Low-Overhead OTFS Transmission With Frequency or Time Domain Channel Estimation

Existing channel estimation techniques for orthogonal time frequency space (OTFS) transmission are mostly based on the pilots embedded in the two-dimensional delay-Doppler domain, which significantly increases the signaling overhead. In this paper, three variants of OTFS transmission, i.e., frequency-domain pilot-aided (FD-PA) OTFS, time-domain pilot-aided (TD-PA) OTFS, and time-domain training sequence (TD-TS) OTFS, are proposed. They enable one-dimensional channel estimation and corresponding equalization to be applied in either frequency or time domain. The relationships among different channel representations for fast fading channels are first formulated with discrete Fourier transform (DFT) and inverse DFT matrices. The transmission frame construction and associated channel estimation methods are then presented in detail for FD-PA-OTFS, TD-PA-OTFS, and TD-TS-OTFS respectively. The theoretical mean-square-error (MSE) for each of the proposed channel estimation methods is also derived. Finally, simulation results are provided to verify the MSE analyses and demonstrate that the proposed FD-PA-OTFS is the most effective transmission technique for high-mobility wireless communications in terms of diversity performance, signaling overhead, and power efficiency.