Pilot Pattern Optimization for Sparse Channel Estimation in OFDM Systems

Han Wang, Qing Guo, Gengxin Zhang, Guangxia Li, Wei Xiang

Abstract: Compressive sensing (CS) based sparse channel estimation requires optimal pilot patterns, whose corresponding sensing matrices should have small mutual coherences, so as to efficiently exploit the inherent channel sparsity. For the purpose of minimizing the mutual coherence of the sensing matrix, we introduce a new estimation of distribution algorithm (EDA) to optimize the pilot pattern so as to improve the channel estimation performance. The proposed scheme guides the optimization process by building and sampling the probability distribution model of the promising pilot indexes, and approaches the optimal pilot pattern iteratively. The algorithm is able to not only preserve the current best pilot indexes, but also introduce diversity by sampling new ones, and hence is unlikely to trap into local minima and more robust than other methods. Simulation results show that our proposed method can generate sensing matrices with smaller mutual coherences than existing methods, and the corresponding optimized pilot pattern performs well in terms of sparse channel estimation.