A blind source extraction method for chaotic signals using unscented Kalman filters

The applications of chaos in wireless communication systems require tackling the blind source extraction problem. A class of existing blind source extraction methods in the framework of kalman filtering suffer great performance degeneration when applying to noisy chaotic mixtures. To address this issue, a new method is proposed. This method incorporates two unscented Kalman filters alternately estimate the extracting vector and the source signal. Simulation results indicate that the proposed algorithm outperforms existing unscented Kalman approach.