Bilinear Vector Approximate Message Passing-Based Off-Grid Channel Estimation for OTFS Systems

Accurately estimating channels with fractional Doppler and delay is a challenging task for orthogonal time-frequency space (OTFS) systems. This paper proposes using the bilinear vector approximate message passing (Bi-VAMP) algorithm to sparsely estimate integer delay and Doppler shifts and learn fractional delay and Doppler shifts separately. The scalar coefficients of uncertain matrices are extended to vector form, transforming the first-order approximation model of the effective delay-Doppler (DD) domain channel response into an uncertain matrix estimation model. Furthermore, considering the characteristics of OTFS channels, this paper provides some suggestions to enhance the convergence robustness of the Bi-VAMP algorithm. Simulation results validate that compared to other on-grid and off-grid methods, the proposed off-grid Bi-VAMP OTFS channel estimation scheme exhibits significant advantages in normalized mean square error (NMSE) performance metrics.