Cognitive FMCW Automotive Radar for Interference Suppression

Automotive radar systems are key sensors in advanced driver-assistance systems (ADAS). However, mutual interference in dense traffic environments severely degrades their performance, increasing both missed detections and false alarms. This paper proposes a cognitive frequency-modulated continuous wave (FMCW) automotive radar framework that employs a unified signal processing approach for suppressing both coherent and long-duration non-coherent interference. The method exploits a silent chirp, inserted between frames, to sense the electromagnetic environment. Interference parameters are estimated using the fractional Fourier transform (FrFT). For coherent interference suppression, the radar adaptively adjusts its transmitted chirp rate to mitigate ghost targets. Non-coherent interference is suppressed using a two-stage strategy that combines iterative orthogonal projection with iterative soft-thresholding. Experimental validation with real-world radar data demonstrates the effectiveness of the proposed framework, particularly in mitigating challenging long-duration non-coherent interference.