2D spectral estimation for unambiguous detection of automotive radar targets

Abstract

Automotive radar systems operating in urban areas phase a hard challenge detecting relevant targets since their signals are corrupted by strong clutter signals from the ground and surrounding buildings and because dense target scenarios may cause ambiguities. The combined FSK-LFMCW waveform by Rohling solves the range/velocity ambiguity within a short measurement time but the FFT used in the processing have limited resolution and large side lobe levels that are troublesome in dense target situations. In this paper we use the FSK-LFMCW waveform in an array system and define the target detection and parameter extraction as a 2D frequency estimation problem. We compare the performance of the non-parametric Capon and APES estimators with the regular FFT processing. The results show that we obtain better accuracy for the estimated target parameters and that we have a larger probability of target detection and a much smaller false alarm ratio.