Pulse compression is the elegant solution to the fundamental radar designer’s dilemma: the desire for long pulses (for high energy and thus long detection range) conflicts with the need for short pulses (for fine range resolution). By modulating the transmitted pulse and applying a matched filter on reception, pulse compression achieves the energy of a long pulse and the range resolution of a short pulse simultaneously. The range resolution is determined by the waveform bandwidth, not the pulse duration, enabling time-bandwidth products exceeding 10,000 in modern radars. This article examines the principles and practice of pulse compression.

Matched Filter Principle

The matched filter maximizes the signal-to-noise ratio at its output for a known signal in additive white Gaussian noise. Its impulse response is the time-reversed, complex-conjugated version of the transmitted waveform. When the received signal correlates with this reference, the output is the autocorrelation function of the waveform — a sharp peak at the target’s range with surrounding sidelobes determined by the waveform’s modulation characteristics. The compressed pulse width is approximately 1/B, where B is the waveform bandwidth, and the processing gain (SNR improvement) equals the time-bandwidth product BT.

Linear Frequency Modulation (LFM)

The linear FM chirp is the most widely used pulse compression waveform. Its frequency sweeps linearly across the bandwidth B during the pulse duration T. After matched filtering, the compressed pulse has a sin(x)/x shape with first sidelobes at -13.2 dB relative to the main lobe — too high for many applications where strong targets can mask weaker neighbors through range sidelobe interference. Amplitude weighting (Taylor, Hamming, Chebyshev) applied in the frequency domain suppresses sidelobes, typically achieving -35 to -45 dB peak sidelobes at the cost of 1–2 dB SNR loss and slight main lobe broadening.

LFM’s primary virtue is Doppler tolerance: a Doppler-shifted LFM chirp produces a compressed pulse at a shifted range (range-Doppler coupling), but the pulse shape and SNR are preserved. This property enables pulse-Doppler processing where the same matched filter serves all Doppler bins. For applications requiring high Doppler resolution, the range-Doppler coupling must be measured and compensated during tracking.

Nonlinear FM and Phase-Coded Waveforms

Nonlinear FM (NLFM) waveforms shape their instantaneous frequency profile to produce inherently low range sidelobes without amplitude weighting, avoiding the associated SNR loss. The frequency-vs-time profile is designed using stationary phase principles to produce the desired power spectrum shape. NLFM typically achieves -35 to -40 dB peak sidelobes with zero SNR loss — attractive for long-range surveillance where every decibel matters. The trade-off is reduced Doppler tolerance compared to LFM.

Phase-coded waveforms divide the pulse into N subpulses (chips), each transmitted with a specific phase from a code set. Barker codes achieve sidelobes of 1/N (peak-to-peak), but the longest known Barker code has only 13 chips, limiting compression ratio to about 11 dB. Polyphase codes (Frank, P1–P4, Zadoff-Chu) provide longer sequences with better Doppler tolerance. Binary phase codes (m-sequences, Gold codes) offer large compression ratios and good LPI characteristics but with higher sidelobes than optimally designed analog waveforms.

Digital Implementation

Modern pulse compression is implemented digitally using fast convolution: the received signal and matched filter reference are transformed to the frequency domain via FFT, multiplied, and inverse-transformed. This approach is computationally efficient (O(N log N) vs. O(N^2) for direct time-domain correlation) and naturally accommodates frequency-domain weighting for sidelobe control. For stretch processing, used with very wideband LFM waveforms where full fast convolution is computationally prohibitive, the received signal is mixed with a replica chirp (deramping), converting range delay to a constant frequency that is resolved by FFT.

Implementation challenges include managing the dynamic range of the compressed output (the processing gain can produce output signals exceeding 80 dB above the input noise floor) and handling eclipsing loss when strong nearby returns saturate the receiver during transmission. Advanced architectures using stretch processing, segmented correlation, and multi-rate processing address these challenges in high-performance defense radar systems.