We propose a maximum likelihood (ML) approach for tracking the direction-of-arrival (DOA) of multiple moving targets by a passive array. A locally linear model is assumed for the target motion, and the multiple target states (MTSs) are defined to describe the states of the target motion, The locally linear model is shown to be strongly locally observable almost everywhere. The approach is to estimate the initial MTS by maximizing the likelihood function of the array data. The tracking is implemented by prediction through the target motion dynamics using the initial MTS estimate. By incorporating the target motion dynamics, the algorithm is able to eliminate the spread spectrum effects due to target motion. A modified Newton-type algorithm is also presented, which ensures fast convergence of the algorithm. Finally, numerical simulations are included to show the effectiveness of the proposed algorithm.
展开▼
