Single channel blind signal separation with Bayesian-MCMC

To solve the problem of single channel blind signal separation with unknown components number, a method based on Bayesian-MCMC is presented. First, single channel mixed signal, whose components have same general parameters, is modeled to change the problem into joint estimation of signal parameters and components number. Then, Bayesian theorem is applied in the joint estimation. Lastly, Bayesian computation of the estimation is accomplished with reversible jump MCMC. Simulation results indicate that the algorithm is practical and effective.