Radio Channels 521386S

Propagation in Satellite Channels II

© Centre for Wireless Communications (CWC), University of Oulu

 Contents 1. Large- and small-scale fading
2. Fading in satellite channels
3. Empirical narrowband models
4. Statistical models
5. Physical-statistical models for
built-up areas
6. Overall mobile satellite
channel

2 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Channel geometry

• High elevation angles, so significant

clutter only close to mobile station

• Atmospheric effects also significant

at SHF (3-30 GHz) and EHF (30-
300 GHz)
Gas attenuation
Rain attenuation
Scintillation

• Closely coupled combination of

large- and small-scale fading
effects

3 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 • Free space loss time-varying
for LEO (500-2000km) and
MEO (5000-12000km)
systems
• Large but constant for GEO
(36000km)

4 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 1. Large- and small-scale fading
Large scale fading:
Due to building or other
obstacle shadowing
KED Theory (Knife edge
diffraction)

Small scale fading:

Constructive or
Large Scale Fading destructive interference
due to multipath
propagation

5 11/7/2022 Small Scale Fading © Centre for Wireless Communications (CWC), University of Oulu
 2. Fading in satellite channels
• Elevation angle from
the mobile to the
satellite large,
minimum angles 8-25°

• Rapid changes in large

scale fading->changes
between Line-of-Sight
(LOS) & Non-Line-of-
Sight (NLOS)

• Small and large scale

fading closely
associated

6 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Local shadowing
Buildings regarded as
diffracting knife-edges
• remember 0.6 x
Fresnel Zone
• shadowing may be
less at high
frequencies

Tree shadowing
• remember 0.6 x
Fresnel Zone
• Attenuation coefficient

7 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Local multipath
• Reflections and rough
surface scattering
produces multipath
and hence fast fading

• Path length differences

small, so wideband
effects modest

• Multiple scattering
weak (e.g. satellite-x-y)

8 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Example of static GNSS (on-ground) multipath
measurement data

Received signal strength vs Angle of Elevation

For GNSS, 0 deg elevation is at horizon & 90 deg is zenith

9 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu
 3. Empirical narrowband models
Models for narrowband propagation in mobile satellite system features

1. Excess path loss & shadowing are dynamic processes

2. No direct deterministic calculation of physical effects

Empirical roadside shadowing (ERS) model predict the probability of fading to a given depth
in presence of roadside trees

(14.1)

% of distance travelled Elevation angle [°]

Extension to 0.8 -20 GHz
.
(14.2)
and are attenuations
at freqs. and
10 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu
 Original ERS model applies only at L band 1.5 GHz, elevation angles
and at fade exceedance percentages from 1 to 20%.

𝐿(𝑃, 𝜃)
𝐿 𝑓

11 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Probability distributions

LoS Rician, k
NLoS Rayleigh

12 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 4. Statistical models LoS Rician, K

Representation of the channel

statistics: mixture of Rice, Rayleigh
and log-normal components
Total narrowband fading signal is
NLoS Rayleigh
composed of a coherent (LOS) and a
diffuse (scattered) parts

( ) (14.5)

The magnitude of a
where is coherent part, and complex Gaussian
are shadowing components, has random variable is a
Rayleigh-distributed
complex Gaussian distribution random variable

13 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Loo model
• For tree shadowing
• Coherent part lognormal
• Multipath is Rayleigh
Total complex fading signal
(14.6)
where is log-normally distributed LOS amplitude, is Rayleigh distributed multipath
amplitude, and are uniformly distributed phases.
The p.d.f. of the fading amplitude ( ) can be approximated by Rayleigh/Log-normal
distributions

(14.7)

where is the standard deviation of multipath part (real or imag), is standard deviation
of and is the mean of
14 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu
 Example prediction: Loo model

Loo and ERS models @ 1.5 GHz, 45° elevation:

and
15 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu
 Lutz model
Two distinct channel states
Appropriate for modelling urban or suburban areas
LoS Rician, k factor
depends on satellite
elevation angle and
carrier frequency

16 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 LOS condition is represented by a ”good” state. The p.d.f. for signal amplitude,
is

NLOS condition is represented by a ”bad” state with p.d.f. found by integrating

the Rayleigh distribution over all possible values of mean power

(14.10)

where mean power varies slowly with log-normal distribution

representing varying effects of shadowing in NLOS.
Proportion of time for which the channel is in the bad state is the time-share of
shadowing, A. So overall p.d.f. of signal amplitude is

(14.11)

17 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Markov chain
The probability distribution of the next state depends only on the current state.
Transitions between states in Lutz model are described by a first-order Markov chain
which represents variations between states by transition probabilities

and (14.12)
In digital communication systems, each state transition is taken to represent the
transmission of one symbol.

18 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Transition probabilities can be found in terms of the mean number of symbol durations spent in
each state:

where and are mean symbol durations in good and bad state, respectively

Time-share of shadowing (the proportion of symbols in the bad state) is

(14.13)

Probability of staying in good state for more than

symbols
Probability of staying in bad state for more than (14.14)
symbols

19 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 An example of the signal variation produced by Lutz model

20 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Lutz highway and city parameters vs. elevation angle

21 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Two-state Lutz model is very useful for analyzing and simulating performance.
Inaccurate in the second-order statistics of shadowing, due to the assumption that transition
between LOS and NLOS is instantaneous.

To overcome this, one may use extra states representing intermediate levels of shadowing
(smaller Rice factor than LOS state)

22 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 5. Physical-statistical models for built-up
areas
• Derive fading distributions directly from distributions of physical parameters
using simple electromagnetic theory
• Hybrid approach which has advantages of both empirical and physical
models
• Require only simple input data (mean building height, variance)
• Applicable over very wide parameter ranges

Building height distribution

• Street canyon environment
• Statistics of building height as input data

23 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Building height distribution
Probability density functions selected to fit the data

Log-normal
⁄ ( ⁄ ) (14.15)

Rayleigh
(14.16)

24 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Time share of shadowing model
parameter for Lutz model
Shadowing assumed to occur when building height exceeds ray height

(14.17)

Assuming Rayleigh height distribution, time-share of shadowing becomes

(14.18)

Shadowing occur exactly when ray is geometrically blocked by building

(14.19)

25 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Comparison with measurements

26 11/9/2022 © Centre for Wireless Communications (CWC), University of Oulu

 6. Overall mobile satellite channel

Time-variant
transversal filter
Satellite process: effects between satellite and Earth surface (tapped delay line)
Terresterial process: all effects in vicinity of the mobile
27 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu
 3GPP NTN channel models

Figure 6.2-3: Combined satellite and terrestrial channels (conceptual

drawing).

28 11/9/2022 © Centre for Wireless Communications (CWC), University of Oulu

 Conclusion
Satellite radio channels consist of satellite and terrestrial processes

• Distance from thousands of kilometres up to 36 000 km -> very large FSPL

component
• Elevation angle (0°) 8° - 25°
• With small elevation angles, the fading is dominated by trees and buildings
• Atmospheric effects
Scintillation
Absorption
Attenuation

29 11/7/2022 © Centre for Wireless Communications (CWC), University of Oulu