IEEE Aerospace and Electronic Systems Magazine - September 2020 - 32

Feature Article:

DOI. No. 10.1109/MAES.2020.2988331

Sidelobe Mitigation in Noise Radar Using Sparse
Signal Processing
Lilian de Martın, Wim van Rossum, Diogo Ribeiro,
Laura Anitori, Radar Technology, TNO

INTRODUCTION
Traditional pulse-Doppler radars use waveforms with a
fixed pulse repetition frequency (PRF) and identical modulation for each pulse, typically a linear frequency modulation (LFM). The advantage of a pulsed system is the
lack of Tx/Rx coupling if there is no simultaneous receive
and transmit, and the reduced dynamic range as compared
to continuous transmission, since responses from close by
targets are not received. However, the drawback of this
mode of operation is the presence of ambiguities in either
range or Doppler, or in both, which is dictated by the sampled nature of such systems. For unambiguous detection
of targets at large ranges, the PRF should be low, while
for unambiguous detection of high velocity targets the
PRF should be high. For medium PRF waveforms, next to
the ambiguities, blind velocities appear for those Doppler
frequencies that fold into the Doppler bins with clutter
present. Also, blind ranges appear due to the switching off
of the receiver during the transmission.
An option to estimate the range and velocity unambiguously is using PRF staggering. Here, multiple bursts are
transmitted in a sequence with different PRFs. Since each
burst samples the ambiguous range and velocity differently, the range and Doppler can be unambiguously estimated by using the principle of the Chinese remainder
theorem. The integration over the bursts is performed
incoherently with an M-out-of-N detector, resulting in
a total dwell time that is N times longer than the time of
a single burst. Ideally, the time needed for unambiguous
detection of targets would be minimized by having a

Authors' current address: Lilian de Martın, Wim van
Rossum, Diogo Ribeiro, and Laura Anitori, Radar
Technology, TNO, The Hague, The Netherlands
(e-mail: lilian.demartin@tno.nl).
Manuscript received September 6, 2019, revised February
19, 2020, April 1, 2020, and ready for publication April 14,
2020.
Review handled by Andy Stove.
0885-8985/20/$26.00 ß 2020 IEEE
32

waveform that allows for unambiguous detection in range
and velocity in a single burst, thus reducing the time on
target needed to detect it.
A way to achieve unambiguous range and Doppler
measurements with a single burst is to use a modulation that
is not repeating over time, as it is the case for noise-like
waveforms [1]. Such waveforms can be varied from pulse
to pulse, if used in combination with a pulsed radar or generated without any repetition for continuous wave radar.
Apart from their ability to suppress ambiguities, continuous
noise waveforms exhibit a low peak power, decreasing their
probability of interception and increasing their capability
for electronic countercountermeasure [2], [3].
However, a major drawback of noise waveforms is the
noise-like behavior of the sidelobes that appear after
matched filter (MF) processing [4]-[6]. For the case of
continuous-wave noise radar, these sidelobes are distributed over the entire range-Doppler plane with an average
level of 1/N, where N is the number of independent samples in a burst [4]. As a result, weak targets will possibly
be masked by the sidelobes of stronger targets, independently of the distance between targets. This is in contrast
to LFM waveforms, in which sidelobes are concentrated
in range and Doppler space and only a limited number of
range-Doppler cells are affected by them. However, it
should be noted that sidelobes are an artifact of the signal processing that is applied on receive-typically linear
filtering-and can be minimized or suppressed by using
advanced processing techniques [5] and waveform design
[7], [8]. On the contrary, ambiguities (or equivalently,
grating lobes) are an inherent property of the transmitted
waveform and cannot be removed by signal processing,
since they are identical to, and hence indistinguishable
from, the true target response.
The problem of sidelobe suppression on receive in
noise radar has been broadly studied in the literature.
Methods include using adaptive lattice filters [9], [10],
stretch processing [11], [12], apodization filtering [13],
inverse filtering [14], CLEAN algorithms [15], [16], or
subtracting a modeled version of the target echo from the
received signal [17]. If the scene is sparse, it is also

IEEE A&E SYSTEMS MAGAZINE

SEPTEMBER 2020

IEEE Aerospace and Electronic Systems Magazine - September 2020

Table of Contents for the Digital Edition of IEEE Aerospace and Electronic Systems Magazine - September 2020