An efficient ML algorithm for localizing closely-spaced sources by passive sensor array

The authors present a novel scheme called maximum log-likelihood-sum (MLSUM) to simultaneously determine the number of closely spaced sources and their locations by uniform linear sensor arrays. Based on the principle of the maximum likelihood (ML) estimator and a orthogonal-projection decomposition technique, the multivariate log-likelihood maximization problem is transformed into a multistage one-dimensional log-likelihood-sum maximization problem. The global-optimum solution of the ML localization is obtained by simply maximizing the single one-dimensional log-likelihood function. This algorithm is applicable in the case of coherent sources and in the case of incoherent sources. The performance of the MLSUM algorithm is compared to that of MUSIC for several simulation examples