Off-Grid OTFS Channel Estimation Based on Equally Distributed Information Quantity Grid Evolution

In high-mobility scenarios, orthogonal time frequency space (OTFS) exhibits a significant advantage over orthogonal frequency division multiplexing (OFDM) due to its sparsity in the delay-Doppler (DD) domain. However, challenges arise in the estimation of fractional paths due to the issue of low grid resolution, leading to notable modeling error in the sparse representation of the channel state information (CSI). Typically, the accuracy of CSI acquisition increases as grid points approach the actual paths, thus the distribution of grid points should be related to the distribution of actual paths in the DD domain. In this letter, we propose a 2D off-grid OTFS channel estimation method based on the equally distributed information quantity. The distribution of grid points is evolved according to the distribution of estimated paths in the delay and Doppler dimension in sequence. For each dimension, two processes, i.e., learning and fission, are carried to evolve the grid and estimate the channel parameters, alternatively. A coarse uniform grid will be evolved into a non-uniform denser grid which represents the CSI in DD domain more accurately. Simulation results indicate our proposed grid evolution method outperforms the off-grid method with uniform grid interval at the same level of complexity. Furthermore, it surpasses the 1D off-grid method and achieves estimation accuracy close to the 2D off-grid SBL combination method at a lower complexity.