Joint Channel and Data Estimation via Bayesian Parametric Bilinear Inference for OTFS Transmission

In high-speed mobile communication environments, an orthogonal time frequency space (OTFS) scheme with robustness to doubly-selective fading channels by spreading symbols in the frequency-time (FT) domain has attracted much attention. However, typical pilot-based channel estimation schemes cause system performance degradation due to the increased overhead of channel state information (CSI) acquisition, and large-scale matrix operations based on the size of OTFS equivalent channels are also problematic in terms of the computational cost. To address this issue, in this paper, we focus on the fact that joint channel and data estimation (JCDE) in the delay-Doppler (DD) domain OTFS systems can be formulated as a large-scale parametric bilinear inference problem, and solve it via Gaussian belief propagation (GaBP) to design a novel low-complexity and high-accuracy JCDE algorithm with the use of relatively short pilot sequences. From computer simulations, we confirm that the proposed method significantly outperforms the conventional two-stage channel and data estimation, and asymptotically approaches the idealized scheme given perfect CSI knowledge.