2D DOA Estimation Using Sparse Representation of Higher-Order Power of Covariance Matrix

Abstract

A novel two-dimensional (2D) direction-of-arrival (DOA) estimation algorithm utilizing a sparse signal representation of higher-order power of covariance matrix is proposed. Through applying the higher-order power of covariance matrix to construct a new sparse decomposition vector, this algorithm avoids the estimation of incident signal number and eigenvalue decomposition. And the hierarchical granularity-dictionary is studied, which forms the over-complete dictionary adaptively in the light of source signals’ distribution. Compared with MUSIC and L1-SVD, this algorithm not only provides a better 2D DOA performance but also possesses the capability of coherent signals estimation. Theoretical analysis and simulation results demonstrate the validity and robust of the proposed algorithm.