Radar’s range is determined in part by its radiated pulse length T, and its resolution is determined by its radiated bandwidth B. But in simple systems, B*T=1, which means that good range and resolution are mutually exclusive.
Pulse compression is a method of breaking this unwanted constraint between range and resolution. Pulse compression radar transmits a modulated pulse which is both long (i.e. good range) and wideband (i.e. good resolution). The pulse compression receiver compresses the long received signal of length T into a short signal of width 1/B. It does this by delaying each sub-part of the input signal spectrum different amounts so that each sub-part arrives at the output at the same instant. The pulse compression ratio is T/(1/B) or B*T.
Matched Filtering is a generalization of pulse compression which states that given a received signal s(t) and spectrum S(f), the Matched Filter is that receiver filter which maximizes the output signal to noise ratio, and it has impulse response s(-t) and frequency response conjugate(S(f)).
Pulse compression subsystems consist of a transmitter waveform generator IF module, either digital or based on a gated limited SAW impulse response, and a unity gain matched filter receiver IF module. A matched filter may give sidelobes so large that they mask a small target in the presence of a large target, so the receiver is often purposely mismatched for sidelobe suppression. Receivers can also have multiple amplitude &/or phase matched &/or tracking channels.
There are several types of pulse compression identified by the modulation used on the transmitted pulse:
LINEAR FREQUENCY MODULATION (LFM) was the earliest and is still the most common type of pulse compression. A LFM Dispersive Filter with flat impulse response is used for the generation of the transmitted pulse. A LFM Dispersive Filter with opposite chirp rate is used in the pulse compression receiver, which is usually mismatched for sidelobe suppression by adding Taylor or Hamming amplitude weighting to the frequency response, at the cost of typically 50% pulse broadening and 1.5 dB signal/noise loss.
NON-LINEAR FREQUENCY MODULATION (NLFM) is used when the mis-matched LFM filter with its attendant pulse broadening and signal/noise loss is unacceptable. The NLFM receiver is a true matched filter. The price is Doppler sensitivity: the maximum allowed Doppler is approximately .1/T.
PSEUDO-RANDOM NOISE MODULATION (PN) is used when FM will not meet system requirements. A SAW matched filter receiver may be used, but sidelobes and Doppler sensitivity are generally large.