Estimation of the source number using array discrete wavelet transform

A crucial step in many array signal-processing applications is an accurate estimation of the unknown number of sources. Standard techniques for source number estimation often have shortcomings in practice when sensor signals are buried in additive noise with unknown parameters. In this paper we present an efficient method for detecting the number of sources from noise corrupted sensor signals by using an array discrete wavelet transform. Denoising and source number estimation are performed in the wavelet transform domain allowing subsequent processing, such as blind source separation, in a very efficient manner. Simulation results are included that illustrate the performance of the method.