2021 IEEE International Conference on Power Electronics, Computer Applications (ICPECA)

Design of the Frequency Modulated Continuous

Wave (FMCW) Waveforms, Simulation of the Real
Road Scenario and Signal Processing for the
2021 IEEE International Conference on Power Electronics, Computer Applications (ICPECA) | 978-1-7281-9004-4/20/$31.00 ©2021 IEEE | DOI: 10.1109/ICPECA51329.2021.9362523

Automotive Adaptive Cruise Control

Xinyue Wang
School of Electrical Engineering Southeast University Nanjing, China
[email protected]

Abstract—Adaptive Cruise Control (ACC) system has means a higher resolution and accuracy.in comparison to
aroused much concern nowadays to increase the ability of road 24GHz radar system. Hence, 77GHz FMCW radar system
scenario indication and enhance the safety of self-driving now owns much more preference in vehicle-mounted radars.
application. FMCW radar-based technology is the main
technique used in ACC system for target detection due to its The basic theory of FMCW radar is to obtain the distance
high resolution and accuracy. The dominating purpose from and speed of targets through the IF signal, which is a mixed
radar point of view is to observe all the objects within detection wave of transmitted signal and returned signal from targets.
range and estimate their range and relative velocity Special waveform can be used in modulation for the target
simultaneously and provide all the target information for estimation.
automotive control system for further processing. A 77GHz
FMCW radar system, implementing Fast Fourier Transform This paper describes how a FMCW radar can detect targets
(FFT) technology for signal processing, is presented in this and results are shown by MATLAB. In section II, the
paper to detect specific targets in the designed road scenario for fundamental theories of radar are introduced and several
automotive application. Basic features of FMCW radar are analyses such as power budget and ambiguity function are
introduced in detail, along with the process of radar system discussed based on the real parameter values of 77GHz radar
simulation and waveform generation discussed. The results in extracted from presented literatures. Section III goes through
regard to targets such as Range-Doppler map are displayed in the principles of saw-tooth and triangular FMCW waveforms,
MATLAB. which are implemented in radar detection simulation platform
presented in section IV, as well as the results. Finally, main
Keywords—ACC, FMCW radar, FFT, target detection points of this paper are concluded in section V.
I. INTRODUCTION The whole radar system generally consists of four parts
With the improvement of living standard and abundance which are shown in Fig. 1. The data processing module is one
of material life, cars have become one of the main of the most important part for accurately derivation of the beat
transportations in daily life. Meanwhile, road traffic accidents signal directly relative to range and velocity measurements. In
happened with significantly increasing frequency, and a large this paper, FFT is used in MATLAB for signal processing.
number of residents die each year because of traffic accidents
[1-2]. Such being the case, Adaptive Cruise Control (ACC) is
proposed to assist drivers and mitigate car crashes.
Radar technology is under development and of great
popularity in ACC system for road scenario estimation in time
in order to obtain simultaneous range and velocity
measurements for car controllers to change or adjust travel
performance and adapt to different road conditions. FMCW
radar technology originated in the late 1960s [3], which as a
high resolution measurement technology has been widely
used in the field of automotive measurement instruments [4-
6]. Compared with classic pulse radar, FMCW radar has the
advantages of short measuring time and low peak average
power ratio. In the traveling radar system, the distance and
velocity information of the targets need to be measured at the
same time, and these parameters need to be updated in a short
time. By using some advanced hardware technology, FMCW
radar can easily generate and process signals in real time. Fig. 1. System structure [3]
Commonly, FMCW radars typically have two frequency
bands: 24GHz and 77GHz, in which 77GHz radar system is
more attractive due to its larger bandwidth which usually

978-1-7281-9004-4/21/$31.00 ©2021 IEEE 815 Shenyang, China

Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply. 22-24, 2021
January
 II. RADAR BASIC observed area, the waveform with partial energy
reduced will be reflected back to the receiver. The
A. Radar in Brief receiver also acts as an amplifier to amplify the
Radar, abbreviation for Radio Detection and Ranging, is a attenuated echo signal to a level where its existence can
general term of sensors radiating electromagnetic energy from be detected.
antennas to propagate in observed spaces. In automotive
Adaptive Cruise Control system, radars mounted on vehicles • Mixer: the mixer is a kind of signal processing module,
transmit modulated waveforms (transmitted signal), which in which the transmitted signal and returned signal are
will be intercepted and reflected by different objects coupled. The output of it is a mixture of the two signals
(returned/echo signal), in order to resolve, discriminated and called intermediate frequency signal (IF signal), the
identify the road scenarios. The transmitted and echo frequency of which is known as beat frequency.
waveforms, processed by the radar mixer, provide the • Filter: this process is to filter out clutters so as to obtain
information needed to extract the relative location and a purer waveform containing target-relative figures.
velocity of targets.
• Signal processor: the beat frequencies of IF signal are
The basic elements of the radar system are illustrated in the keys to the calculation of target states and are
Fig. 2. derived in this stage, which, for example, can apply the
The main components and their functions can be Fast Fourier Transformer.
summarized as follows: • Data processor: the information of IF signal is
• Antenna: the antenna of radar system includes two processed mechanically during this process and
parts, one of which is the transmitter and the other one parameters of targets such as ranges, velocities and
is the receiver. The waveform with electromagnetic angular position are outputs of the processor.
energy is transmitted into a specific volume in space
through
g the transmitter. Once there is an object
j in the

Fig. 2. Elementary block diagram of a radar system

B. Radar Classification TABLE I. RADAR SYSTEMS LETTER OR BAND CLASSIFICATION

Historically, radars were first designed to apply in military Frequency Range in GHz
Letter Designation
affairs. It is primarily for this reason the most common radar (IEEE Standard)
systems are classified by the letters or band designations used HF (High Frequency) 0.003-0.03
by the military [7]. Table 1 presents the information associated VHF (Very High Frequency) 0.03-0.3
with radar systems letter or band classification not revise any UHF (Ultra High Frequency) 0.3-1.0
of the current designations. L-band 1.0-2.0
Besides, the type of waveform is another popular classifier S-band 2.0-4.0
of radar system nowadays. Based on this sorting scheme, there C-band 4.0-8.0
are two types of radars: pulse radars and Continuous Wave X-band 8.0-12.5
(CW) radars. Ku-band 12.5-18.0
K-band 18.0-26.5
Pulse radars: Pulse radars radiate series of rectangular
waveforms repeatedly. In this category, the system is Ka-band 26.5-40.0
classified according to the Pulse Repetition Frequency (PRF) V & W or MMW (Millimeter
Normally > 34.0
Wave)
into low PRF, medium PRF and high PRF.
CW radars: In comparison, CW radars employ a

816
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
continuous sine waveform for range measurement and always
use doppler shift frequency to measure the relative velocity of
moving targets [8]. Frequency modulated continuous
waveform (FMCW) radars are one of the most widely used
radars in ACC systems. There are many methods of
modulation, among which, the saw-tooth and triangular waves
are two common application. The main idea of this paper is
the design of FMCW radar system and the principles of
FMCW radar will be detailedly introduced in Part III.
C. Radar Equations
For the purpose of estimating target range, the radar range
equation serves as a function of radar characteristics [8].
Besides, it significantly guides for the radar design.
The main parameters and their symbols in radar equations
are as followed.
x Transmitted signal energy ( ): The electromagnetic
power that possessed by transmitted waveform. Fig. 3. Relationship between received power and distance
x Received signal energy (  ): The electromagnetic
power that returned back to the antenna. Assuming that the received signal is equal to the minimum
x Antenna aperture ( ): The effective area of receiving detective signal ( ), the maximum range can be defined as
antenna that collects the returned power  .  

x Gain of antenna (G): The value of antenna gain is  = (  )  (3)
!"#
defined by equation (1):
Where, $ is Boltzmann’s constant,

= (1)
%& is the standard temperature in Kelvin,
Where  is the wavelength of radar signal in meters (). ' is system noise bandwidth,
According to the above parameters, the power density at * is the noise figure of receiver, $%& ' * is the thermal
  noise according for noise introduced by the practical receiver,
distance  from the radar antenna is   .
 
,+ or SNR is signal-to-noise ratio.
x RCS (  ): RCS represents the radar cross section,
which is a value to show the target characteristic. Substitute equation (2) into (3) with consideration of
Proportional to the target parameters such as physical losses, the maximum range is equal to:
shape, size, material and orientation, the amount of
    4
returned energy is related to property of targets. RCS   = ( ) -./ 0# 1# ( ,2)3
 (4)
56788
is the parameter that lumps all the target characters.
Equation (2) shows the simplest form of radar equation D. Radar Cross Section
  
It is because of the presence of echo signals that targets in
  =    (2) observed area become traceable and sometimes classified.
   
Therefore, radar systems must be critically designed to be able
Where,  is the signal power the receiver received to quantify the echo of different bodies, which is inseparably
(Watts), aligned with target characteristics such as size, material, shape,
and orientation. In order to obtain the echo that can
 is the transmitted power (Watts), characterize targets as accurately as possible, radar cross
 is the transmitter gain, section ( :         been widely
accepted as the designation, is introduced as an effective
 is the distance or radius to the target (), parameter that can describe the echo peculiarities from
specific bodies.
 is the radar cross section of target ( ),
Due to the operation principle of radars mentioned briefly
 is the effective area of the receiver ( ). in section II, the electromagnetic wave transmitted from
Equation (2) demonstrates that the far the distance antennas dispersedly propagates energy, the spatial
between targets and the receiver, the less power the receiver distribution of which is known as scattering. The specific body
will detect. The relationship between received power and that is detected is called scatter and the returned energy from
range can be graphically plotted by MATLAB, which is that target is also known as backscattering, which constitutes
shown in Fig. 3. the echo of the target. As the definition in reference [8], :
explicitly represents the intensity of target echo. Assumed that
the object radiates the power extracted from an incident wave
in all directions, if absolutely all the reflective energy in each
direction is equally propagated back to the transmitter and the
scatted power density calculated is that on the surface of a

817
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
large sphere of radius equal to the range to the radar ( ) The basic formula of power budget analysis is equation (2),
centred on the scatter, the value of : would be the same as which gives the solution of received power. According to the
the projected area of target seen by the transmitter in the unit theory of radar antenna, the relationship between the receiver
of  . The formal definition of : is effective area and gain is

|B | P
 = ;< 4A |B8 | (5)   = (7)
>? / 

Where, C& is electric field intensity of transmitted wave Where,  is the gain of the receiver antenna.
impinging on the target, The radar equation (2) can be rewrote as
CD is electric field intensity of returned signal back to the
radar. [2]   P 
  = ( )Q  R
(8)
The RCS  can also be expressed in decibels ( E' )
referenced to  Except for the :, all the parameters to some extent are
controllable and the received signal power can be obtained. If
 (E') = 10 ;IJ4& (( )) (6) long distance detection is acquired, the transmitted power is in
need of a certain large value so that the echo signal can be
However, the equation (6) is relatively limited for the RCS caught.
of targets, except simplest scatters, varies greatly due to the
After the derivation of received power, it is the next step
orientation of the object and the characteristics of echo are
to analyze the factors that affect the receiver processing
strongly related to the nature of target surface and object size,
process.
leading to the notion of an equivalent sphere not always
meaningful. For instances, for electrically small target who are 1) Noise Power: Noise is the main unwanted
less than a wavelength or so in size, the fluctuation is small electromagnetic energy contributing to the interference of the
because of incapability of long wavelength to resolve target collection of wanted energy. For simplification, the radar
details. In comparison, large targets can be indispensably echo operation environment can be assumed to be noise-free that
sources and contribute to increasingly different echo
no external noise coinciding with desired energy. Besides, the
characteristics. Besides, in practice, the energy transmitted
from transmitter can be absorbed during the propagation or by radar itself will not be capable the generate extra noise during
targets and there is no guarantee that the reflective energy is the operation process. However, there is still an unavoidable
precisely equal in all direction. Thus, it is quite difficult to noise called thermal noise existing. Quantitively, the thermal
estimate : for different objects. Normally, the values of noise is
: for specific targets are determined by various
measurements. For example, reference [9] introduced that + = $%& ' (9)
PNA (Agilent:E8363B) in anechoic chamber room is applied
in UWB (Ultra-wide Bandwidth) measurement of the vehicle Where, N is the thermal noise power (Watts).
and the human body for the waveform with central frequency
of 10, 24 and 79 GHz. Reference [10] focus on : At a standard temperature of 290 K, the noise power
measurements of various pedestrians and it presents validation equals to $%& = 1.38 × 10VW (J/K) × 290(Z) = 4 ×
between the measured : patterns and a full-wave EM 10V4 [/]^. However, in reality, the total noise power must
simulation. According to reference [9]-[12], following table contain extra distributions, larger than the ideal noise power.
shows the common values of : for different objects To account for the additional noise, the practical noise power
detected by 77GHz radar system. All the selected values will can be expressed as the ideal noise power multiplied by a
be used in this paper. figure * .

TABLE II. VALUES OF : +_ = $%& ' * (10)

Radar Cross Section Radar Cross
Radar target
(KL ) Section (MN)
Where, +_ is the total noise power (Watts).
Pedestrian 0.5 O-3 2) Singal-to-noise Ratio: As the signal passes through the
Automobile 100 20
Pickup Trunk 200 O23
system, the receiver signal power is related to a parameter
Bicycle 2 O3 known as signal-to-power ratio, defined as the ratio of receiver
signal power to the noise power.
E. Power Budget
For the insurance of recognition of the echo signal, the `a Dbc bd P
 ,+ = = (11)
energy contained by the signal should be achieved and the 2_D bd 27
minimum power need to be taken into consideration as the
radar system peculiarity. Allowing for the factors such as the Since the detection ability of a radar system can be reveal
size of antenna, propagation losses, antenna gain and signal- by the weakest echo signal that can be identified [12], the
to-noise ration quantities, which can considerably affect the minimum detectable signal, corresponding to a minimum
received power, power budget analysis before the design of signal-to-noise ratio with a certain level of noise is expressed
the radar system is indispensable so as to better evaluate the as
system performance.
 = $%& ' * (,+) (12)

818
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 Associate with equation (9), for long distance radar Where, FSPL represents the free-space path loss, the
detection, the maximum range towards the target in observed formula of which is
space is achieved when the receiver signal power gets the
minimum.  g 
*e = f h (15)
  P    P 
  = ( = ( (13)
)Q !"# )Q -./ 0# 1# ( ,2)!"# Where, E is the distance between the transmitter and
receiver (m)
For different targets, the plot of signal-to-noise ration
versus detection range is illustrated by Fig. 4. Equation (15) can be rewrote in terms of frequency

 gi 
 *e = f h (16)
`

Where, j is the signal frequency (Hz).

Transfer FSPL in the unit of decibels

 gi   gi
 *e = 10 ;IJ4& f h = 20 ;IJ4&
` `


= 20 ;IJ4& E + 20 ;IJ4& j p 20 ;IJ4& (17)
`

In addition of the propagation losses (Lqrss ) which contain

both losses in transmission path and received path, equation
(14) can be modified as

  P    P  4
Fig. 4. Relationship between signal-to-noise ratio versus detection range   = ( )Q !"#
= ( )Q -./ 0# 1# ( ,2)!"#
 (18)
56788

3) Propgation loss: There is one factor, crucial and

F. Ambiguity Function
imperative practically, omitted through all the above analysis.
It is known as the system losses, which cannot be avoided In reality, it is one of the key technologies in radar system
to optimize the cutback of above factors that influence the
during power propagation..
receiver. For the correction of a known signal with the
Free-space model and TRP model [13] are two commonly
presence of additive noise, for example, white noise, matched
used propagation models. In this paper, free-space model is
filters are widely applied in radar due to its ability of
chosen for simplification.
maximizing the attainable SNR at output so as to achieve an
There is only one line-of-sight path between the explicit performance of receiver signal.
transmitter and receiver in free-space model. The received
The simplified block diagram of radar receiver is shown in
power in this model is defined as [7-10]
Fig. 5.
 =     *e (14)

Fig. 5. Block diagram of radar receiver

The output of above system is defined as radar’s ambiguity is a moving target with a Doppler frequency jg , the formula
function [11-14], which represents the envelope or modulus of of the filter output is
the matched filter output with the signal input a Doppler
Ğ
shifted version of original signal. In other words, it provides a  v(w, jg ) = yVĞ t(u) z (u p w){}~  i€ 
Eu (19)
time response of the matched filter to a given energy signal
received coincided with a time delay and Doppler shift.
The modulus square is referred to as ambiguity function
Denote t(u) as the complex envelope of the signal. If there [15].

819
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 Ğ 
|v(w, jg )| = yVĞ t(u) z (u p w){}~  i€ 
Eu (20) Simplify above equation and it follows that
“/
 Ž‘ V‡
The ambiguity function is able to indicate the range and  }(w, jg ) =
‡/
yV“/ {  (ˆ’i€) 
Eu (31)
Doppler information of targets at the same time, for which it
is normally used in waveform design process to provide
insight of different waveforms and tells their appropriateness The result of the integration of equation (31) is
with reference to specific radar detection requirement. In “
addition, the ambiguity function can be used to determine the ‡ D( ‡/ (ˆ‡’i€ )(4V ))
ˆi€ “/
range and Doppler resolution for different waveforms. }(w, jg ) = { (1 p ) “ (32)
‡/ ‡/ (ˆ‡’i€ )(4V )
“/

C is prescribed for signal }(u) as

When it comes to the condition that p& Œ  Œ 0, similar
Ğ analysis can be carried out due to the function symmetry
 C = yVĞ|}(u)| (21)
property expressed by equation (25). Therefore, the ambiguity
function of a upchirp LFM signal can be derived.
The properties of ambiguity function are listed.

x The maximum value of ambiguity function measured D• ‡/ (ˆ‡’i€ )f4V h–
|“|
at the point (w, jg ) = (0,0). |}(w, jg )| = ”f1 p
|‡|
h
“ /
” (33)
‡/ |“|
‡/ (ˆ‡’i€ )f4V h
“/
{ƒ}|v(w, jg )| } = |v(0,0)| = (2C ) = 4C (22)
Where the frequency of LFM signal can be indicate as f =
|v(w, jg )| <= |v(0,0)| (23) ‰w + jg , increasing linearly with time, which will be
introduced in section III.
x The ambiguity function is formed in line with
symmetry. Correspondingly, the ambiguity function for a down-chirp
signal is
 |v(w, jg )| = |v(pw, pjg )| (24)

|“|
D• ‡/ (ˆ‡Vi€ )f4V h–
|‡| “
x The total volume under envelope is constant. |}(w, jg )| = ”f1 p h |“|
/
” (34)
‡/ ‡/ (ˆ‡Vi€ )f4V h
“/
 †|v(w, jg )| EwEjg = (2C ) (25)
In conclusion, for LFM signals, the comprehensive
x v(j) denotes the Fourier transform of the signal }(u). formula of its ambiguity function can be presented as
With reference to Parseval’s theorem, equation (25)
can be written. |“|

D• ‡/ ˜ˆ‡’1™š"P› i€ œf4V h–
|‡| “ /
?  |}(w, jg )| = ”f1 p h |“| ” (35)
|v(w, jg )| = yV? v z (j)v(j p jg ){ V i‡
Ej  (26) ‡/ ‡/ ˜ˆ‡’1™š"P› i€ œf4V
“/
h

x Assume that |v(w, jg )| is the ambiguity of signal Where, *`ž is set to be the chirp figure. For upchirp
}(u) and signal }4 (u) is supposed to be signal }(u) signal, *`ž = 1, and for downchirp signal, *`ž = p1.
with an additional quadratic phase modulation, the
ambiguity function of }4 (u) is To attain the ambiguity function, values of w& , ‰ and
*`ž are ought to be known inputs. For a particular LFM
 }4 (u) = }(u){  ˆ
(27) signal, w& can be the sweep time for each chirp and ‰ is
determined by its bandwidth and sweep time. The theory will
|v4 (w, jg )| = |v(w, jg + ‰u)| (28) be displayed in section III. Thus, consider the design
requirement in this paper, the inputs of ambiguity and their
For LFM (linear frequency modulated) signal used in this values for a 77GHz FMCW radar system are:
paper, the formula of the signal is defined in reference [21].
TABLE III. VALUES OF :
4 
 }(u) = {‹u( ){  ˆ
(29) Symbol Description Values
Š‡/ ‡/ Pulse width/
w& 7.33 microseconds
Sweep time
Under the case when 0 Œ  Œ & , the integration is limited '[ Bandwidth 300 MHz
‡ ‡ Upchirp = 1
between p / and / p  [21]. *`ž Chirp figure
  Downchirp = -1
Submitting equation (29) into equation (19), the ambiguity
III. FMCW WAVEFORM
function for LFM complex envelope is
FMCW radar system represents for frequency modulated
}(w, jg ) = continuous waveform. The transmitter of a FMCW radar
4 Ğ  V‡ emits an electromagnetic signal with frequency changing over
y {‹u f‡ h {‹u f ‡ h {  ˆ
‡/ VĞ
{ ( ˆ(V‡)  i€ )
Eu (30)
time in a sweep across a specified large bandwidth. The chirp
/ /
signal transmitted then hits targets and subsequently returned

820
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
back towards the receiver, accompanied with a time delay and target.
frequency shift due to a relative range and velocity. By mixing
the transmitted signal and the returned signal, the frequency • Doppler resolution: ¢¤ = 1 /t .Doppler resolution
difference, also referred to as beat frequency between these determines how well the radar can resolve targets with
two can be detected as a signal in an intermediate frequency different speeds.
range. The range measurements, as well as velocity, are in A. Advantages of FMCW Radar
need of the beat frequencies as known quantities. Reference [23,25,28,29] mentioned the advantages of
In this paper, the parameters of the radar system are FMCW waveforms. In conclusion, the features and merits are
defined as followed. listed below.
• Carrier frequency: j` = 77]^ . On the market of x Accurate short-range measurements.
ACC system, 77GHZ radar, due to the high range x Low sensitivity to interfering with extremely high
resolution and accuracy [25] are currently popular used degree of range resolution.
for high preference automatic applications [26-27]. x Easy integration.
• Wavelength: ‹ = 1{8 ,t . According to the light x Low measurement time.
propagation speed and carrier frequency, the x Low computation complexity.
wavelength can be calculated by equation (36). x Capability to use a lower transmitted power to achieve
the same detective range in comparison with classic
` pulse radar due to its average-to-peak power ration
 ¡= (36)
i™ equal to 100% duty cycle.

• Unambiguous range:  = 200 . This value B. Analysis of FMCW Waveforms

typically determines the size of observed area. In other Given to different modulation methods, there are
words, it is the range of farthest target the radar can essentially two types of linear frequency modulated (LFM)
detected. waveforms – saw-tooth LFM waveform and triangular LFM
waveform.
• Range resolution: ¢ = 0.5 . The resolution how far
apart targets need to be so that they are distinguishable 1) Saw-tooth FMCW: After modulation, the frequency of
as different objects. the continues waveform changes with time. Fig. 6 illustrates
• Unambiguous Doppler: ¤ = 230 $/¥ . This up-chirp waves.
parameter represents the maximum velocity
y of the

Fig. 6. Saw-tooth LFM waveform

`
Where, u is the sweep time of single chirp. '§ = (38)
¢
Reference [28] explains bandwidth as a representation of
information, which is determined by signal pulse width or any According to Fig. 3, the sweep slope of this upchirp can be
internal modulation of the signal. From equation (37), the defined as:
value of bandwidth can be derived in equation (38).
0¦
$= (39)
` !
¢ = (37)
0¦

821
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 The frequency of the wave is proportional to the sweep The equation of this signal in time dimension is:
time.
(u) = ‹It˜¨(u)œ = ‹It(2Aj` + A$u  ) (42)
j(u) = j` + $u (40)
Take signal chirp as an example, the solid line in Fig. 6
The phase of the signal is given by: represents the transmitted signal. After a delay period, the
returned signal (dashed line) transmitted back to antenna,
¨(u) = 2A y j(u)Eu = 2Aj` + A$u  (41) which is demonstrated in Fig. 7.

(a) stationary target (b) moving target

Fig. 7. Single upchirp FMCW waveform

According to equation (42), the transmitted signal can be 4 g¬­ -

jª1 =  = $w = (51)
written as:  g `

Compared equation (50) with (47), the frequency of mixed

% = ‹It(2Aj` u + A$u  ) (43)
signal is the beat frequency, which consists of the unknown
value – relative range of target.
and the phase is:
Moving target: If the target has a constant velocity ¤, the
 ¨ . = 2Aj` u + A$u  (44) beat frequency include an extra Doppler shift, shown as jg in
Fig. 5.
Stationary target: The received signal is back towards the
®
antenna after a delay time w: jg = (52)

 w= (45) j© = jª1 =
-
p jg (53)
`
`

Where,  is the initial range of the target. 2) Triangular FMCW: Equation (53) reveals that the beat
Based on Fig. 5, then the received signal and its phase is: frequency contains two unknow parameters – target range
and velocity. However, there is only one equation,
 = ‹It(2Aj` (u p w) + A$(u p w) ) (46) insufficient to resolve the target. Such being the case, it is
necessarily available to add a downchirp signal, which, with
¨ = 2Aj` (u p w) + A$(u p w) (47) the upchirp signal, forms the triangular FMCW waveform.
The waveform of a triangular FMCW and the its time-
The beat frequency between these two signal equals to: frequency plot are drawn in Fig.8 and Fig. 9.
-
j© = w$ = (48)
`

The range measurement is done by mixing the transmitted

and echo signal. In the mixer of the receiver channel, signal
from local oscillator is multiplied with the echo signal with
phase ¨ .

ª1 = %   = ‹It(¨ . )‹It(¨ )

« ‹It(¨ . p ¨ ) = ‹It(¨ª1 ) (49)

¨ . p ¨ = 2Aj` w p A$w  + 2A$w  u (50)

Fig. 8. Triangular FMCW waveform

822
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 Similar to saw-tooth waveform, the range and velocity Which is the same as equation (38). The sweep bandwidth
measurements are detailly introduced as followed. is inversely proportional to range resolution which should be
taken into great consideration when both high range resolution
and high bandwidth are needed. From several reference, the
77GHz FMCW are widely used in ACC system for its high
resolution. That’s why the 77GHz waveform is chosen in this
paper.
C. Spectrum and Spectrogram
In signal processing module, a spectrum describes the
frequency component of an entire signal, which converts a
time series signal into frequency sequence. With Fast Fourier
Transformer technology, a physical signal can be decomposed
into several discrete frequency. A simple process of getting a
spectrum is shown in Fig. 10 and the spectrum of an up-chirp
LFM waveform is plotted in Fig. 11.

Fig. 9. Frequencies of Triangular FMCW waveform and IF Signal

The beat frequency has two values related to target range

and Doppler shift.

¯ž -
jª1 = p jg (54)
`

-
jª1g_§ = + jg (55)
`

The solutions of above equation sets are:

` ¯ž
R= ˜jª1 + jª1g_§ œ (56)
-

4 ¯ž
jg = ˜jª1g_§ p jª1 œ (57)


Combining equation (52) and (57), the velocity of the

target is derived.

¯ž
¤ = (jª1g_§ p jª1 ) (58)
 Fig. 10. Block diagram of getting spectrum

Both Fig.2 and Fig.3 illustrate that the frequency of carrier

waveform increases linearly with time and the sweep slope
can be given by the quotient of the change of frequency over
that of time ¢j ,¢u. Equation (47) can be rewritten as:
¢³ µ
 f± = k =   (59)
¢´ ¶

In order to obtain a specific range resolution ¢ , the

frequency resolution is:
¢³ ¢µ
 ¢f± =   (60)
¢´ ¶

For a spectral resolution ¢j© , the signal needs to be

capable for a minimum observation dwell time equal to
1,¢j© = %g .
¢´ ¶ Fig. 11. Spectrum of up-chirp LFM waveform
 T· =  (61)
¢³ ¢µ
However, spectrum only indicates the information of
When it comes to the total sweep frequency, signal frequency and amplitude, omitting the varying with
time. Spectrogram visualized the signal frequency variation
¢³ ¶ process with time axis, which plot the graph with time as
 B¹ = T· = (62)
¢´ ¢µ horizontal axis and frequency as vertical axis, as well as grey
level indicating the amplitude. Applying short Fourier

823
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
Transformer, spectrogram is a spectrum of short chunks of
signal, which take a small segment of the signal and derive the    ¡ 
 = ([)
frequency of it at that time duration, thus giving the change (4A)W 
tendency of frequency with time.
= 10;IJ10( ) +  + 
Fig. 12 illustrate the process of doing short Fourier
Transform of a linear frequency waveform, from which it can +10;IJ10(¡ ) + 10;IJ10()
be seen that the frequency of this wave is increasing linearly
with time. p10;IJ10((4A) ) p 10;IJ10( ) (E') (63)

Which is proportional to the ratio of RCS over the fourth

of range for each target.
B. FMCW Waveform Generation
According to the basic theory of FMCW waveforms
introduced in section III, the up and down chirps radiated to
space need to be generated critically for the detection of road
scenario with high accuracy.
The frequency and bandwidth of this FMCW waveform
are concluded in TABLE V.

TABLE V. FREQUENCY AND BANDWIDTH OF FMCW WAVEFORM

Parameter Value
Carrier frequency (ÆÇ ) 77 ÂÈÉ
Bandwidth (BW) 300 MÈÉ

Because the maximum range is set to be 200m, the

Fig. 12. Spectrogram of up-chirp LFM waveform
maximum time delay can be calculated by equation (45).
×&&
From the spectrum, we can only easily tell the bandwidth w = O 1.3333 × 10VÊ (s) (64)
`
of the up-chirp is 300MHz. While in Fig. 12, the spectrogram
of this signal not only show the bandwidth, but also clearly To make sure that the echo signal is returned back to the
shows its frequency changing trend over time. receiver before the next sweep chirp is transmitted, the sweep
IV. RADAR SYSTEM SIMULATION time for each single chirp should be 5~6 times the maximum
delay time [1]. The coefficient of 5.5 is chosen here.
A. Radar System Setup
In ACC system simulation, major radar parameters such u = 5.5w = 7.3333 × 10VÊ (s) (65)
as the transmitter, the receiver, and the antenna should be
taken into consideration with other components omitted for Where, u is the sweep time of single up or down chirp.
simplify. Substituting equation (64) into (39), the sweep slope of the
Based on the power budget analysis in section II, signal is
following values of radar parameters for 77GHz ACC radar
simulation platform are chosen to use in this paper [30-31]. W&&×4&Ë ÌÍ
$= = 4.0909 × 1013 (66)
Î.WWWW×4&ŽË D
TABLE IV. PARAMETERS OF 77GHZ ACC RADAR SYSTEM
Because the maximum velocity of target is 230 km/h, so
Name of Parameter Value the maximum Doppler shift based on equation (52) is
Antenna aperture (º» ) ¼. ½¼ × ¾½V¿ KL
Antenna gain 27.0042 dB
®!ÏÐ ×(W&÷W.Ê)
Transmitted power (ÀÁ ) 3.2 mW jg = = (67)
(W×4&Ñ )÷(ÎÎ×4&Ò )
Transmitter gain (ÂÁ ) (9+ Antenna gain) dB
Receiver gain (ÂÃ ) (15+ Antenna gain) dB
Receiver noise figure (ÄÅ ) 4.5 dB Thus, the maximum beat frequency is

Assume that the maximum target range this radar system !ÏÐ -
j© = + jg (68)
can detect is 200m, substituting these data into the equation `
(4), the minimum SNR is equal to 4.5632dB. Normally, 20dB
of SNR is chosen. Besides, the range resolution is set to be 1) Saw-tooth wavefroms: According to the theory
0.5m. introduced in section III, the transmitted signal of an up-chirp
can be expressed as
Combine equation (1) and (2), the received power of echo
signal is:
% =  ‹It(2Aj` u + A$u  ) (69)

Where A´ is the amplitude of transmitted wave which is

proportional to the transmitted power.

824
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
  «  (70) According to equation (52), the Doppler frequency can be
calculated which is in the unit of KHz, much smaller than the
For a specific target, the signal hitting on the target and bandwidth in MHz. Such being the case, it is not very apparent
being transmitted back is of Doppler frequency shown in the spectrogram.
After mixing the transmitted and received signal, the IF
 =  ‹It(2Aj` u + A$u  + 2A(jg p $w)u) (71) signal is produced, shown in Fig. 15. In Fig. 15, IF signal is
presented with a constant frequency and the beat frequency
Where AÔ is the amplitude of echo wave which is can be extracted.
proportional to the received power.

 «  « (72)


Õ
To be simplify, assume A´ = 1 and AÔ = .
µ

All the parameters can be computed in MATLAB. Thus,

the transmitted signal is generated, which is shown in Fig. 13.

Fig. 15. IF signal and beat frequency

2) Trangular waveforms: A triangular waveform is a

combination of two sawtooth sweep, one is up-chirp and
another is down-chirp. With reference to Fig. 3, the triangular
waveform can be mathematically expressed.
The frequency of the waveform is

Fig. 13. Up-chirp FMCW transmitted signal $u, 0 Πu Πu

j (u) = Ö (73)
p$u + 2$u , u < u Π2u
From the spectrogram of this signal, the bandwidth equal
to 300MHz can be told. Where u is the sweep time of single up-chirp or down-
Take the target car whose RCS is 100m2, speed is 30km/h chirp.
and initial range is 60m as an example, the echo signal from it Then its phase is calculated by equation (41).
is shown in Figure. 14, in which the amplitude of received
signal is reduced compared to transmitted signal. It indicates A$u  , 0 Πu Πu
the effect of factors in radar equations and is a reflection of ¨(u) = Ö  (74)
power budget analysis. pA$u + 4A$u u, u < u Π2u

The signal can be described in the form of exponent which

contains its amplitude and phase information at the same time.

%} = { ¬() = ‹It˜¨(u)œ + Øt<Ù(¨(u)) (75)

The frequency of returned signal is that of the transmitted

signal shifted a delay time along time axis, along with a
Doppler shift.

j (u) = j (u p w) + jg (76)

The function ‘horzcat’ can be used to concatenate these

two chirps and the result is shown below.
fd The figure of the up-chirp part is the same of Fig. 13 and
the down-chirp is drawn in Fig. 16.


Fig. 14. Up-chirp FMCW received signal

825
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 C. Plots of Ambiguity Function
Based on the equations mentioned in section II, the
ambiguity function can be produced in MATLAB and the plot
of it will show exactly the performance of the waveform,
whose x-axis is range information corresponding to target
delay time, y-axis shows the Doppler effect and z-axis is the
amplitude. The ambiguity function of the above saw-tooth
wave is shown below in Fig. 18.

Fig. 16. Down-chirp FMCW waveform

The bandwidth (300MHz) of the signal is clearly shown in

its spectrogram.
Also take the target used in section IV as an example, the
whole triangular waveform is generated in Fig. 17

Fig. 18. Ambiguity function

In Fig. 18, the maximum point of its ambiguity function

appears at zero point which corresponds to equation (35), the
first property of the function.
Change the plot to y-z version, the zero-Doppler cut is
shown in Fig. 19 and is to determine the range resolution
corresponding to -3dB is marked. The range resolution due to
the values are approximately 7.038-6.598=0.44 m, which
meets the requirement of 0.5 m.

(a) Transmitted signal

Fig. 19. Zero-Doppler cut

Besides, from the zero-Range cut in Fig. 20, the Doppler

resolution can be used to design the velocity resolution
requirement.

(b) Received signal

Fig. 17. Triangular FMCW waveform

826
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 be a returned signal back to the receiver with different
amplitude and frequency. The detection process is concluded
in the flow diagram Fig. 22.
Based on the theory in section III, the phase of the IF signal
for up-chirp and down-chirp is

¨¯ž = 2Aj` w p A$w  + 2A$w  u p 2Ajg u

Ü (77)
¨g_§ = 2Aj` w p A$w  + 2A$w  u + 2Ajg u

After FFT of the IF signal, each peak value of its spectrum

corresponding to the beat frequency for each target returned
signal. In addition, the frequency axis can be converted into
range axis according to equation (47).

Fig. 20. Zero-Range cut

Looking the plot of function in the x-y axis, which gives

the relationship of range and doppler in Fig. 21, from the
function point of view, the max point is of whose range equals
to zero and Doppler shift equals to zero.

Fig. 22. Flow chart of detection

The result of detection plotted in MATLAB is shown in

Fig. 23, one of which the x-axis is the value of beat frequency,
while another one’s x-axis is the target range. In essence, there
is no difference between that two figures because the beat
frequency indicates the range information. The value of range
axis can be obtained by the following equation.
iÏÐ"8 `
D = (78)
-
Fig. 21. Doppler-Range cut
In equation (78), D is the range axis, jD is the axis of
D. Multiple Targets Detection beat frequency, c is the speed of light and k is the sweep slope
Radar targets in ACC application are objects constituting of the waveform.
the road scenario, such as vehicles, pedestrians, obstacles,
vegetations and road shoulders. Considered the characteristics
of such targets, specific values according to radar parameters
are in need of being set to meet up the detection requirements.
Hence, as to the targets, four different targets are supposed in
this road scenario and the maximum target velocity should not
exceed 230km/h.

TABLE VI. TARGET PARAMETERS

Radar Radar Cross Velocity Initial Range
Number
target Section (MN) (ÚK/Û) (K)
Pedestrian 1 -3 10 80
Ordinary 20 -40 50
2
Car 20 30 60
Pickup
1 23 70 160
Trunk
Bicycle 1 3 15 110

The signal generated in section IV which is shown in Fig.

17(a), is propagating into space and for each target, there will Fig. 23. Result of detection

827
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 When zoom in the beat frequency corresponding to each whose column number is equal to the number of chirps
signal, two waves of frequency are presented in Fig. 24, one transmitted to an object N¶àáÔâ .
¯ž
of which is the beat frequency related to the up-chirp (jª1 )
¯ž After first Fourier Transform along vertical axis, the
and another one is from the down-chirp (jg_§ ).
spectrum of all the IF signals are combined in a NFFT ×
Thus, consider the two beat frequency values separately N¶àáÔâ _chirp matrix (NFFT is the points number of FFT), in
for up and down chirp, the actual range after sweep duration which the range information of the target is saved. Then take
should be calculated by equation (56) with the effect of both the second Fourier along horizontal axis to get the Doppler
delay time and Doppler shift, as well as the velocity calculated frequency and the Doppler frequency axis can be converted to
by equation (58). velocity by equation (52). As a result, NFFT × N¶àáÔâ _chirp
E. Range-Doppler Map matrix is generated whose 3-D plot has target relative velocity
as x-axis, target range as y-axis and amplitude as z-axis.
For a moving target, if a waveform consisting of sequences Change the 3-D plot into x-y dimension, the Range-Doppler
of chirps with short duration t Þ , the phase information is illustrated in Fig. 26.
appears in each single chirp will change. Thus, different
echoes’ data can be separated in Doppler domain. As said in In Fig. 27, the five bright spots represent the five targets
section II, Range-Doppler map, based on the ambiguity detected. The values of x-coordinate indicate their velocities
function is a figure contains the range and velocity together by and the y-coordinates show their initial ranges. All the
applying FFT into two dimensionalities. The basic theory and coordinates meet the deign values in table V.
processing process are shown in Fig. 25. Compared to the Fig. 21, it can be said that the range-
The IF signal obtained from each chirp is stored in a Doppler map of targets is the movement of the maximum
column vector and all of the signals incorporate a matrix point in ambiguity function from point (0,0) to where the
target is.

Fig. 24. Beat frequency

828
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
 Fig. 25. Process of plot Range-Doppler map

Fig. 26. 3-D plot of signal matrix after two FFTs

Fig. 27. Range-Doppler map

V. CONCLUSION
With the rapid development of intelligent transportation
today, the research and development of vehicle-mounted radar
is particularly urgent.
In this project, a saw-tooth waveform and a triangular

829
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.
waveform are proposed for target estimation based on 77GHz [10] Belgiovane, D., Chi-Chih Chen, Ming Chen, Chien, S.Y.-. and Sherony,
linear frequency modulation radar technology. From R. (May 2014) 77 GHz radar scattering properties of pedestrians. IEEE,
pp. 735.
triangular FMCW waveforms, the beat frequencies obtained
[11] Bel Kamel, E., Peden, A. and Pajusco, P. (Mar 2017) RCS modeling
respectively from up-chirp and down-chirp contains targets’ and measurements for automotive radar applications in the W band.
range and Doppler shift information, by which the ranges and Euraap, pp. 2445.
velocities of detected objects can be calculated and figures [12] M. I. Skolnik, Introduction to Radar Systems. New York: McGraw Hill,
showing targets’ condition can be plotted. With a sequence of 2001
chirps, range-Doppler map of targets can be displayed after 2- [13] Sizov, V., Cherniakov, M. and Antoniou, M. (2007) 'Forward
D FFT, which tells the range and velocity of targets at the scattering radar power budget analysis for ground targets', IET Radar,
same time. The results are confirmed to be right compared to Sonar & Navigation, 1(6), pp. 437. doi: 20060174.
given parameters of targets. [14] Rappaport, T.: ‘Wireless communications: principles and
practice’(Prentice-Hall PTR, 1996)
From ACC system perspective, radar system is one of its [15] Parsons, J.D.: ‘The mobile radio propagation channel’ (John Wiley
most important sensors, leading to the imperative to &Sons, 2000, 2nd edn.)
improvement of estimation level for complex road scenario. [16] Poole, Ian. "Free Space Path Loss: Details, Formula, Calculator". radio-
The radar system in this paper is too simple to deal with electronics.com. Adrio Communications Ltd. Retrieved 17 July 2017.
practically complicated situation. Thus, it is the next step to [17] Jankiraman, M., Design of Multifrequency CW Radars, Raleigh, NC:
simulate more integrated road scenario, for example, taking SciTech Publishing Inc., 2007.
the target orientation and acceleration into account. [18] Levanon, N., Radar Principles, New York, NY: John Wiley & Sons,
1988.
Additionally, high-precision target parameter estimation
algorithm and the enhancement of range and Doppler [19] Mahafza, B. R., and A. Z. Elsherbeni, MATLAB Simulations for Radar
Systems Design, Boca Raton, FL: Chapman & Hall/CRC, 2004.
resolution are also the emphases needed to be kept pace with.
[20] Pace, P. E., Detecting and Classifying Low Probability of Intercept
Radar, Norwood, MA: Artech House, Inc., 2004.
REFERENCES
[21] Mahafza, B.R.Radar Systems Analysis and Design Using MATLAB.
[1] Kim, Sung Soo, and M. Won. The Development of a Real-Time Chapman and Hall/CRC.
Multibody Vehicle Dynamics and Control Model for a Low Cost
Virtual Reality Vehicle Simulator: An Application to Adaptive Cruise [22] Jankiraman, M., Design of Multifrequency CW Radars, Raleigh, NC:
Control. Virtual Nonlinear Multibody Systems. Springer SciTech Publishing Inc., 2007.
Netherlands,2003. [23] M.Jankiraman.: 'FMCW Waveform', in London,: 'FMCW Radar
Design' (Artech House Press, 2018, 1st edn.), pp. 19-20
[2] Vahidi, A., and A. Eskandarian. "Research advances in intelligent
collision avoidance and adaptive cruise control." IEEE Transactions on [24] Pace, P. E., Detecting and Classifying Low Probability of Intercept
Intelligent Transportation Systems 4.3(2004):143-153. Radar, Norwood, MA: Artech House, 2009.
[3] O’Hara F, Moore G. A high performance CW receiver using feedthru [25] Rohling, H. and Meinecke, M.-. (2001) Waveform design principles
nulling[J]. Microwave Journal, 1963, 6(9): 63-71. for automotive radar systems. IEEE, pp.
[4] Komarov I V, Smolskiy S M. Fundamentals of short-range FM [26] Karnfelt, C., Peden, A., Bazzi, A., El Haj Shhade, G., Abbas, M. and
radar[M]. Artech House, 2003. Chonavel, T. (Oct 2009) 77 GHz ACC radar simulation platform. IEEE,
pp. 209.
[5] Stove A G. Linear FMCW radar techniques[C]//IEE Proceedings F
(Radar and Signal Processing). IET Digital Library, 1992, 139(5): 343- [27] Stove, A.G. (1992) 'Linear FMCW radar techniques', IEE Proceedings
350. F Radar and Signal Processing, 139(5), pp. 343. doi: 10.1049/ip-f-
[6] Jia H, Kuang L, Zhu W, et al. A 77 GHz frequency doubling two-path 2.1992.0048.
phased-array FMCW transceiver for automotive radar[J]. IEEE Journal [28] Merrill Skolnik, Radar Handbook, Third Edition 2008
of Solid-State Circuits, 2016, 51(10): 2299-2311. [29] [Marco Roggero, John Zhao and Giorgia Zucchelli, ' Design of FMCW
[7] BASSEM R. MAHAFZA.: ‘Definitions and Nomenclature', in Radars for Active Safety Applications', IETE Journal of Research, pp.
Huntsville, Alabama, USA: 'Radar Systems Analysis and Design Using 1-2. doi: 10.1080/03772063.2019.1615392.
MATLAB' (CRC Press, 2013, 3rd edn.), pp. 3–9 [30] Karnfelt, C., Peden, A., Bazzi, A., El Haj Shhade, G., Abbas, M. and
[8] Merrill Skolnik, Radar Handbook, Third Edition 2008 Chonavel, T. (Oct 2009) 77 GHz ACC radar simulation platform. IEEE,
pp. 209.
[9] Motomura, T., Uchiyama, K. and Kajiwara, A. (Jan 2018)
Measurement results of vehicular RCS characteristics for 79GHz [31] Zhou, W., Li, Y. and Zhang, J. (Dec 2018) Estimating Target Situation
millimeter band. IEEE, pp. 103. Based on 77GHz Radar Sensor for ACC System. IEEE, pp. 2436.

830
Authorized licensed use limited to: SASTRA. Downloaded on September 07,2023 at 09:56:41 UTC from IEEE Xplore. Restrictions apply.