Efficient Channel Estimation Utilizing Channel Sparsity in the Delay Domain for OTFS Systems

In this paper, we consider orthogonal time frequency space (OTFS) systems in the presence of fractional Doppler shifts. For OTFS systems, an embedded pilot-aided channel estimation (CE) method has been proposed. In this method, pilot and guard symbols are appropriately placed in the delay-Doppler (DD) domain, and thresholding is then applied to the received signal for CE. However, this method suffers from a significant degradation in CE accuracy due to inter-Doppler interference, which is caused by fractional Doppler shifts. To this end, we propose CE methods using the fast iterative shrinkage-thresholding algorithm (FISTA) and coordinate descent (CD) method, exploiting channel sparsity in the DD domain. Furthermore, these proposed methods reduce the size of the received signal and the measurement matrix by thresholding that takes advantage of channel sparsity in the delay domain. Computer simulations demonstrate that the proposed methods outperform the conventional threshold-based and sparse Bayesian learning-based CE methods.