Compressive beamforming using greedy algorithms.

Direction of Arrival (DOA) estimation is a topic of great interest in many fields like electromagnetic, seismic/geophysical and acoustic sensing. High resolution algorithms for DOA estimation have been widely used in applications like SONAR, RADAR and wireless communications to resolve closely-situated sources. However these algorithms are very sensitive to the signal to noise ratio (SNR), the number of snapshots and correlation between sources. The DOA estimation problem can be viewed as a sparse representation problem as the signals impinging on an array are intrinsically sparse in the spatial domain. Hence compressed sensing techniques can be applied in DOA estimation. This paper explores the formulation of the DOA estimation as a sparse representation problem and compares the performance of different compressive beamforming techniques. Compressive beamforming is done using l(1) minimization and greedy algorithms. It is shown that greedy algorithms are faster than l1 minimization. An improvement to greedy algorithm is proposed in this paper so that prior information about sparsity is not needed. Resolution is further improved using multiband signals. Thus the study shows that compressive beamforming can give 2 to 3 degree resolution from single snapshot which is better than existing methods.