Joint Channel Estimation and Equalization for OTFS Based on EP

Orthogonal time-frequency space signaling (OTFS) is a novel modulation scheme that can exploit full time-frequency (TF) diversity when coupled with a proper receiver. Toward this end, this paper presents a joint channel estimation and equalization approach for OTFS based on the expectation propagation (JCEE-EP). First, We apply discrete prolate spheroidal basis expansion model (DPS-BEM) to describe the time-varying (TV) channel and obtain a Gaussian prior of BEM coefficients via LS estimator. Based on the delay-Doppler (DD) domain to time domain transformation, we then decompose the joint posterior probability into a couple of factors and approximate each one by Gaussian distributions according to the EP principle. As the moment-matching (MM) step in channel factor update is computationally intractable, we resort to the first-order conditional EP (CEP) method for an approximate solution. Finally, the bit error rate (BER) simulation results demonstrate the superiority of our proposed JCEE-EP over the existing approaches.