Estimating Multiple Target Locations in Multi-Path Environments

This paper addresses the problem of multi-target localization with distributed multiple-input-multiple-output (MIMO) radar systems without angular resolution capability-i.e., only based on the measurements of the signal propagation delays. In particular, we address the case that multiple targets cannot be distinguished by their “signatures” or other unique characteristics. Further complications in realistic settings arise from the following factors: 1) the direct path (DP) between transmitter/receiver and target might be blocked, and 2) indirect paths (IPs) arising from multi path propagation might be present. We present in this paper a novel algorithm that can accurately localize multiple targets even under these difficult circumstances. This algorithm is based on an iterative approximation to an exhaustive maximum likelihood (EML) estimation of the target locations, yet avoiding the exponential complexity that an EML requires. It utilizes a clustering technique that gives the number and locations of targets, and incorporates an IP detection mechanism based on the likelihood of several propagation delay measurement (PDM) correspond to DPs from the same target. Despite its lower complexity, our algorithm achieves almost the same performance as the EML in a variety of simulation settings. We finally study the choice of algorithm parameters and the configurations of the MIMO system nodes to achieve the best localization performance.