Unitary Root-MUSIC Combined with Subspace Modification for Circular Microphone Array

Abstract

A unitary root signal classification (Root-MUSIC) combined with subspace modification for uniform circular microphone arrays is proposed. Firstly, the subspace is modified by eliminating the two undesirable terms in the covariance matrix, which cause the subspace leakage, thus improving the estimation accuracy of subspace. Secondly, a real-valued Root-MUSIC algorithm for circular microphone arrays is developed by unitary transformation. The proposed method can reduce the computational complexity of complex-valued Root-MUSIC by 75%. Some simulation results demonstrate the effectiveness of the proposed method