DOA estimation from sensor array measurements taken at two distinct times

This paper investigates the problem of direction of arrival estimation using the spatial response of a uniform line array of sensors at two distinct times, with the emphasis given to small KL product where K is the spatial bandwidth of the signal and L the array length. An analogy is drawn between this problem and that of time delay estimation (TDE) between shifted versions of a time signal. It is seen that the condition of long observation time needed by conventional asymptotic TDE techniques is violated when KL is small. To overcome this difficulty, an exact maximum likelihood (ML) estimator of the DOA does not rely on any asymptotic conditions. Simulation results indicate improved performance compared with an approximate ML estimator based on large KL assumption. At high signal-to-noise ratio, the proposed ML estimator is nearly optimal and important simplifications are possible