Abstract:

 To reduce computation complexity in the particle-filtering based blind source separation, a novel eparation algorithm of co-frequency digital modulated mixture is proposed, which combines the particle filtering lgorithm with support vector machine. Formulae for assigning particle weights and expression for stimating a posterior distribution function are derived. Performance of the proposed algorithm is analyzed in wo aspects: statistic characteristics of the estimated posterior density function and its computation complexity. he estimated posterior density function is shown to be close to the true function. Both the theoretical nalysis and simulation results show that the proposed algorithm can reduce processing time without loss of
performance in terms of bit error probability as compared with the particle-filtering based blind source separation lgorithm.

Key words:  digital modulated mixture, ingle channel blind source separation, particle filtering, support vector achine