Joint channel estimation and signal detection for AFDM transmission with time-domain pilot arrangement

Affine Frequency Division Multiplexing (AFDM) has shown great potential in high-mobility communication scenarios. However, efficient pilot design and low-complexity receiver algorithms remain challenging. In particular, conventional subcarrier-embedded and superimposed pilot schemes often suffer from excessive guard interval overhead and interference between pilots and data symbols. To overcome these limitations, this paper introduces a novel time-domain block-type pilot structure that enables effective pilot-data separation while reducing the overhead of guard intervals. Building on a parametric bilinear model derived from a time-domain basis expansion model (BEM), we propose a joint channel estimation and signal detection framework for AFDM systems. To facilitate the implementation of this framework with high computational efficiency, we develop a low-complexity sampling-based Expectation Propagation (EP) algorithm, which leverages Dirac mixture sampling approximations to simplify the computation of intractable integrals in bilinear message passing. Through benchmark comparisons with the Genie-aided bound and other established algorithms in both channel estimation and data detection, simulation results demonstrate the effectiveness of the proposed algorithm.