DOA estimation for unknown number of signals using particle swarm optimization

In this paper, the direction of arrival estimation for unknown number of signals is formulated as a constrained optimization based on the Maximum Likelihood criterion. For that, we propose a fitness function that depends on the number of signals, array size and the measured signal at the array output such that the correct combination of these parameters minimizes the fitness. To minimize the fitness, we provide an algorithm based on the hybridization of the particle swarm optimization. In this algorithm, an initial estimate of the parameters is obtained using the real valued optimizer then the exact parameters are resolved using the binary optimizer. Simulation results have shown that the proposed algorithm provides higher probability of detection compared to that of the conventional techniques used to deal with the current problem. In addition to that the algorithm shows low sensitivity to the maximum number of signals assumed to be incident on the array.