Efficient method of M-PSK demodulation based on particle filtering

Abstract

An efficient particle filtering algorithm for demodulation of M-PSK signals at the background of non-Gaussian noise is proposed. The state-space model of the observation signal is formulated including the dynamics of channel parameters' updating. The resulting estimation of informative symbols and channel parameters is done in two parallel contours. The simulations for QPSK signals have shown that for a sufficiently high number of particles the proposed method outperforms classical demodulation approach based on Gardner and Costas loops.

Method for the single-channel blind separation based on particle filtering

“…The modeling of non-Gaussian noise was done as in [4]. Different numbers of particles were tried: 500  .…”

“…In this paper, we propose the method for the single-channel blind separation of BPSK (binary phase shift keying) signals based on the particle filtering which is the generalization of demodulation algorithm presented in [4]. The key idea of particle filters is to use a stochastic grid approximation for the conditional probability distribution of the symbols with particles (values of the grid) evolving randomly in time (see, e.g.…”