Progress In Electromagnetics Research, PIER 98, 33–52, 2009

SIDELOBES REDUCTION USING SIMPLE TWO AND

TRI-STAGES NON LINEAR FREQUENCY MODULA-
TION (NLFM)

Y. K. Chan, M. Y. Chua, and V. C. Koo

Faculty of Engineering & Technology
Multimedia University
Jalan Ayer Keroh Lama, Bukit Beruang, Melaka 75450, Malaysia

Abstract—The Linear Frequency Modulation (LFM) waveform is

the most commonly and extensively used signal in practical radar
system. However a compressed LFM signal at the receiver will
produce the first sidelobe at a level of −13 dB to the peak of
the main lobe. A weighting function is needed to apply in order
to reduce the sidelobes. However, the penalty of mismatch loss
is clearly evident. It may reduce output SNR, typically by 1 to
2 dB. Every single dB of additional SNR can have great effects in
reducing false alarm rates in target detection applications. In an
effort to achieve low autocorrelation sidelobe level without applying
weighting function, Non-Linear Frequency Modulation (NLFM) signal
has been investigated. This paper describes the sidelobe reduction
techniques using simple two-stage FM waveform, modified two-stage
FM waveform and tri-stage FM waveform. Simulation results of the
proposed NLFM signal are presented. Sidelobe reduction of more than
−19 dB can be achieved by this design without any weighting technique
applied.

1. INTRODUCTION

The matched filter and pulse compression concepts are the basic
of radar processing algorithms [1, 2]. Since radar return is always
susceptible to noise and interference from all kinds of objects
illuminated by the antenna beam, the receiver must be optimized. The
matched filter is a filter whose impulse response, or transfer function
is determined by a certain signal, in a way that will result in the
maximum attainable signal to noise ratio [3–6]. Pulse compression
Corresponding author: Y. K. Chan ([email protected]).
34 Chan, Chua, and Koo

involves using a matched filter to compress the energy in a signal into

a relative narrow pulse.
Pulse compression is a classical signal processing technique to
increase the range resolution of transmitted pulse without having
to increase the peak transmit power [1, 3, 5, 7]. Instead of a fixed
frequency pulse, the transmitted pulse is modulated by a specific phase
or frequency pattern during a wider pulse interval. The receiver uses a
pulse-matched filter to pass reflected pulses that match the pattern of
the outgoing pulse and at the same time, reject noise and other signals.
The LFM, or chirp waveform, has superior performance in pulse
compression radar since they can be easily generated and processed.
Many diverse techniques and devices have been developed to provide
the required pulse compression processing for these signals [2].
However, the LFM has large sidelobes with respect to the mainlobe.
Reducing the sidelobes can be accomplished by linear filtering the
output, i.e., applying window functions or data tapering. However,
since the cumulative filtering is no longer precisely matched to the
signal, it necessarily reduces output SNR as well, typically by 1–2 dB.
The non-linear frequency modulation (NLFM) signal is another
continuous phase modulation waveform applicable to pulse compres-
sion radar. It has been claimed to provide range sidelobe suppression
since the modulation is designed to provide the desired amplitude spec-
trum with reduced sidelobes [3, 8–10]. The NLFM requires no time or
frequency weighting for range sidelobe suppression.

2. LINEAR FREQUENCY MODULATION (LFM)

WAVEFORM

The LFM waveform is the most widely discussed pulse compres-

sion waveform in the literature and most extensively used in prac-
tice [3, 4, 11]. The phase function of the LFM signal consists of a
quadratic phase coefficient k which results in a linear frequency over
the duration of the signal. Thus the chirp waveform can be described
by the Re{s(t)} with
½ µ ¶¾
1 2
s(t) = exp j2π fc t − kt , −τ /2 ≤ t ≤ τ /2
2 (1)
= 0, elsewhere
where fc is the carrier frequency of transmitted waveform, k is the
chirp rate or the sweep rate of the waveform. The bandwidth of the
signal is given by
B = kτ (2)
Progress In Electromagnetics Research, PIER 98, 2009 35

The instantaneous frequency f (t) can be obtained by differentiating

the argument of the exponential in Equation (3),
1 d(2π(fc t + kt2 /2))
f (t) = = fc + kt (3)
2π dt
Although the LFM signal has a duration of τ , it can behave like
a pulse with duration equivalent to the inverse of its bandwidth, i.e.,
τeq = 1/B. The signal processing that allows this to happen is known
as pulse compression. The amount of this compression is given by
τ /τeq = τ B = D, which is the time-bandwidth product of the waveform.
The LFM chirp exhibits the interesting property of possessing very
large time-bandwidth products [4, 6]. The transform of the chirp is
essentially flat over its range of frequencies. When the time-bandwidth
product of the LFM signal increase, the signal’s spectrum shape will
become more rectangular.
The matched filter of the LFM has an impulse response, h(t), that
is time inverse of the signal at receiver input as given below
½ µ ¶¾
1 2
h(t) = K cos 2π fc t − kt , −τ /2 ≤ t ≤ τ /2 (4)
2
where K is factor that result in unity gain. Since the echo from the
target at time to is delayed replica of the transmitted signal, the return
signal is given as
½ ¾
1 2
s(to ) = cos 2π (fc + fd ) to + kto (5)
2
where s(to ) represent the return echo from target and fd is the shifted
in frequency cause by the Doppler effect.
From above equation, the general output of matched filter can be
written as
Z τ /2 ½ ¾ ½
1
g (to , ωd ) = K cos (ωc + ωd ) t + k (2π) t2 cos ωc (to − t)
−τ /2 2
¾
1
+ k (2π) (to − t)2 dt (6)
2
where ωc = 2πfc and ωd = 2πfd .
The closed form solution of above Equation can be obtained
through a considerable amount of trigonometric and algebraic
manipulation. The result of this calculation is given as,
½ ¾
ωd + 2πkto
n³ sin (τ − |to |)
ωd ´ o 2
g (to , ωd ) = G cos ωo + to ,
2 ωd + 2πkto
2
−τ /2 ≤ t ≤ τ /2 (7)
36 Chan, Chua, and Koo

7
x 10 Modulation Signal Real part of chirp signal

1.5 0.5

1 0

0.5 -0.5

0 0.5 1 0 0.5 1
Time in sec -5 Time in sec -5
x 10 x 10
Autocorrelation Function
0

-10

-20

-30

-40
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Time -6
x 10

Figure 1. Linear FM waveform.

where |to | is the absolute value of to . Note that the matched filter
output is in the form of sinXX (sinc X). The sidelode levels of LFM
correlation function is approximate −13.2 dB. Figure 1 shows the plots
of LFM signal, the modulation signal and its auto-correlation function.
The LFM signal is generated based on the specification of pulse width
10 µsec and bandwidth of 20 MHz.
LFM increases the bandwidth and subsequently improved the
range resolution of the signal by a factor equal to the time
bandwidth product. However, relatively high sidelobes remain
in the autocorrelation function. Such autocorrelation function is
unacceptable in some radar applications, where more than one target
is present, giving rise to echo of different amplitudes. Three major
techniques have been implemented to obtain lower sidelobes level,
i.e., weighting in time domain, frequency domain weighting and
NLFM [12, 13].
Time domain weighting is equals to the amplitude modulation
of the transmitted signal. However such implementation will lead
to reduction of transmitted power, and therefore a signal to noise
ratio loss will occur. Frequency domain weighting spectrum shaping
using well knows weighting windows such as Hann and Hamming [13].
However the implementation of weighting windows may lead to the
Progress In Electromagnetics Research, PIER 98, 2009 37

penalty of mainlobe broadening. NLFM waveform is designed such

that its matched filter response satisfies the sidelobe requirements.
Since the receiver is matched with to the signal shape, no mismatch
losses as in time domain weighting and frequency domain weighting.

3. NON-LINEAR FREQUENCY MODULATION (NLFM)

WAVEFORM

NLFM is a general class of continuous phase coding in which the sweep

rate is not restricted to a constant as compare with LFM. For an
arbitrary FM modulation signal,
s(t) = exp[jφ(t)] (8)
where φ(t) is the frequency modulation function. Differential the phase
modulation function will obtain f (t), the instantaneous frequency
of the NLFM signal. Figure 2 shows one of the example of
NLFM waveform; the Cosine Modulation signal, its waveform and its
autocorrelation function.
NLFM signal has been investigated in an effort to overcome the
mismatch loss but still achieve low auto correlation sidelobe levels.
However, it was found that most of the NLFM signal is not able to

7
x 10 Modulation Signal Real part of chirp signal

1.5 0.5

1 0

0.5 -0.5

0 0.5 1 0 0.5 1
Time in sec -5 Time in sec -5
x 10 x 10
Autocorrelation Function

-5
-10
-15
-20
-25
-30
-1.5 -1 -0.5 0 0.5 1 1.5
Time -6
x 10

Figure 2. Cosine modulation.

38 Chan, Chua, and Koo

provide sufficient sidelobe reduction. Only special designed NLFM

modulation signal is able to obtained lower sidelobe levels. Precision
NLFM signal are more difficult to design. Windowing or Weighting
function has been design and used to reduce the sidelobes [14–17].

3.1. Simple Two Stages NLFM

A simple two stage NLFM signal has been investigated. This NLFM
modulation function has two LFM stages as shown in Figure 3. It can
be divided into two parts which has two distinct LFM sweep rate. The
frequency of each part is linearly swept through the given time frame.
The instantaneous frequency of this NLFM can be written as,

 B
 1 t, 0 ≤ t ≤ T1
f (t) = T1 (9)
 B
 B1 + 2 t, T1 ≤ t ≤ T2
T2
Thus, the phase of this two stages NLFM signal can be derived by
integrating Equation (9).


 B t2
Z  1× , 0 ≤ t ≤ T1
φ(t) = f (t)dt = T1 2 (10)

 B2 t2
 B1 × t + × , T1 ≤ t ≤ T2
T2 2
A few simulations have been carried out for T = 10 µs and
B = 20 MHz with different values of T1 , T2 , B1 and B2 . Several

Frequency

B2

B1

Time, t
T1 T2

Figure 3. Modulation signal for simple two stages NLFM waveform.

Progress In Electromagnetics Research, PIER 98, 2009 39

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 4. Two stages NLFM waveform response for configuration 1.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 5. Two stages NLFM Waveform response for configuration 2.

40 Chan, Chua, and Koo

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 6. Two stages NLFM waveform response for configuration 3.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 7. Two stages NLFM waveform response for configuration 4.

Progress In Electromagnetics Research, PIER 98, 2009 41

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 8. Two stages NLFM waveform response for configuration 5.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 9. Two stages NLFM waveform response for configuration 6.

42 Chan, Chua, and Koo

potentially suitable functions have been investigated with varying the

parameters stated above. Figure 4 to Figure 9 show some of the
promising functions and their corresponding autocorrelation functions
output. The highest sidelobe level for each of the configuration is listed
in Table 1.

Table 1. Highest sidelobe level for two-stages LFM signal.

Configuration 1 2 3 4 5 6
Sidelobe
−16.15 −14.88 −16.8 −17.7 −15.4 −15.4
Level (dB)

3.2. Modified Two Stage NLFM Signal

Instead of LFM sweep for both segments, one of them can be
modulated using polynomial function of order 3 to 5. Two possible
NLFM modulation signals are shown in Figure 10. It is a combination
of LFM chirp signal followed by a polynomial function or LFM chirp
signal after a polynomial function. Both configurations are able to
produce promising low sidelobes as shown in Figure 11 to Figure 16.

Table 2. Highest sidelobe level for modified two stages LFM signal.
Configuration 1 2 3 4 5 6
Sidelobe
−15.65 −15.4 −19.2 −16.8 −14.78 −16.5
Level (dB)

Frequency

B1

B2

Time, t
T1 T2
T
(a)
Progress In Electromagnetics Research, PIER 98, 2009 43

Frequency

B1
B

B2

Time, t
T1 T2

T
(b)
Figure 10. Modulation signal for modified two stage LFM signal.
6 Frequency
x 10 Real part of chirp signal
1

15 0.5

10 0

5 -0.5

0 -1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 11. Modified two stages NLFM waveform response for

configure 1.

All of them are simulated with T = 10 µs and B = 20 MHz but with

different LFM sweep rates and polynomial functions. Table 2 lists the
corresponding sidelobe level of some potentially NLFM functions which
may apply to modern radar system for the aid of better performance.
44 Chan, Chua, and Koo

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 12. Modified two stages NLFM waveform response for

configure 2.
6
x 10 Frequency Real part of chirp signal
1

15 0.5

10 0

5 -0.5

0 -1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 13. Modified two stages NLFM waveform response for

configure 3.
Progress In Electromagnetics Research, PIER 98, 2009 45

7
x 10 Frequency Real part of chirp signal
2 1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 14. Modified two stages NLFM waveform response for

configure 4.
7
x 10 Frequency Real part of chirp signal
2 1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 15. Modified two stages NLFM waveform response for

configure 5.
46 Chan, Chua, and Koo

6
x 10 Frequency Real part of chirp signal
1

15 0.5

10 0

5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 16. Modified two stages NLFM waveform response for

configure 6.

3.3. Tri-stages NLFM

Beside the possible configurations shown in Section 3.1 and Section 3.2,
another possible FM modulation i.e., Tri-stages NLFM modulation
signal can be applied in order to suppress the sidelobe level. Its general
waveform is shown in Figure 17, which can be subdivided into three
stages. Each of the segments is linearly sweep with a particular chirp
rate. It composes of two LFM segments that increase the modulation
rate at the leading and trailing edges of the waveform.
The instantaneous frequency of this NLFM can be written as,

 B1

 t, 0 ≤ t ≤ T1

 T1 B

2
f (t) = B1 + t, T1 ≤ t ≤ T2 (11)

 T2

 B

 B1 + B2 + 3 t, T2 ≤ t ≤ T3
T3
Thus, the phase of this two stages NLFM signal can be derived by
Progress In Electromagnetics Research, PIER 98, 2009 47

Frequency

B3

B
B2

B1

T1 T2 T3 Time, t

Figure 17. Modulation signal for Tri-stages NLFM waveform.

Table 3. Highest sidelobe level for Tri-stages LFM signal.

Configuration 1 2 3 4 5 6
Sidelobe
−14.26 −17.37 −17.35 −19.2 −18.45 −17.13
Level (dB)

integrating Equation (10).


 B 1 t2

 × , 0 ≤ t ≤ T1

Z  T1
 2
2
B2 t
φ(t) = f (t)dt = B1 × t + × , T1 ≤ t ≤ T2 (12)

 T2 2

 2
 B1 × t+B2 × t+ B3 × t , T2 ≤ t ≤ T3

T3 2
Simulations have been done with various configurations. It can be
shown that the tri-stage NLFM waveform is able to produce lower
sidelobe as compared to conventional LFM signal. Some of the
simulation results are shown in Figures 18–23. It can be observed
that some of the configurations are able to suppress the sidelobe for
better radar target detection. Table 3 listed the highest sidelobe level of
autocorrelation function for some potentially tri-stage NLFM function.
48 Chan, Chua, and Koo

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

-5
x 10

Figure 18. Tri-stages NLFM waveform response for configuration 1.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

-5
x 10

Figure 19. Tri-stages NLFM waveform response for configuration 2.

Progress In Electromagnetics Research, PIER 98, 2009 49

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

-5
x 10

Figure 20. Tri-stages NLFM waveform response for configuration 3.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 21. Tri-stages NLFM waveform response for configuration 4.

50 Chan, Chua, and Koo

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-50
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-5
x 10

Figure 22. Tri-stages NLFM waveform response for configuration 5.

7
x 10 Frequency Real part of chirp signal
1

1.5 0.5

1 0

0.5 -0.5

-1
0 0.5 1 0 0.5 1
Time -5 Time -5
x 10 x 10
0

-10

-20

-30

-40

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

-5
x 10

Figure 23. Tri-Stages NLFM waveform response for configuration 6.

Progress In Electromagnetics Research, PIER 98, 2009 51

4. CONCLUSION

This paper presents the detailed description of LFM signal in term of

match filter response and pulse compression. The main characteristics
of LFM waveform include linear frequency ramp, flat topped spectrum
and its autocorrelation function can be approximated by a sinc function
with −13.3 dB sidelobes.
The Two-stages LFM functions are attractive since they are
capable of reducing sidelobes level with simple implementation scheme.
The simulation results show an improvement of 1.58 dB to 4.4 dB
depends on the configuration applied. It has been shown that
the Modified Two-stages NLFM signal capable of achieving better
sidelobe reduction as compared to simple Two-stages NLFM signal.
A highest sidelobe suppression of −19.2 dB can be achieved. Beside,
the Tri-stages NLFM signal is investigated with various configuration
and the simulation results show the autocorrelation function exhibits
attenuated sidelobes to less than −19 dB in one of the NLFM waveform.
In summary, the NLFM has been demonstrated to be an
effective technique for sidelobes suppression. More importantly, its
implementation scheme is simple to achieve without any SNR-robbing
sidelobe filtering or window functions.

REFERENCES

1. Chan, Y. K. and V. C. Koo, “An introduction to synthetic

aperture radar (SAR),” Progress In Electromagnetics Research B,
Vol. 2, 27–60, 2008.
2. Chan, Y. K. and S. Y. Lim, “Synthetic Aperture Radar (SAR)
signal generation,” Progress In Electromagnetics Research B,
Vol. 1, 269–290, 2008.
3. Skolnik, M. I., Radar Handbook, McGraw-Hill, New York, 1970.
4. August, W. R., High Resolution Radar, Artech House, Boston,
1996.
5. Fowle, E. N., “The design of FM pulse compression signals,” IEEE
Transactions on Information Theory, Vol. 10, No. 1, 61–67, 1964.
6. Mahafza, B. F., Introduction to Radar Analysis, CRC Press, New
York, 1998.
7. Keeler, R. J. and C. A. Hwang, “Pulse compression for weather
radar,” Record of the IEEE 1995 International Radar Conference,
529–535, 1995.
8. Barton, D. K. Modern Radar System Analysis, Artech House,
Norwood, MA, 1988.
52 Chan, Chua, and Koo

9. Skolnik, M. I., Introduction to Radar Systems, McGraw-Hill, New

York, 1980.
10. Cook, C. E. and M. Bernfeld, Radar Signals: An Introduction to
Theory and Application, Artech House, Boston, 1993.
11. Jakowatz, C. V., D. E. Wahl, P. H. Eichel, D. C. Ghiglia, and
P. A. Thompson, Spotlight-mode Synthetic Aperture Radar: A
Signal Processing Approach, Kluwer Academic Publishers, Boston,
1996.
12. Divito, A., A. Farina, G. Fedele, G. Galati, and F. Studer,
“Synthesis and evaluation of phase codes for pulse compression
radar,” Rivista Tecnica Selenia, Vol. 9, No. 2, 12–24, 1985.
13. Harris, E. J., “On the use of windows for harmonic analysis with
the discrete fourier transform,” Proc. IEEE, Vol. 66, No. l, 51–83,
1978.
14. Morgan, D. P., “Nonlinear chirp radar waveforms with improved
tolerance to doppler shifts,” IEE Colloquium on Physics and
Device Applications of Acoustic Waves (Digest No. 69), 8/1-5,
London, UK, May 22, 1987.
15. Collins, T. and P. Atkins, “Nonlinear frequency modulation
chirps for active sonar,” IEE Proceedings — Radar, Sonar and
Navigation, Vol. 146, No. 6, 312–16, Dec. 1999.
16. Johnston, J. A. and A. C. Fairhead, “Waveform design and
doppler sensitivity analysis for nonlinear FM chirp pulses,” IEE
Proceedings F — Communications, Radar and Signal Processing,
Vol. 133, No. 2, 163–175, Apr. 1986.
17. Varshney, L. R. and D. Thomas, “Sidelobe reduction for matched
filter range processing,” Proceedings of the 2003 IEEE Radar
Conference, 446–451, 2003.
18. Doerry, A. W., “SAR processing with non-linear FM chirp
waveforms,” Sandia Report, Sandia National Laboratories, 2006.