Correlated Block-Sparse Channel Estimation and Phase Optimization in RIS-OTFS Systems

For a reflecting intelligent surface (RIS)-aided orthogonal time-frequency space (OTFS) system, we propose a two-dimensional off-grid sparse channel estimation model that exploits the fact that delay-Doppler (DD) locations of non-zero channel gains in a cascaded RIS-OTFS channel are sum of non-zero DD locations in the base station-RIS and RIS-user channels. Using this model, we estimate channel by considering multiple OTFS frames, which results in a block sparse DD domain channel. The multiple frames allow us to model the temporal correlation in a multi-frame OTFS system. We propose a novel correlated coupled prior to capture the DD sparsity of the channel, and estimate it by developing a sparse Bayesian framework. We maximize the achievable rate for our RIS-OTFS system by developing a gradient descent method, which has a high complexity. We also propose a low-complexity dimension-wise sinusoidal maximization (DSM)-based DD domain channel power maximization to obtain sub-optimal phase shifts. We show that the proposed channel estimation and phase optimization algorithms provide substantial gain over their existing counter-parts. We also show the robustness of an RIS-OTFS system over its non-RIS counterpart in a high-mobility scenario.