Two-Stage Joint BEM-OTFS Channel Estimation Algorithm Based Sparse Bayesian Learning Algorithm

In this paper, a two-stage BEM-OTFS channel sparse Bayesian learning (SBL) estimation algorithm is proposed to address the problem of increasing error of the compressed sensing channel estimation algorithm at low resolution when OTFS is embedded in the pilot frequency mode and the pilot frequency overhead is too large. The algorithm uses a complex exponential basis expansion model (CE-BEM) at the transmitter side to design a design mode with lower pilot overhead, and a linear solution is used to obtain the roughly estimating channel matrix. After detecting the rough estimated channels, some of data symbols detected are used as pseudo-pilot for the second stage estimation. In the second stage, a sparse expression about the basis parameters is constructed using the discrete prolate spheroidal basis expansion model (DPS-BEM), which is solved by the SBL algorithm to calculate a more accurate channel after the sparse Bayesian learning (SBL) algorithm. The simulated numerical analysis shows that the present algorithm has ideal performance for OTFS symbols with lower Resolution, which is better than the SBL algorithm which is based on the DD domain pilot input-output relationship in general. Meanwhile, the present algorithm also has some advantages in spectral efficiency.