Sparsity-Driven Moving Target Detection in Distributed Multistatic FMCW Radars

TL;DR

This paper introduces a scalable, sparsity-based framework for detecting moving targets in distributed FMCW radar systems, achieving high-resolution parameter estimation with reduced computational complexity.

Contribution

It presents a generic, low-complexity detection scheme that effectively handles non-static targets sharing common support across multiple radar signals.

Findings

Achieves fast, high-resolution moving target detection.

Outperforms previous sparsity-driven methods in accuracy.

Demonstrates robustness through extensive simulations.

Abstract

We investigate the problem of sparse target detection from widely distributed multistatic \textit{Frequency Modulated Continuous Wave} (FMCW) radar systems (using chirp modulation). Unlike previous strategies (\emph{e.g.}, developed for FMCW or distributed multistatic radars), we propose a generic framework that scales well in terms of computational complexity for high-resolution space-velocity grid. Our approach assumes that \emph{(i)} the target signal is sparse in a discrete space-velocity domain, hence allowing for non-static target detection, and \emph{(ii)} the resulting multiple baseband radar signals share a common support. By simplifying the representation of the FMCW radar signals, we propose a versatile scheme balancing complexity and detection accuracy. In particular, we design a low-complexity, factorized alternative for the Matching Pursuit algorithm leveraging this simplified model, as well as an iterative methodology to compensate for the errors caused by the model simplifications. Extensive Monte-Carlo simulations of a K-band radar system show that our method achieves a fast estimation of moving target's parameters on dense grids, with controllable accuracy, and reaching state-of-the-art performances compared to previous sparsity-driven approaches.