An AFDM off-grid Channel Estimation Based on Sparse Bayesian Learning

Affine frequency division multiplexing (AFDM) modulation is a linear frequency modulation multi-carrier communication system that offers performance comparable to Orthogonal time frequency space (OTFS) modulation in time-varying channels. In this paper, we apply the off-grid model and sparse Bayesian learning (SBL) framework to address channel estimation challenges fractional Doppler. We exploit the joint sparsity of channel parameters to estimate the on-grid components of Doppler shift by indexing, with off-grid components estimated as hyperparameters via the EM algorithm. Furthermore, our proposed method is compatible with both embedded single pilot and sequential pilots.