WCSE 2017
ISBN: 978-981-11-3671-9 DOI: 10.18178/wcse.2017.06.149

Variational Bayesian Method for Joint Sparse Channel Estimation and Data Detection in Frequency-Selective Rayleigh-Fading Systems

Zhang Kai, Yu Hongyi, Hu Yunpeng, Shen Zhixiang

Abstract— This paper deals with the problem of data detection in sparse frequency selective Rayleighfading channels. An iterative scheme for joint channel estimation and data detection is proposed based on maximum a posteriori (MAP) criterion. We begin with a general sampled representation of the nonlinear sparse multipath channel, which is parameterized by a linear counterpart at a resolution commensurate with the sampling frequency. Then, a hierarchical system model is developed, in which priors for all model parameters, i.e., channel coefficients, symbols, and noise precision parameter, are properly chosen. In particular, the channel sparsity is utilized by defining parametric heavy-tailed priors for its coefficients. And then the optimal joint estimation framework based on MAP criterion is developed. Since direct maximization of the posterior probability density function is infeasible, the variational Bayesian method is invoked as a feasible method to simplify the problem. The estimation of the channel and data are dealt with jointly and iteratively. The superiority of the proposed schemes with respect to several typical methods is presented by both theoretical analysis and Monte Carlo simulations.

Index Terms— Sparse channel estimation, Rayleigh-Fading channels, maximum a posteriori estimation, variational Bayesian inference

Zhang Kai, Yu Hongyi, Hu Yunpeng, Shen Zhixiang
The National Digital Switching System Engineering and Technological R&C Center of China, CHINA

[Download]


Cite: Zhang Kai, Yu Hongyi, Hu Yunpeng, Shen Zhixiang, "Variational Bayesian Method for Joint Sparse Channel Estimation and Data Detection in Frequency-Selective Rayleigh-Fading Systems," Proceedings of 2017 the 7th International Workshop on Computer Science and Engineering, pp. 855-863, Beijing, 25-27 June, 2017.