Randomized Multiple Candidate Iterative Hard Thresholding Algorithm for Direction of Arrival Estimation

The sparse recovery problem by l0 minimization which is of central importance in compressed sensing (CS)-based algorithms for direction of arrival (DoA) estimation has attracted considerable interest recently. This paper proposes a greedy algorithm called randomized multiple candidate iterative hard thresholding (RMC-IHT) which generates a set of potential candidates using the iterative hard thresholding algorithm and selects the best candidate based on the a priori knowledge of the distribution of the signal and noise matrices. We also consider the case of correlated sources and develop a version of RMC-IHT for this scenario. Simulation results illustrate the improvement achieved by RMC-IHT.