Waveform Design Principles for Automotive

Radar Systems
Hermann Rohling, Marc-Michael Meinecke
Technical University of Hamburg-Harburg/ Germany
Department of Telecommunications
Eissendorfer Strasse 40, D-21073 Hamburg
phone: +49-40-42878-3028, fax: +49-40-42878-2281
e-mail: rohling@tu-harburg.de

Abstract—This paper presents a high performance 77GHz radar types of classical pulse waveform with ultra short pulse
FMCW radar sensor for automotive applications. Powerful length (10 ns) or alternatively continuous wave (CW) transmit
automotive radar systems are currently under development for signal with a bandwidth of 150 MHz are considered. The main
various applications. Radar sensor based comfort systems like advantage of CW radar systems in comparison with classical
Adaptive Cruise Control (ACC) are already available on the
market. The main objective from a radar sensor point of view is
pulse waveforms is the low measurement time and low
to detect all targets inside the observation area and estimate computation complexity for a fixed high range resolution
target range and relative velocity simultaneously with a high system requirement. Two classes of CW waveforms are well
update rate. known in literature: The linear frequency modulated (LFM)
and the frequency shift keying (FSK) CW waveform types.
FMCW radar sensors have the advantages of very high range FSK uses at least two different discrete transmit frequencies
resolution but a serious task occurs in multiple target situations to
(see Fig. 1).
suppress so-called ghost targets. In classical FMCW waveforms
this has been solved in using multiple chirp signals with different
slope. But in this case a long measurement time (approximately
fT (t) (a) fT (t) (b)
50ms) is needed which is a contradiction to a high update rate. CPI 0 CPI 1 Chirp 0 Chirp 1
Therefore, in this paper a new waveform design is presented
which has all advantages of FMCW radars but needs an f Step fSweep
extremely low measurement time (10ms) in a radar sensor with 3
different antenna beams. t t
0 TCPI 2TCPI 0 TChirp 2TChirp
Index Terms—Automotive radar, waveform design, FMCW Fig. 1. Two CW waveform principles: (a) FSK modulation, (b) LFM
modulation.
I. INTRODUCTION

This paper describes a new waveform design for automotive

7 7 GHz radars are already on the market as Adaptive Cruise
Control (ACC) systems for high performance automotive
applications. The most important requirement for these radars
applications based on CW transmit signals which lead to an
extreme short measurement time. The main idea of this new
waveform is based on a combination between LFM and FSK
is the simultaneous target range and velocity measurement
CW waveforms in an intertwined technique. Unambiguous
with high resolution and accuracy even in multi-target
range and velocity measurement with high resolution and
situations. The well-known waveforms need a relatively long
accuracy can be required in this case even in multi-target
measurement time (50-100ms). Future developments will be
situations. After an introduction into FSK and LFM waveform
more concentrated on safety applications like Collision
design techniques in section II and III the combined and
Avoidance (CA) or Autonomous Driving (AD). In this case the
intertwined waveform will be described in detail in section IV.
requirements for reliability (extreme low false alarm rate) and
reaction time (extreme short measurement time) are much
higher compared with ACC systems. To give a general idea of II. PURE FSK MODULATION PRINCIPLE
the most important requirements for automotive radar systems
the maximum range for automotive radars is 200m, the range
resolution is 1m and the velocity resolution is 2.5km/h, Pure FSK modulation as shown in Fig. 1 (a) uses two
respectively. discrete frequencies f A and f B (so-called two frequency
To meet all these system requirements specific waveform measurement) [1] in the transmit signal. Each frequency is
design techniques must be considered. For ACC systems both transmitted inside a so-called coherent processing interval
(CPI) of length TCPI (e. g. TCPI = 5 ms ). Using a homodyne
receiver the echo signal is down converted by the where ∆v describes the velocity resolution resulting from the
instantaneous frequency into base band and sampled N times. λ
The frequency step f Step = f B − f A is small and will be CPI duration TChirp ( ∆v = = 0.8 m/s , λ is the
2 ⋅ TChirp
chosen in dependence of the maximum unambiguous target
wavelength of 4 mm @ 77 GHz and TChirp = 2.5 ms ).
range. The time-discrete receive signal is Fourier transformed
in each CPI of length TCPI and targets will be detected by
an amplitude threshold (CFAR). Due to the small frequency
step in the transmit signal a single target will be detected at the ν detected peaks for
same Doppler frequency position in the adjacent CPI’s but Target 1
with different phase information on the two spectral peaks. Target 2
The phase difference ∆ϕ = ϕ B − ϕ A in the complex spectra R
is the basis for the target range R estimation. The relation Fig. 2. : example R-v-diagram for two targets measured with up chirp.
between the target distance and phase difference is given by
the following equation Fig. 2 shows an example inside the R- v -diagram for a two
c ⋅ ∆ϕ target situation measured with a single chirp. Each line drawn
R=− . (1)
in Fig. 2 corresponds to a measured spectral line at index κ
4π ⋅ f Step
indicating all solutions for equation 2 and all possible
To achieve an unambiguous maximum range measurement combinations of target Range R and velocity .
of 150 m a frequency step of f Step = 1 MHz is necessary. In
this case the target resolution only depends on the CPI length For reason of resulting range-velocity ambiguities further
measurements with different chirp gradients in the waveform
TCPI . The technically simple VCO modulation is an
are necessary to achieve an unambiguous range-velocity
additional advantage of this waveform. But this FSK measurement even in multi-target situations. The well known
waveform does not allow any target resolution in the range up-/ down-chirp principle as it is depicted in Fig. 1 (b) is
direction, which is an important disadvantage of this described in detail in [6]. LFM waveforms can be used even in
measurement technique. Especially in automotive traffic multi-target environments, but the extended measurement time
environment more than a single fixed target will occur is an important drawback of this LFM technique.
simultaneously inside an antenna beam. These fixed targets
can not be resolved by a FSK waveform.
f(t) v
(a) (b)

III. PURE LINEAR FREQUENCY MODULATION Tchirp t R

PRINCIPLE
Fig. 3. : (a) waveform for use in multi target situations and (b) corresponding
example R-v -diagram for a two target situation and the related intersection
Radars which apply pure linear frequency modulation points
technique (LFM) modulate the transmit frequency with a
triangular waveform [STO92]. The oscillator sweep is given In multi target situations a waveform as shown in Fig. 3 (a)
by f Sweep . A typical value for the bandwidth is is used which consists of 4 different chirp signals. In general,
the frequencies modulation will be different in each of the 4
f Sweep = 150 MHz to achieve a range resolution of chirps. In each chirp signal all targets are detected which still
fulfil equation (2). The detected spectral lines from all 4 chirp
c
∆R = = 1 m . In general, a single sweep of the signals can be drawn in a single R- v -diagram (Fig 3 (b))
2 ⋅ f Sweep where the gradient of a single line is dependent on the chirp
LFM waveform gives an ambiguous measurement in range R sweep rate. In multiple target situations many intersections
and relative velocity v. The down converted receive signal is between lines of different and adjacent chirps appear as the
sampled and Fourier transformed inside a single CPI. If a example in Fig 3 (b) shows. If such an intersection point
spectral peak is detected in the Fourier spectrum at index κ occurs which has no physical representation of a reflection
(normalized integer frequency) the ambiguities in target range object it is called a ghost target. A real target is represented by
and velocity can be described in a R - v -diagram by the an intersection point between all considered 4 lines.
following equation
v R v R
κ= − ⇔ = +κ , (2)
∆v ∆R ∆v ∆R
 IV. CONCEPT OF COMBINED FSK AND LFM velocity v0 :
WAVEFORMS c ⋅ ∆R ( N − 1) ⋅ ∆ϕ − π ⋅ κ
R0 = ⋅ (4)
The combination of FSK and LFM waveform design π c − 4 ⋅ ( N − 1) ⋅ f Shift ⋅ ∆R
principle offers the possibility of an unambiguous target range
and velocity measurement simultaneously. The transmit
( N − 1) ⋅ ∆v c ⋅ ∆ϕ − 4π ⋅ f Shift ⋅ ∆R ⋅ κ (5)
waveform consist in this case of two linear frequency v0 = ⋅
modulated up-chirp signals (the intertwined signal sequences π c − 4 ⋅ ( N − 1) ⋅ f Shift ⋅ ∆R
are called A and B). The two chirp signals will be transmitted
in an intertwined sequence (ABABAB...), where the stepwise This new intertwined waveform shows that unambiguous
frequency modulated sequence A is used as a reference signal target range and velocity measurements are possible even in
while the second up-chirp signal is shifted in frequency with multi-target environment. An important advantage is the short
f Shift . The received signal is down converted into base band measurement and processing time.

and directly sampled at the end of each frequency step. The v Target f Sweep >
combined and intertwined waveform concept is depicted in Detection!
Fig. 4.
phase
measure
fT (t) ment f Shift >
v0

re cy
R0

t
R

en
su en
B

m
ea u
f Sweep

m freq
A
B
A
fT ,B B fSweep Fig. 5. Graphical resolution principle of ambiguous frequency and phase
fShift fIncr =
fT , A A N −1 measurements.
t
0 TChirp V. SYSTEM EXAMPLE
Fig. 4. Combined FSK-LFMCW waveform principle.

In this section a waveform design based on the new

Each signal sequence A or B will be processed separately by
intertwined signal is developed as an example for automotive
using the Fourier transform and CFAR target detection
techniques. A single target with specific range and velocity applications. The signal bandwidth is f Sweep = 150 MHz to
will be detected in both sequences at the same integer index fulfill the range resolution requirement of 1 m. The stepwise
κ =κA =κB in the FFT-output signal of the two processed frequency modulation is split into N = 256 separate bursts
spectra. In each signal sequence A or B the same target range 150 MHz
and velocity ambiguities will occur as described in Equation 2. of f Incr = = 588 kHz each. The measurement
255
But the measured phases ϕA and ϕB of the two (complex) time inside a single burst A or B is assumed to be 5 µs
spectral peaks are different and include the fine target range resulting in a chirp duration of the intertwined signal of
and velocity information which can be used for ambiguity TChirp = 2.56 ms which results in a velocity resolution of
resolution. Due to the coherent measurement technique in
∆ϕ = ϕ B − ϕ A can λ
sequence A and B the phase difference ∆v = = 2.7 km/h .
be evaluated for target range and velocity estimation. The 2 ⋅ TChirp
measured phase difference ∆ϕ can be described analytically
by the following equation: The important waveform parameter f Shift is optimized on
π v f Shift the basis of high range and velocity accuracy. The highest
∆ϕ = ⋅ − 4π ⋅ R ⋅ , (3)
accuracy occurs if the intersection point in the R - v -diagram
N − 1 ∆v c
results from two orthogonal lines as it is depicted in Fig. 7. For
where N is the number of frequency steps (or receive
this reason the frequency shift between the signal sequences A
signal samples) in each transmit signal sequence A and B. In
this first step ∆ϕ is ambiguous but it is possible to resolve
1
and B is f Shift = − ⋅ f Incr = −294 kHz (the related
these ambiguities by combining the two measurement results 2
of Equation 2 and 3. The intersection point of the two waveform is shown in Fig. 6). In this specific case Equation 4
measurement results is shown in Fig. 5 in a graphical way. The and 5 turn into
analysis leads to an unambiguous target range R0 and relative
R0 N −1 κ
= ⋅ ∆ϕ − , (6)
∆R 2π 2
v0 N −1 κ
= ⋅ ∆ϕ + . (7)
∆v 2π 2

fT (t)

f Sweep
A
B
A
B fIncr
fT, A A
fT,B B fShift t
Fig. 8. Experimental car of the Technical University Hamburg Harburg
Fig. 6. Combined FSK-LFM waveform with optimized frequency shift. equipped with a 77GHz far range radar sensor.

v Target frequency
Detection! measurement VII. CONCLUSION

v0 90° The proposed intertwined CW waveforms show high

performance in range and velocity measurement accuracy. The
R0 R main advantage is the short measurement time in comparison
phase to classical LFM waveforms while the resolution and accuracy
measurement
performance is unchanged. Compared with a pure FSK
Fig. 7. R - v -diagram for the combined waveform with optimized frequency waveform the intertwined waveform allows resolution in
shift. velocity and range simultaneously. The properties of the new
intertwined CW waveform technique are quite promising. This
concept is a good basis for high performance automotive radar
VI. EXPERIMENTAL CAR systems with different safety applications (e.g. pre crash)
which require ultra short measurement and processing time.
The new waveform has been tested in our experimental car in
realistic street situations. The following picture shows the test
car which is equipped with a 77 GHz radar sensor, a smart
brake buster and a throttle control system for automatic REFERENCES
driving. The radar sensor detects all targets inside the [1] Artis, Jean-Paul; Henrio, Jean-François: "Automotive Radar
observation area and measures target range and velocity Development Methodology" International Conference on Radar
Systems, Brest/ France, 1999.
simultaneously. The relevant target is selected by signal [2] Klotz, Michael; Rohling, Hermann: "A high range resolution radar
processing and the car control system is activated by this system network for parking aid applications" International Conference
information. In this case the car follows the detected relevant on Radar Systems, Brest/ France, 1999.
object with controlled distance. [3] Klotz, Michael; Rohling, Hermann: "24 GHz Radar Sensors for
Automotive Applications" International Conference on Microwaves and
Radar, MIKON2000, Wroçław/ Poland, 2000.
The experimental car was used for radar tests in more than [4] Levanon, N; “Multifrequency Complementary Phase-Coded Radar
40000 km on public streets. Many important experiences have Signals” German Radar Symposium, GRS 2000, October 2000, Berlin,
been gained since this time. The implemented sensor is a 77 pp 431-436
[5] Rohling, Hermann; Meinecke, Marc-Michael; Mende, Ralph: "A
GHz FMCW radar with a separated transmit and receive
77 GHz Automotive Radar System for AICC Applications" International
antenna. The target azimuth angle is measured by the receive Conference on Microwaves and Radar, MIKON98, Workshop, Kraków/
signal in 3 adjacent and overlapping beams. The new radar Poland, 1998.
waveform shows very good results and especially extreme [6] Rohling, Hermann; Meinecke, Marc-Michael; Klotz, Michael; Mende,
Ralph: "Experiences with an Experimental Car controlled by a 77 GHz
short measurement time.
Radar Sensor" International Radar Symposium, IRS98, München, 1998.
[7] Stove, A. G.: "Linear FMCW radar techniques" IEE Proceedings-F, Vol.
139, No. 5, Oct. 1992