DOA estimation of uncorrelated, partly correlated and coherent signals using alternating oblique projection

The problem of direction-of-arrival (DOA) estimation is important in array signal processing. To estimate the DOAs of uncorrelated, partly correlated and coherent signals, a new iterative DOA estimation algorithm, named AOP-DOA, is proposed by using alternating oblique projection (AOP). In each iteration, the oblique projection approach is employed to separate the received signals, then the DOA of each separated signal is estimated one after another. After theoretical analysis on the relationship between the proposed AOP-DOA and the conventional alternating projection based maximum likelihood estimator (AP-MLE), an AOP&AP-DOA algorithm, which is a combination of AOP-DOA and AP-MLE, is developed to reduce the computational complexity ofAOP-DOA. Extensive experiments validate the effectiveness and complexity of the proposed two algorithms. Particularly, AOP&AP-DOA keeps the merits of AOP-DOA, but exhibits superiority over AOP-DOA in terms of computational complexity when proper adaptive grid refinement strategy is applied.