Off-Grid Channel Estimation With 2D Pilots for MIMO-OTFS in Fractional Delay-Doppler Channels

Orthogonal time frequency space (OTFS) modulation is a promising technique for high-mobility communication, yet it faces challenges related to high pilot overhead. Two-dimensional (2D) pilots offer an effective solution to reduce this overhead, while inappropriate pilot sizes would lead to performance degradation in channel estimation, pilot overhead, and peak-to-average power (PAPR). Recognizing these challenges, this paper first examines the trade-offs among these factors with respect to the pilot size. Moreover, an off-grid channel estimation approach is proposed for 2D pilots in fractional delay-Doppler (DD) channels. The proposed algorithm includes two innovative processing methods for 2D pilots, which can convert the 2D-pilot response into an impulse-pilot response. One features a simple design with low complexity in processing 2D-pilot responses. The other enhances performance by exploiting sequence characteristics and a cyclic filling structure, maintaining lower complexity compared to existing methods. Combined with off-grid, the proposed algorithm can perform channel estimation. The simulations demonstrate that the proposed channel estimation algorithm achieves superior performance compared to baseline methods, in terms of normalized mean square error (NMSE) and bit error rate (BER) performance.