Multi-Target Detection in FMICW Radar Using Machine Learning

Abstract

This study proposes an effective method for detecting multiple targets in frequency-modulated interrupted continuous wave (FMICW) radar by using periodic signal characteristics and applying machine learning. FMICW radars are typically used for detecting a single target, but this method overcomes the challenges of range-Doppler coupling and spurious wave generation caused by the switching of transmit and receive signals, to detect multiple targets. The method first uses FMICW radar to generate an RD-Map by applying MLE (maximum likelihood estimation) , and then enhances the targets by identifying the intersection of lines generated by range-Doppler coupling. Thus, by applying a machine-learning-based object detection algorithm to the enhanced images, multiple target detection becomes possible.