Robust blind identification algorithm in stationary noise environments

Abstract

Blind channel identification aims to extract the channel impulse response with little prior statistical knowledge of the source signal. In this study, a novel blind channel identification algorithm is proposed based on second-order statistics. This algorithm utilises the special structure of the covariance matrix to identify the channel impulse responses. The proposed algorithm also outperforms the conventional blind identification algorithm with a low signal-to-noise ratio (SNR) and a few observations. Theoretically, the performance of this algorithm is independent of the SNR. This algorithm can be used in both white or coloured noise environments provided that the noise is stationary. This algorithm is robust to parameter estimation errors and has a simple and efficient implementation. Simulation examples are provided to demonstrate the superior performance of this algorithm.