Blind Source Separation Based on Signal Temporal Predictability

Based on the fact that the temporal predictability of signals is often different, new algorithm using second order statistics for blind source separation (BSS) is proposed in this work. The novel measure of signal temporal predictability is developed. Based on which, the proposed method is less time-consuming and requires minimal assumptions regarding the probability density functions of the source signals. Also, the online mode is provided. Simulations show that the proposed algorithm is robust with the best precision for batch mode, and its convergence is quite well for online mode.