Source Enumeration Based on Spatial Correlation Function for Independent/Dependent Sources

Abstract

The detection of the number of sources when the sources may be dependent and more than the sensors is a challenging problem. This paper proposes a new method to address this problem, which is mainly based on the spatial correlation function and the Gerschgorin disk estimator (GDE). Compared to the fourth-order cumulant-based source enumeration methods presented recently, the proposed method requires much fewer samples to accurately estimate the source number and can work well even when the sources are dependent. Simulation results show that the proposed method possesses superior detection performance over the existing methods for source enumeration under an unbalance noise environment, and testify the effectiveness of the proposed algorithm for both the independent and dependent sources.