onboard spacecraft operations.
For instance, by continuously querying the EPS on the available power, the OBC
can decide to turn off non-critical subsystems to prevent vital systems from shut-
ting down from lack of power. Secondly, the OBC must be able to command the
EPS to disable or enable different subsystems throughout the various phases of the
mission. Since the amount of transmitted data between these two subsystems is
small, a low-speed data link is sufficient, although there is a new trend to incorpo-
rate high-speed standard link such as SpaceWire4 to satisfy increasing demand for
data volume.
The OBC is also responsible for receiving, interpreting, and executing com-
mands from ground operators via the radio receiver. Using low-speed radio
transmitters, the OBC also sends packets of housekeeping data, or telemetry, to the
ground station. The purpose of the housekeeping data is to give the operators on the
ground an overview of the spacecraft health and its general condition.
Some small satellites only have a single low-speed transmitter, so the housekeep-
ing and payload data are combined over the same link. For larger satellites with
payloads capable of producing vast amounts of data, a dedicated high-speed data
link is used to store the data on an onboard storage system. When the satellites pass
over a ground station, the OBC commands the high-speed radio transmitter to
retrieve and transmit the previously stored payload data through another dedicated
high-speed link from the onboard storage system. This approach frees the OBC
from having to process large amounts of data and allows it to devote its internal

4
SpaceWire is a spacecraft communication network based in part on the IEEE 1355 standard of com-
munications. It is coordinated by the European Space Agency (ESA) in collaboration with international
space agencies including NASA, JAXA, and RKA

8
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

resources for time critical operations and communicates with the PL and all other
subsystems through the low-speed data links. This would include the requirements
to retrieve information on the health, perform critical interventions as well as to
command these subsystems to perform various actions according to the operational
arrangement of the mission.

3.3 Typical command and telemetry processing subsystem

The telemetry, tracking, and control (TT&C) subsystem of a satellite provides a

connection between the satellite (space segment) and the ground facilities (control
or user segment). The purpose of the TT&C function is to ensure the satellite
performs correctly. As part of the satellite bus, the TT&C subsystem is required for
all satellites regardless of the mission type. The TT&C subsystem has three specific
tasks that must be performed to ensure a successful mission:

1. Telemetry: the collection, processing of health, and status data of all spacecraft
subsystems, and the transmission of these data to the control segment on the
ground. This requires not only a telemetry system on the spacecraft but also a
global network of ground stations around the world, unless the satellite space
network includes intersatellite links that can relay the data to designated satel-
lite and downlink to the appropriate ground station. Figure 8 below illustrates
the processing of telemetry data by the C&DHS. Here the different health
information and status information sent from various subsystems are col-
lected by the telemetry input interface, fed to the C&DHS processor, buffered,
encrypted, and sent down to the ground station.

2. Tracking: the determination of the satellite’s exact location by the control

segment and where it is going via the reception, processing, and transmitting
of ranging signals. This requires a ranging system on the spacecraft and a data
collection ground network for this tracking function to work.

3. Command and control: the reception and processing of commands for contin-
uous operation of the satellite. Usually a ground system is required, although
advanced spacecraft designs have evolved toward “autonomous operations”
so that many of the control functions can be automated onboard and do not
require ground intervention except under emergency conditions. A typical
command processing scenario is illustrated in Figure 9 where serial command
bit stream from the command receiver is received by the command input inter-
face, where the relevant commands are extracted and sent to the appropriate
subsystems via a serial or parallel interface.

Figure 8.
Telemetry processing by C&DHS.

9
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 9.
Command and control message processing by C&DHS.

3.4 Typical mission data processing subsystem for communication applications

For communications payload, the onboard switching systems are designed to

make more efficient use of a satellite communication network, especially those
that employ multi-beam technology that entails onboard switching to interconnect
uplink and downlink beams with a high degree of efficiency.
Figure 10 below summarizes the functional block diagram of a channelized
transponder processor assuming a digital implementation of the channelized tran-
sponder filtering and switching function. Any signal within the receiver bandwidth
is down-converted to an intermediate frequency (IF) or baseband and digitally
sampled. These samples are digitally filtered, stored, and routed to the switch port
corresponding to the desired downlink beam. This routing is achieved by a simple
readdressing of the stored digital samples within a common output buffer memory
or by a more traditional digital switch implementation.
For most sensing payload and as shown in Figure 4 above, the sensor analog
data are collected onboard, digitized, buffered if necessary, and transmitted down
to ground station for processing. This is due to the complexity of sensing mission

Figure 10.
Channelized processor for communications payload.

10
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

data processing and the lack of onboard computational power to accomplish these
tasks. An example of onboard PL processing for passive electro-optical (EO) remote
sensing is shown in Figure 11 below, where the reflected light from earth is passing
through a combination of optical lenses and charge coupled device5 (CCD) whose
output is an analog signal that would be conditioned by analog filters before being
digitized, compressed, and sent down via a mission data downlink to the ground
station for processing. There, the data are decompressed, and image is enhanced by
appropriate algorithms and displayed for users.
Typical data volume collected by sensing payload is large, and peak rates can
produce data at much higher speeds than TT&C; thus, a separate downlink for
mission data is needed. Depending on the system, this mission data downlink to a
ground station can either be performed using a dedicated mission direct downlink,
or indirectly via a relay broadband communications satellite. Sensing satellite can be
positioned in GEO, MEO, or LEO orbits, and can have many possible mission data
downlink architectures based on mission requirements. For example, a LEO sensing
satellite can either buffer its mission data until within view of a dedicated ground
station for downlink, or it can forward its mission data to a relay satellite that can
ensure that the mission data can be downlinked to a designated ground station.
Another example of active remote sensing is a synthetic aperture radar (SAR)
mission, where returned radar signals are collected onboard and sent to the ground to
be correlated and form an image of the ground surface. This type of remote sensing
does not heavily depend on sun light and other weather affects. Applications for SAR
include agriculture, geology, geohazards, ice, oil spills, and flood monitoring. Several
emerging applications such as forestry, ship detection, and others are possible [1].
An example of a SAR mission is the NASA-ISRO Synthetic Aperture Radar (NISAR)
[5], which is a collaborative earth-science mission between NASA and the Indian

Figure 11.
Onboard image processing for an EO application.

5
A CCD is an integrated circuit etched onto a silicon surface to form light-sensitive elements repre-
sented by what are called pixels. Photons incident on this surface generate a charge that can be read by
electronics as a voltage and turned into a digital copy of the light patterns falling on the device.

11
Satellite Systems - Design, Modeling, Simulation and Analysis

Space Research Organization (ISRO). The sensing payload features an L-band SAR
instrument and an S-band SAR instrument. The simultaneous dual-frequency radar
system at peak rates will produce data at gigabit-per second speeds, which drives
the data-volume requirements at a minimum of 35 Terabits per day of radar science
data to the ground. This is a direct mission downlink system with three designated
ground stations. The payload communication system uses a 70-cm high-gain antenna
with two synchronized transmitters in a dual-polarization configuration with each
transmitter providing 2.4 Gbps of coded data with an aggregate rate of 4.8 Gbps.

4. Software defined radio (SDR) and applications to satellite

Traditional communications systems are designed for and constrained to a spe-

cific waveform(s) operating over predetermined frequencies, bandwidths, and signal
modulation types. This paradigm works well when the requirements and constraints
of the communication link and network protocol are well understood prior to design.
As a result, most radios in today’s world have very dedicated uses. A car key fob
is designed only to unlock or lock your car door, while a smart phone radio connects
to the Internet through various wireless communication protocols. Although these
examples vary in complexity of the hardware, they both cannot operate outside the
confines of their physical layer implementation. Consequently, RF hardware with a
narrow focus is not suitable for applications with a broader communication scope.
A single software defined radio (SDR) with a flexible RF front-end combined
with modern computing power can be used for the above applications plus more.
In addition, a radio with a flexible hardware and software architecture can also lead
to more innovation in the communications industry. Because of the rapid develop-
ment nature of software, an engineer or researcher can experiment with novel ideas
and SDR waveforms that would not be achievable with a traditional radio.
SDR in the satellite communications industry has become a growing trend,
particularly in the commercial and defense industries. In the following section, an
overview of SDR will be given and applications of SDR in satellite communications
will be discussed.

4.1 Overview of SDR

Before going into SDR basics, some of the SDR advantages are [6]:

• Interoperability: an SDR can seamlessly communicate with incompatible

radios, or work as a bridge between them. For example, different branches
of the military and law enforcement can use many incompatible radios, thus
hindering communications during joint operations. A single multichannel SDR
can work with all these different radios and provide interoperability.

• Efficient use of resources under varying conditions: for example, a low-power

waveform can be selected if the radio is running low on battery, while a high-
throughput waveform can be used to quickly download a file. This flexibility is
one of the first reasons why SDR became popular.

• Opportunistic frequency reuse in SDR using cognitive radio6 (CR) technol-

ogy: if the “owner” (or primary user) of a spectrum band is not using it,

6
A cognitive radio (CR) is a radio that can be programmed and configured dynamically to use the best
wireless channels in its vicinity to avoid user interference and congestion.

12
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

an SDR-CR can “borrow” the spectrum until the owner comes back. This
technique has the potential to dramatically increase efficient use of radio
frequency spectrum.

• Reduced obsolescence: an SDR can be field upgraded to support the latest com-
munications standards. This capability is especially important to radio with
long life cycles such as those in satellite communications.

• Lower cost: a single SDR can be adapted for use in multiple markets and for
multiple applications. For example, a single radio can be sold to cell phone and
automobile manufacturers to significantly reduce cost.

• Research and development: SDR can be used to implement many different

advanced waveforms, e.g., code division multiplexing access (CDMA) or
orthogonal frequency division multiplexing (OFDM), for real-time perfor-
mance analysis. Performance studies can be conducted much faster and often
with higher fidelity than simulations.

On the other hand, some of the disadvantages for SDR are:

• Cost is the most common argument against SDR. A single key fob is based on
a very inexpensive ASIC7; however SDR is heavily reliant on FPGA,8 which is
much more expensive. This is even more significant for high-volume, low-
margin consumer products.

• The second most common argument against SDR is increased power consump-
tion with increased DSP complexity and higher mixed-signal/RF bandwidth.
Power consumption in an FPGA or GPP for flexible signal processing can easily
be 10 times higher than in ASIC. Also, wideband analog-to-digital converters
(ADCs), digital-to-analog converters (DACs), and RF front-ends consume
more power than their narrowband equivalents.

• Increased time and cost to implement the radio: it can take much more engi-
neering effort to develop software/firmware for multiple waveforms than for
one, especially if it must be compliant with a military standard such as JTRS9.

• Changing specifications and requirements: this usually happens when the SDR
design must support not only a set of baseline waveforms but also anticipate
additional waveforms.

• Increased schedule risks: since SDR is still a relatively new technology, it is

more difficult to anticipate schedule problems. Also, it is difficult to thor-
oughly test the radio in all the supported and anticipated modes.

• Limited technical scope: SDR only addresses the physical layer and will require
cooperation from upper layers for throughput improvements.

7
Application-specific integrated circuit (ASIC) is designed for specific purpose.
8
Field Programmable Gate Array (FPGA) is an integrated circuit designed to be configured by a
designer after manufacturing thus much more general purpose.
9
The Joint Tactical Radio System (JTRS) aims to replace existing radios in the U.S. military with a single
SDR to enable new frequencies and waveforms added via software or firmware upload.

13
Satellite Systems - Design, Modeling, Simulation and Analysis

4.1.1 SDR basics

The general definition for a SDR is a radio with some or all its physical layer behav-
ior defined through means of software [7, 8]. SDRs are incredibly valuable devices as
they allow the end user the ability to traverse the RF spectrum at variable sampling
rates. The fundamental qualities that make up an SDR are the flexible specifications
and the ability to transform the analog signal using digital signal processing (DSP).
A radio can be categorically separated into receivers and transmitters. For this
section, the receiver implementation will be considered as it is generally more
interesting and complex. A block diagram of an SDR receiver is shown below in
Figure 12. The following sections will present the anatomy of the SDR that differ-
entiates it from a traditionally designed radio.

4.1.2 RF front-end

The purpose of the RF front-end (RFFE) is to isolate the desired signal received
by the antenna from interference signals. To achieve this, the signal of interest must
be brought down to lower frequency for digital conversion while mitigating the side
effects from filtering during the frequency conversion process. A flexible RFFE
for SDR must be designed so that the frequency and bandwidth are controllable by
software. Depending on the system requirements and the available RF component
specifications, there are several ways to achieve this.
One of the most common RFFE designs for analog radios is the heterodyne
receiver. A heterodyne receiver, shown in Figure 13 below, works by mixing down
the received signal from its carrier frequency to a lower intermediate frequency
(IF). The signal at IF can now be more conveniently filtered, amplified, and pro-
cessed. A super-heterodyne receiver uses a fixed IF that is lower than the carrier
frequency but higher than the signal bandwidth and often uses two stages of down
conversion to reduce the filtering requirements at each stage.
Another popular RF front-end architecture generally used for low-power
applications is called zero-IF. A zero-IF receiver, shown in Figure 14 below, uses a
single mixing stage with the local oscillator (LO) set directly to the desired carrier
frequency to convert directly to baseband in-phase and quadrature signals. Because
mixers tend to have high power consumption and only low-pass filters are required,
the simpler zero-IF provides improved power efficiency over a heterodyne architec-
ture. However, the zero-IF implementation is more susceptible to IQ imbalances of
the in-phase and quadrature oscillators, which will produce anomalies in the signal
constellation. LO leakage may also self-mix through the RF ports creating a large
DC bias. Both issues can be corrected using digital signal processing.

Figure 12.
A block diagram of an SDR.

14
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

Figure 13.
Heterodyne receiver.

Figure 14.
Zero-IF receiver.

The analog-to-digital converter (ADC) is responsible for converting a continu-

ous-time signal to a discrete-time one. To translate signals from the analog to digital
domain, an ADC must perform two fundamental steps: sampling and quantization.
Sampling is the process of reading voltages at discrete-time intervals. Quantization
is the process of converting these voltage readings into binary outputs. ADC perfor-
mance can be evaluated based on various parameters, such as: signal-to-noise ratio
(SNR), dynamic range, bit resolution, sampling rate, and power dissipation. The
ADC dictates the DSP limitations of the SDR. Generally, the sampling rate should
be at least twice the desired bandwidth of your signal. The ADC should be chosen to
match the capability of your processor and specifications of the signals of interest.

4.1.3 Digital front-end (DFE)

The two main functions of a digital front-end are sample rate conversion (SRC)
and channelization. Once a signal has become digitally converted, the samples
need to be further primed for digital processing. Operating the ADC at a fixed rate
simplifies its clock generation; however, it may be necessary to convert the sampling
rate to match the sampling rate required to demodulate certain waveforms. Most
wireless signals generally operate with specific symbol or chip rates that are speci-
fied by their respective standard. Depending on the RFFE design and signal type,
channelization may be required to select the channel of interest.
SRC represents a classic sampling theorem problem. Converting sampling rates
can introduce undesirable effects such as aliasing, an effect that causes frequency
components to overlap. SRC can be achieved digitally through the processes of deci-
mation and interpolation. To mitigate aliasing, decimation is performed by using an
anti-aliasing filter followed by subsampling, which is essentially removing samples
at certain intervals. Interpolation is a method of calculating values to add values
in between samples. Channelization works by using digital down conversion, the
process of digitally mixing down a signal to baseband with a numerically controlled
oscillator.

15
Satellite Systems - Design, Modeling, Simulation and Analysis

4.1.4 Digital signal processing

SDRs have an array of devices to choose from for the required DSP application,
each with their own strengths and weaknesses. An SDR may integrate multiple pro-
cessor types and partition the signal processing chain to optimize each processor.
The following criteria should be considered when evaluating the various processor
types: flexibility, modularity, and performance. The three digital hardware choices
this section will consider are the general-purpose processor (GPP), digital signal
processor (DSP), and the field programmable gate array (FPGA).
A GPP is the typical microprocessor designed to handle a wide variety of generic
tasks that can be found in your everyday personal computer. They are generally
designed to have large instruction sets and highly capable of implementing and
performing complex arithmetic tasks such as modulation/demodulation, filtering,
fixed/floating point math, and encoding/decoding. Some commonly used GPP
architectures are x86/64 and Advanced RISC Machine (ARM). The advantage of
using a GPP is the wide availability, flexibility, and ease of programmability. Several
GPP-based SDRs, such as Universal Software Radio Peripheral (USRP) and the
LimeSDR, operate by digitizing the baseband signal and performing the required
digital signal processing on computers. These types of SDRs are popular among uni-
versity researchers and hobbyists due to the relative ease of obtaining and develop-
ing their applications.
Because the GPP was designed with such a broad focus, latency, speed, and
power efficiency may be a limiting factor depending on the application. Many wire-
less communication standards have strict real-time and large processing bandwidth
requirements that most modern CPUs cannot meet due to processor architecture
and operating system design. .
A DSP is a microprocessor optimized for digital signal processing applications
with the ability to be programmed with high-level languages. Although a GPP can
contain much of the same functionality, the DSP performs the same digital signal
processing operations more quickly and efficiently due to its reduced instruction set
computer (RISC) architecture and parallel processing. The reduced instruction set
limits the essentials but contains optimizations for common DSP operations such as
multiply accumulate (MAC), filtering, matrix operations, and fast Fourier transform
(FFT). DSPs are commonly sold in two variants: optimized for power efficiency and
optimized for performance; and are used in applications such as base stations and
edge devices. Power consumption is also minimized by reducing the silicon footprint
that would be in GPPs sophisticated cache and peripheral subsystems.
Although DSPs have been commonly deployed in the past decades, they serve as a
middle ground between GPPs and FPGAs with regard to flexibility, performance and
efficiency. Field-programmable gate array (FPGA) offers more parallelism, higher
data rates, and better power efficiency than DSP, but is not well suited for control
applications, such as implementing the network/protocol stack. This is due to the
limited amount of memory in FPGA and for this reason it is often paired with GPP.
A FPGA is an array of programmable hardware logic blocks, such as general
logic, memory, and multiplier blocks, that are wired together via a reconfigu-
rable interconnect to generate an integrated circuit for several designs with
the ability to quickly switch between configurations. FPGA configurations are
programmed using hardware description language (HDL), which is also used for
ASIC. Because a FPGA functionality is defined at the hardware level and can be
implemented using parallelism, it can perform DSP algorithms at much higher
rates than DSPs and GPPs. FPGA consumes more power and requires more space
than ASICs but provides more programmability and flexibility than ASIC. A big
consideration for using FPGAs for SDR is the domain knowledge requirement

16
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

for developers. Developing on FPGAs can be time consuming and require an

extensive understanding of the target hardware architecture.
When the system requirements exceed the capabilities of a singular processor
type, a comprehensive solution may include a combination of the above processor
types. A common processing architecture in the defense industry comprises of a
FPGA, DSP, and GPP. In this paradigm, the FPGA is responsible for high data rate
signal processing tasks, such as sampling and filtering, the DSP handles demodula-
tion and protocol, and the GPP performs control-related tasks, such as the user inter-
face and algorithmic processing. Implementing such a system can become a complex
management task to coordinate the processing flow; however, the system can benefit
greatly by optimizing overall performance based on the strength of each processor.

4.2 Applications of SDR to satellites

For space applications, SDR has unique challenges such as extreme radiation and
temperature environment, autonomous operational requirements, limitations on
size, weight and power (SWAP), and the need for reduced development time and
increased reliability in agile prototyping. In this section, recent applications of soft-
ware defined radio to satellite, as well as the current status of radiation-hardened
SDR components, are presented.

4.2.1 NASA STRS

Recognizing early on that a standard and open architecture is needed to encour-

age reuse and portability of software, NASA developed an open architecture speci-
fication for space and ground SDRs called the Space Telecommunications Radio
System (STRS) [9]. From this standard, several compliant systems have been built
and demonstrated in radios on the International Space Station (ISS) and several
ground stations. It was also the intention of NASA that the STRS architecture should
be used as baseline for many future NASA space communications technologies.
In a nutshell, the STRS standard consists of hardware, configurable hardware
design, and software architectures with accompanying description, guidance,
and requirements. The three main hardware functionalities are connected by the
Hardware Interface Description10 (HID) and described and shown in Figure 15 below:

1. General processing module (GPM) consists of the general-purpose processor;

appropriate memory; spacecraft bus (e.g., MILSTD-1553, Space Wire); inter-
connection bus (e.g., PCI); and the components to support the configuration
of the radio.

2. Signal processing module (SPM) where signal processing is used to handle the
transformation of digital signals into data packets. Its components include ASICs,
FPGAs, DSPs, memory, and connection fabric/bus (e.g., PCI, flex-fabric).

3. RF module (RFM) handles the RF functionality to transmit/receive the ap-

propriate digital signal. Its components include RF switches, digital-to-analog
converter (DAC), analog-to-digital converter (ADC), diplexer, filters, low-
noise amplifiers (LNAs), and power amplifiers (PAs).

10
In a HID, the hardware architecture requirements are written so that the hardware provider defines
the functional modules of the system and publishes the functions and interfaces for each module and for
the entire STRS platform [Wikipedia].

17
Satellite Systems - Design, Modeling, Simulation and Analysis

Figure 15.
NASA STRS’ three main hardware functionalities.

In STRS terminology, software includes source code, object code, executables,

etc. implemented on a processor. As shown in Figure 16, the STRS software
architecture uses three primary interfaces: the STRS APIs, STRS hardware abstrac-
tion layer11 (HAL) specification, and the Portable Operating System Interface12
(POSIX®). The STRS APIs provide the interfaces that allow applications to be
instantiated and use platform services.
Configurable hardware designs are the items and designs, such as hardware
description language (HDL) source, loadable files, data tables, etc., implemented in
a configurable hardware device such as a FPGA.
STRS encourages the development of applications that are modular, portable,
reconfigurable, and reusable. The STRS software, configurable hardware design,
metadata, documentation for STRS applications, STRS devices, and operating
environments (OEs) are submitted to NASA STRS Application Repository to allow
applications to be reused in the future with appropriate release agreements.

4.2.2 SDR applications in CubeSats

CubeSats13 are increasingly popular spacecraft platforms for mission-oriented

experiments that can be accomplished via quick prototyping and launches [10–12].

11
Hardware abstraction layer (HAL) is a layer of programming that allows a computer operating system
to interact with a hardware device at a general or abstract level rather than at a detailed hardware level
[Wikipedia].
12
The Portable Operating System Interface (POSIX) is a family of standards specified by the IEEE
Computer Society for maintaining compatibility between operating systems. POSIX defines the applica-
tion programming interface (API) for software compatibility with variants of Unix and other operating
systems [Wikipedia].
13
CubeSats are a class of Small Satellites (SmallSats) weighing between 1 kg and 10 kgs that use stan-
dard size and form factor of 1 U (one unit) of 10 cm x 10 cm x 10 cm. A standard 3 U CubeSat (10 cm
x 10 cm x 34 cm, 5 and 6 kg) has been demonstrated to support real mission, and larger CubeSats (6 U,
12 U, and 24 U) are developed.

18
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

Figure 16.
STRS software architecture layers.

This short development timeline is due to the use of commercial-off-the-shelf

(COTS) technology that typically has limited resilience to the space environment.
Therefore, CubeSat usage has largely been limited to experiments or applications
where high availability is not the main objective.
In general, SDR technology will allow for on-orbit flexibility via reconfigurabil-
ity of data management, protocols, multiple access methods, waveforms, and data
protection. SDR processing requirements are inherently scaled to the application.
The availability of modular, high-performance sequential and parallel processors
that are resilient to radiation upsets allows the tailoring of hardware architectures to
the application and to the CubeSat platform. This is especially suitable for mis-
sions that require the flexibility to support multiple TT&C and mission data from
multiple satellites and ground stations [13–15].
Given the provided mission flexibility, implementing an SDR on a CubeSat
could significantly increase the required processing capacity and thus the size,
weight, power and cost (SWAP-C) of the SDR implementation. Consequently, most
current CubeSat SDR design and implementation are still customized depending
on the mission requirements. In [16], some of the current COTS SDR hardware
and software platforms such as GomSpace, Ettus Research USRP, EPIQ Solutions,
Lime Microsystems, FunCube, and RTL SDR are described and categorized in
decreasing cost and mass to illustrate the heterogeneous nature of SDR in CubeSat
applications. Also described are a number of space and ground segment systems
built to be (or have been) launched using these COTS SDRs or components thereof.
What would be needed is a standard for CubeSat SDR similar to NASA STRS to
ensure that hardware and software reuse can be incorporated into future CubeSat
developments.

4.2.3 SDR applications in emitter location from space

A pioneering commercial application of SDR in space is the HawkEye 360

(HE360) system [17] that was launched on 3 December 2018. HE360 system
consists of three identical spacecrafts and their primary payload is a SDR with

19
Satellite Systems - Design, Modeling, Simulation and Analysis

custom RF front-end along with VHF Ku-band antennas. This Pathfinder mission14
was to enable onboard reception and geolocation of different types of terrestrial RF
signals using signal processing technique to combine received data from all three
payloads15.
One commercial application of this mission is the detection and geolocation
of a maritime vessel’s automatic identification system (AIS), which broadcasts the
locations generated by GPS-enabled receiver. The locations generated by AIS can be
disabled or spoofed, therefore not reliable. Another application would be to allow
regulators, telecommunications companies, and broadcasters to globally monitor
spectrum usage and identify areas of interference. The system can also be used to
help large area search and rescue operations by quickly locating activated emer-
gency beacons.
The SDR developed for the Pathfinder payload consists of an embedded pro-
cessor system and three baseband processors. The baseband processor was built
around the Analog Devices 9361 (AD9361) System on Chip (SoC) product, which
is a highly integrated RF transceiver that combines high-speed ADCs and DACs,
RF amplifiers, filtering, switching plus more. The HE360 payload supported up to
three receiver channels (one AD9361 per channel) that can be simultaneously pro-
cessed on separate frequencies. In addition, the signal processing subsystem takes
advantage of open-source software and firmware code to allow system development
to proceed without knowing the final space hardware. GNURadio16 was selected
for being a free and open-source toolkit for SDR and widely used in small space
projects for ground software processing.

4.2.4 Radiation hardening and its current status on SDR hardware

In space, most semiconductor electronic components are susceptible to radia-

tion damage, thus radiation-hardened (or rad-hard) components are required and
normally developed based on their COTS equivalents with variations in design
and manufacturing17 to reduce the susceptibility to radiation. Consequently,
rad-hard components tend to lag behind most recent COTS developments.
Depending on mission requirements, rad-hard products are typically selected and
tested using popular metrics such as total ionizing dose18 (TID), and single event
effects19 (SEEs).

14
A Pathfinder mission is usually a demonstration to prove that a system can successfully achieve a
specific objective before a full mission can deploy.
15
By comparing time-of-arrival (TOA) and frequency-of-arrival (FOA) measurements between pairs of
receivers, the position of a signal can be computed.
16
GNU is a recursive acronym for “GNUs Not Unix!“ chosen because GNU’s design is Unix-like but
differs from Unix by being free software and containing no Unix code.
17
Design trade-offs choosing more better radiation tolerance integrated circuit technology, low-power
Schottky vs. Emitter Couple Logic (ECL) vs. bipolar over CMOS. Rad-hard chips are often manufactured
on insulating substrates (silicon on insulator, silicon on sapphire, silicon carbide, gallium nitride) instead
of the usual semiconductor wafers. Shielding the package against radioactivity to reduce exposure of the
bare device.
18
TID causes slow gradual degradation of the device’s performance and is measured in rads.
19
SEEs are caused by a single energy particle and can be (a) non-destructive Single Event Upset (SEU)
causing transient pulses in logic or support circuitry or as bitflips in memory cells or registers, (b) poten-
tially destructive Single Event Latch-up (SEL) that may be cleared by a power reset, and (c) destructive
Single Event Burnout (SEB) or Single Event Gate Rupture (SEGR), which is irreversible.

20
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

Per US DoD MIL-PRF-38535 J standard [18], an ideal integrated circuit for space
applications is the qualified manufacturing line20 (QML) Class V with radiation
hardness assurance21 (RHA) level identified in the part specification. From the
perspective of payload designer and developer, only Class V is space quality and
should be the main factor for selecting SDR hardware components.
The FPGA is perhaps the most important component of an SDR and has a long
history for manufactured QML class V parts where rad-hard Xilinx and Actel
(now Microsemi) FPGAs were studied [19]. Currently, Xilinx is the major player
for space-qualified QML level V products used in actual payloads with many more
devices under development. The rad-hard DSP products also follow the QML
process, with Texas Instrument (TI) currently taking the lead for in-flight payloads
with many offerings in space-qualified RF components in addition to DSP. Similarly,
space-qualified GPP follows the same QML path as FPGA and DSP, and the current
on-flight rad-hard GPPs based on the following architecture are [20].

• RISC PowerPC: RAD750, RAD5500.

• RISC MIPS: RH-32, Mongoose-V, KOMDIV-32.

• Motorola 68,000 Series: Coldfire M5208

• ARM Microcontroller: Vorago VA10820

5. Summary and conclusions

In the first section of this chapter, an overview of the satellite bus and payload
subsystems are presented for command and data handling subsystem (C&DHS);
communications subsystem (CS); electrical power subsystem (EPS); propulsion
subsystem (PS); thermal control subsystem (TCS); attitude control subsystem
(ACS) also known as guidance, navigation and control (GNC) subsystem; and
structures and mechanics subsystem (S&MS). A significant portion is spent on
describing the C&DHS and CS with much details on how they are related to other
satellite subsystems for continuous operation.
There are distinctive functional separations between the satellite bus and pay-
load that are discussed at a high level with some examples given; however, there are
currently no existing standard on their interfaces due to legacy satellite design and
development. Examples were given for mission-specific sensing and communica-
tions payloads, showing that pretty much all mission payloads are very customized
in design in legacy systems.
The second section of this chapter covers software defined radio (SDR) as a new
technology with an overview and how SDR is being applied to satellite design and
development in both space and ground segments. There has been a NASA standard
for SDR that has been used for traditional and large satellites and shown to have
some advantages over non-SDR approach.
However, recent rapid developments of Small Satellites (SmallSats), which
CubeSat is a subset of, have resulted in an explosion of SDR applications to build

20
For QML microcircuits, the manufacturer is required to develop a program plan that meets or exceeds
the performance detailed in these appendices of MIL-PRF-38535 J standard.
21
RHA is quantified in terms of the radiation level in Total Ionizing Dose or TID.

21
Satellite Systems - Design, Modeling, Simulation and Analysis

Pathfinder missions that can lead to successful follow-on projects. There remains to
be a standard to be defined for SDR for this CubeSat application. Regardless, SDR is
providing a path forward to a common framework that may enable a more generic
building block for a future concept called Software Defined Satellite that will
change missions based on a software upload.
Since SDR is becoming an important part of a satellite, radiation hardening
of the relevant SDR components is described in some detail. The area is evolving
slowly despite fast changing technology due to the additional design and manufac-
turing steps taken to ensure minimum effects of radiation on microelectronics. The
selection of the appropriate rad-hard FPGA, DSP, and GPP components should be
an important factor in design trade-offs when SDR is being considered for future
missions.

Acknowledgements

The first author, Dr. Hung H. Nguyen, would like to express bountiful apprecia-
tion for his wife, Thuy Le Nguyen, for her constant support during this effort.

Author details

Hung H. Nguyen1* and Peter S. Nguyen2

1 The Aerospace Corporation, Chantilly, Virginia, USA

2 The Raytheon Corporation, Sunnyvale, California, USA

*Address all correspondence to: [email protected]

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

22
Communication Subsystems for Satellite Design
DOI: http://dx.doi.org/10.5772/intechopen.93010

References

[1] Pelton J, Madry S, Camacho- [10] Quintana-Diaz G, Birkeland R.

Lara S, editors. Handbook of Satellite Software-Defined-Radios in Satellite
Applications. Vol. 1. Springer; 2013. Communications, ESA 4S Symposium,
DOI: 10.1007/978-1-4419-7671-0 Sorrento, Italy, May 2018

[2] Griffin M, French J. Space Vehicle [11] Alvarez J, Rice M, Samson R,

Design, Second Edition, AIAA Koets A. Increasing the Capability of
Education Series; 2004. Available CubeSat-based Software-Defined Radio
from: https://arc.aiaa.org/doi/ Applications, 2016 IEEE Aerospace
book/10.2514/4.862403 Conference, 5-12 March 2016

[3] Eger G III. Orion’s Command [12] Maheshwarappa M, Bowyer M,

and Data Handling Architecture, Bridges C. Software Defined
AIAA Space 2008 Conference & Radio (SDR) Architecture to Support
Exposition; 2008 Multi-Satellite Communications, 2015
IEEE Aerospace Conference, 7-14 March
[4] Yra P, Genna M, McMahon S, 2015
Kerns K, Tiede R, Laird M, et al.
Next-Generation Spacecraft Command [13] Asundi S, Fitz-Coy N. Design
& Data Handling System Based on the of Command, Data and Telemetry
RAD750 Processor, 28th AIAA ICSSC- Handling System for a Distributed
2010; 2010 Computing Architecture CubeSat,
IEEEAC Paper #2181, 2013
[5] Kobayashi M et al. NASA’s high-
rate Ka-band downlink system for the [14] Joul S, Lightsey G, Horton S,
NISAR mission. Acta Astronautica. Anandayuvaraj G. A reusable command
2019;159:358-361. DOI: 10.1016/j. and data handling system for university
actaastro.2019.03.069 cubesat missions. In: IEEE Aerospace
Conference; 2014. DOI: 10.1109/
[6] Grayver E. Implementing Software AERO.2014.683636
Defined Radio. Springer; 2013. DOI:
10.1007/978-1-4419-9332-8_12 [15] Stott D, Burek R, Eiseneich P,
Kroutil J, Schartz P, Sweitzer G. The
[7] Reed J. Software Radio: A Modern MSX command and data handling
Approach to Radio Engineering. system. John Hopkins APL Technical
Prentice Hall; 2002. DOI: 10.1109/ Digest. 1996;17(2):143-152
MS.2003.1207473
[16] Hentschel T, Henker M,
[8] Akeela R, Dezfouli B. Software Fettweis G. The Digital Front-End
defined radios: Architecture, state- of Software Terminals. Dredsen
of-the-art, challenges. Computer University of Technology; 1999. DOI:
Communications. September 2018;128. 10.1109/98.788214
DOI: 10.1016/j.comcom.2018.07.012
[17] Sarda K, Roth N, Zee R, CaJacob D,
[9] Space Telecommunications Orr N. Making the Invisible Visible:
Radio System (STRS) Architecture Precision RF-Emitter Geolocation from
Standard, NASA Technical Standard Space by the HawkEye 360 Pathfinder
NASA-STD-4009A. Approved: 2018- Mission, 32nd Annual AIAA/USU
03-14 Superseding NASA-STD-4009 Conference on Small Satellites, August
(Baseline) 2018

23
Satellite Systems - Design, Modeling, Simulation and Analysis

[18] MIL-PRF-38535J. Performance

Specification: Integrated Circuits
(Microcircuits) Manufacturing, General
Specification for (28 December 2010),
US DoD

[19] Roosta R. A Comparison of

Radiation-Hard and Radiation-Tolerant
FPGAs for Space Applications, Jet
Propulsion Laboratory, California
Institute of Technology, NASA
Electronic Parts and Packaging
Program. 30 December 2004

[20] Radiation hardening [Internet].

2020. Available from: https://
en.wikipedia.org/wiki/Radiation_
hardening [Accessed: 11 May 2020]

24