Robust and Efficient Structure-Based Radar Receive Processing

Abstract

Legacy radar systems largely rely on repeated emission of a linear frequencymodulated (LFM) or chirp waveform to ascertain scattering information from an environment.The prevalence of these chirp waveforms largely stems from their simplicityto generate/process and robustness to physical hardware distortion. However,this traditional design philosophy may lack the flexibility and dimensionality neededto address the dynamic “complexification” of the modern radio frequency (RF) environmentor achieve current operational requirements where unprecedented degrees ofsensitivity, maneuverability, and adaptability are necessary.Over the last couple of decades analog-to-digital and digital-to-analog technologieshave advanced exponentially, resulting in tremendous design degrees of freedomand arbitrary waveform generation (AWG) capabilities that enable sophisticated designof emissions to better suit operational requirements. However, radar transmitters(TX) typically require high-power amplification (HPA) to contend with two-way propagationloss. Thus, transmitter-amenable waveforms are effectively constrained to beboth spectrally contained and constant amplitude, resulting in a non-convex NP-harddesign problem. While isolating the global optimal can be intractable for even modesttime-bandwidth products (TB), locally optimal TX-amenable solutions that are “goodenough” are often readily available. However, traditional matched filter (MF) basedestimation may no longer satisfy operational requirements with sub-optimal emissions.Using knowledge of the radar TX-to-receiver (TX-RX) chain, a discrete linearmodel can be formed to express the relationship between observed measurementsand the complex scattering of the environment. This structured representation enablesmore sophisticated least-square (LS) and adaptive estimation techniques that can designedto better satisfy operational requirements, improve estimate fidelity, and extenddynamic range. However, receive dimensionality can be enormous as the aggregatedegrees of freedom (DoF) for a coherent processing interval (CPI) is multiplicativewith respect to bandwidth, pulse duration, and number of pulses. Brute force implementationof model-based estimation techniques that leverage these DoFs may haveunwieldy computational burden on even the most cutting-edge signal processing hardware.Additionally, a discrete linear representation is fundamentally an approximationof a dynamic and continuous physical reality where model errors may induce bias, createfalse detections, or restrict effective dynamic range. As such, these structure-basedapproaches must be both computationally efficient and robust to reality.Here discrete receive models and algorithms for pulsed-radar range-Dopplerestimation are introduced. Modifications and alternative solutions are then proposedto improve estimate fidelity, reduce computational complexity, and enhance practicalrobustness.