Algorithm Research On LFMCW Radar Multiple Targets Detection

The Linear Frequency-Modulated Continues Wave (LFMCW) radar, with its no blind area, high range resolution, low transmit power and low probability of intercept (LPI), has been an important topic in the radar research field and attracted more and more attention recently. The problem of LFMCW radar target detection, in essential, is the beat signal processing. The beat signal model of the stationary, uniform motion and accelerated motion target is sinusoid, Linear Frequency-Modulated (LFM) signal and polynomial-phase signal (PPS) respectively. The information of the target is hidden in the parameters of its beat signal. There still remain some problems on the LFMCW radar target detection, especially on the multiple targets detection under complex background. In this dissertation, algorithms of LFMCW radar multiple targets detection are studied mainly.The following is the summarization of the main work in this paper:1. With respect to the multiple targets detection the beat signal of which is sinusoid, some classical frequency estimation methods based on DFT are discussed and their performance is compared. A new multiple targets detection algorithm in all kind of noise based on modified Genetic Algorithms (GA) is proposed.2. With respect to the multiple targets detection the beat signal of which is LFM signal, several typical time-frequency analysis methods are reviewed and compared the performance of three multiple component LFM signal detection means, Radon-Wigner,Radon-STFT and Radon-Ambiguity-FRFT, by experiments. More over, for the Symmetrical-alpha Stable (SαS) noise in which the algorithms based on second-order or high-order moments degrade greatly and cannot work normally, a new technique grounded on fractional lower order moment is proposed.3. With respect to the multiple targets detection the beat signal of which is PPS, methods based on High-order Ambiguity Function (HAF) and Product High-order Ambiguity Function (PHAF) is introduced mainly. With this background, a novel method based on estimation one by one and"CLEAN"technique is proposed. It overcome the demerit of HAF and PHAF, could detect multiple component PPS effectively.