Interference Mitigation via MMF-MAD-ISP for Automotive FMCW Radar in Complex Environments

Abstract

Automotive radar is a key sensing modality in advanced driver-assistance systems. In dense traffic environments, mutual radar interference can significantly degrade detection performance. This article presents a robust moving mean filtering (MMF)-median absolute deviation (MAD)-iterative subspace projection (ISP) algorithm designed for realistic operating conditions, where multipath propagation of interference signals introduces chirp distortions. The proposed method exploits the observation that, after dechirping and low-pass filtering, noncoherent interference manifests as sudden amplitude spikes within individual chirps. Interference samples are first detected using an MAD criterion following MMF, and subsequently reconstructed via the ISP method. The proposed approach accommodates multiple interference sources as well as extended static reflectors, such as guardrails. Compared with existing methods, it achieves effective interference suppression in complex traffic scenarios while maintaining reasonable computational time. The performance of the method is validated through numerical simulations and experimental measurements.