DOA Estimation Using Log Penalty under Large Arrays

This paper proposes a new direction-of-arrival (DOA) estimation algorithm, which is suitable for the scenario that the number of sensors is large, and is comparable with the number of samples in magnitude. Instead of utilizing classical subspace technique, sparse-recovery-based approach with log penalty is exploited. In detailed implementation, we use DC (Difference of Convex function) decomposition to solve the non-convex optimization problem, and weighted L-1-norm penalty to provide the initial estimation, where the weights are constructed via the orthogonality between the noise subspace and signal subspace in large-scale random matrix theory framework. As a result, an improved DOA estimation performance is achieved. Simulation results validate the effectiveness of the proposed algorithm.