Page 39 - Fundamentals of Radar Signal Processing

are present. Resolution and sidelobes in range are determined by the radar

waveform, while those in angle are determined by the antenna pattern.
In radar tracking, the basic figure of merit is accuracy of range, angle, and
velocity estimation. While resolution presents a crude limit on accuracy, with
appropriate signal processing the achievable accuracy is ultimately limited in
each case by the SIR.
In imaging, the principal figures of merit are spatial resolution and dynamic

range. Spatial resolution determines what size objects can be identified in the
final image, and therefore to what uses the image can be put. For example, a
radar map with 1 km by 1 km resolution would be useful for land use studies,
but useless for military surveillance of airfields or missile sites. Dynamic range
determines image contrast, which also contributes to the amount of information
that can be extracted from an image.
The purpose of signal processing in radar is to improve these figures of

merit. SIR can be improved by pulse integration. Resolution and SIR can be
jointly improved by pulse compression and other waveform design techniques,
such as frequency agility. Accuracy benefits from increased SIR and
interpolation methods. Sidelobe behavior can be improved with the same
windowing techniques used in virtually every application of signal processing.
Each of these topics are discussed in the chapters that follow.

Radar signal processing draws on many of the same techniques and
concepts used in other signal processing areas, from such closely related fields
as communications and sonar to very different applications such as speech and
image processing. Linear filtering and statistical detection theory are central to
radar’s most fundamental task of target detection. Fourier transforms,
implemented using fast Fourier transform (FFT) techniques, are ubiquitous,
being used for everything from fast convolution implementations of matched

filters, to Doppler spectrum estimation, to radar imaging. Modern model-based
spectral estimation and adaptive filtering techniques are used for beamforming
and jammer cancellation. Pattern recognition techniques are used for
target/clutter discrimination and target identification.
At the same time, radar signal processing has several unique qualities that

differentiate it from most other signal processing fields. Most modern radars are
coherent, meaning that the received signal, once demodulated to baseband, is
complex-valued rather than real-valued. Radar signals have very high dynamic
ranges of several tens of decibels, in some extreme cases approaching 100 dB.
Thus, gain control schemes are common, and sidelobe control is often critical to
avoid having weak signals masked by stronger ones. SIR ratios are often
relatively low. For example, the SIR at the point of detection may be only 10 to
20 dB, while the SIR for a single received pulse prior to signal processing is

frequently less than 0 dB.
Especially important is the fact that, compared to most other DSP
applications, radar signal bandwidths are large. Instantaneous bandwidths for an