Page 300 - Fundamentals of Radar Signal Processing

Choosing ΔF > 1/τ allows generation of a wide total bandwidth with fewer
pulses and therefore a shorter data collection time, but the resulting
undersampling creates aliasing that appears as extra range ambiguities (also
called grating lobes). Figure 4.42b shows a similar view of another case with
ΔF · τ = 2.5 but the other parameters unchanged. The bandwidth is now 20/τ Hz
so the resolution in delay is correspondingly finer. However, there are now five
peaks along the zero-Doppler axis, representing five range ambiguities, within ±

1 pulse length.

4.9 The Stepped Chirp Waveform

The stepped chirp waveform is a stepped frequency waveform that substitutes
an LFM constituent pulse for the constant-frequency pulse used above. It can
achieve very wideband operation without resorting to stretch processing,
thereby avoiding the restriction of short range windows. In addition, it avoids
the array frequency steering effects mentioned previously so long as the
individual pulse bandwidth is not too large.

The stepped chirp waveform can allow a large frequency step ΔF > 1/τ
without suffering the aliasing seen in the conventional stepped frequency
waveform. Careful design is needed to relate the LFM pulse bandwidth and
length to the RF step size in order to achieve effective suppression of the
ambiguities. Details and sample parameter sets are given in Levanon and

Mozeson (2004). Processing of the waveform requires individually
demodulating and matched filtering each individual pulse, and then post-
processing the ensemble to construct a new signal with the full bandwidth. This
post-processing can be performed in either the time or frequency domain.
Details are given in Keel and Baden (2012).

4.10 Phase-Modulated Pulse Compression Waveforms
The second major class of pulse compression waveforms is referred to as phase

coded waveforms. A phase coded waveform has a constant RF but an absolute
phase that is switched between one of two or more fixed values at regular
intervals within the pulse length. Such a pulse can be modeled as a collection of
N contiguous subpulses x (t) of duration τ , each with the same frequency but a
n
c
(possibly) different phase