A blind source separation method based on Kalman filtering

According to Nonlinear Principal Component Analysis (NPCA) criterion, a blind source separation algorithm based on Kalman filtering is proposed in this paper. The convergence property of the algorithm is analyzed. The performance of the algorithm is evaluated by using several different kinds of sources. The effect of the number of iteration steps and the observation noise for the performance are investigated. The results show that this algorithm can separate chaotic as well as other sources from linear instantaneous mixtures effectively.