Iterative Channel Estimation for AFDM with High Spectral Efficiency

This paper addresses channel estimation for the affine frequency division multiplexing (AFDM) systems without guard symbols and with the existence of fractional Doppler shifts. As a promising candidate for the next-generation waveforms, AFDM is robust to Doppler shifts and can achieve the optimal diversity order. However, the existing channel estimation schemes for AFDM usually require guard symbols, resulting in low spectral efficiency, or only consider the integer Doppler, causing performance degradation for practical cases. In this work, a novel channel estimation scheme for AFDM is proposed, where short Zadoff-Chu (ZC) sequences are utilized as the pilot. The estimation of the channel gains, Doppler shifts, and delays is performed based on the correlation of the pilot sequence and iteratively with the data detection. Simulation results demonstrate the performance superiority of the proposed scheme.