Decentralised tracking for human target in multistatic ultra-wideband radar

A global signal processing chain for detection and localisation of moving human targets using a multistatic ultra-wideband radar system with widely separated receiving antennas is proposed. The design options for the system are presented and justified, and the required signal processing steps are summarised. Special attention is then devoted to the critical issue of target association. A slow-time feature, video time density function, is proposed to associate target responses in different receiver channels. Video time density function is extracted from range-Doppler video sequences for each target. It indicates the slow-time evolution characteristic of the moving human targets, and thus is invariant to radar observation angles (provided that the target can be detected). Auction algorithm-based global nearest neighbour approach is extended to the measurement-to-measurement association between receivers. The decentralised tracking approach proposed is experimentally verified on measured data.