Blind particle filtering for detection in a time-varying frequency selective channel with non-Gaussian noise

In this paper, we present efficient particle filtering and smoothing algorithms to solve the problem of blind detection in a time-varying frequency selective channel with additive non-Gaussian noise. The proposed algorithms are efficiently implemented via a combination of the optimal importance distribution and the principle of Rao-Blackwellization. The proposed particle smoothing algorithms which results in significantly improved performance, employ the method of delayed sampling, delayed weights, or a combination of the former. Simulation results are provided to illustrate the effectiveness of the proposed algorithms.