Abstract :
Noisy blind source separation (BSS) is studied. The proposed method mainly consists of two stages. The first step is to enhance each mixture by particle filtering so that (as in the theory) the noisy mixtures become noise-free. The second is to extract the sources by BSS algorithms. For particle filtering, the state-space model is defined by the time-varying autoregressive model of each mixture, and the measured space is set by each observed noisy mixture. Various simulations prove that the approach is effective
Keywords :
autoregressive processes; blind source separation; particle filtering (numerical methods); signal denoising; state-space methods; BSS algorithms; noisy blind source separation; noisy mixtures; particle filtering; sequential Monte Carlo methodology; state-space model; time-varying amoregressive model;
