Improving sequential monte carlo blind equalization in OFDM for sparse multipath channles
Xunyi Yu, Ping Lin, Zhiqiang He, Weiling Wu · 2007
We apply Sparse Bayesian learning (SBL) to blind OFDM equalization for sparse multipath channels in this paper. We present an improved SBL sequential Monte Carlo (SMC) blind equalizer and a low complexity Sparse Bayesian blind equalizer in OFDM systems. By assuming a parameterized channel prior and using EM algorithm to estimate the parameter from the Monte Carlo sample, the algorithm could produce an accurate sparse channel estimate. Better channel estimate makes the trial distributions in the original SMC more accurate and improves the BER performance. Based on the observation that increasing the particle number to sample the signal space is inefficient, we proposed a novel low complexity Sparse Bayesian blind equalizer. Simulation results show that both algorithms outperform original SMC algorithms significantly under sparse channel conditions.