Due to the high complexity of the Support Agnostic Bayesian Matching Pursuit（SABMP）algorithm in the support set search

a modified Bayesian sparse channel estimation algorithm is proposed in this paper.The modified Bayesian algorithm can improve the channel estimation accuracy by exploiting the sparse characteristics of the channel.The high-dimensional sparse channel is estimated by calculating the channel expectation.In order to improve the accuracy of the support set search

the idea of Viterbi path decoding is introduced in the search process of dominating support set.All the paths with probability difference below the threshold are preserved.Then

the channel estimation on the chosen support set is carried out and the channel expectation is calculated as the final channel estimate.Theoretical analysis and simulation results show that the lower computational complexity is maintained and the channel estimation accuracy is kept at the same time.