Maximum likelihood estimation of direction-of-arrival and detection for broad-band sources via dynamic programming

An analysis is presented to estimate the number of broadband sources by sensor arrays and to compute the maximum likelihood estimator (MLE) of their direction-of-arrival (DOA) based on the dynamic programming algorithm. The approach is devised in the frequency domain. With proper selection of the number of elements, the distance between elements, and the step size, the global maximum for the log-likelihood function can be obtained. Otherwise, the global maximum is not guaranteed. Simulation results for broadband sources that demonstrate the performance of the proposed method are presented and compared with that of the MUSIC algorithm which employs coherent signal-subspace processing