An improved approach of blind source separation for delayed sources using taylor series expansion

This paper proposes blind source separation (BSS) method to retrieve the signals of wireless communication environment (WCE) mainly based on the statistic properties of the wireless channel model. First, a statistical model for three-dimensional (3-D) spatial domain channel and its analytical expression of the probability density functions (PDF) of the time of arrival (TOA) is introduced. Then, BSS method based on second-order blind identification (SOBI) is proposed to address the separation problem of delayed sources based on the statistical channel model. The improved approach is much efficient in terms of separation quality, especially when the noise is considered. The performance of this proposed method is validated through simulations, which are performed under different delay time and signal to noise ratio (SNR) conditions. The results show that this new approach not only provides a good performance when the delay is large, but also is robust in the noise environment.