A Model-Driven Deep Learning-Based Receiver for OFDM System With Carrier Frequency Offset
Xincong Lin, Yushi Shen, Chunxiao Jiang · IEEE Communications Letters · 2024
Many state-of-the-art deep learning (DL)-based approaches have been applied to design orthogonal frequency-division multiplexing (OFDM) receivers. However, the issue of DL-based receiver design in the presence of carrier frequency offset (CFO) has often been disregarded and not adequately addressed. In this letter, we propose a model-driven DL-based receiver for an OFDM system with CFO. Drawing upon the state evolution theory, we establish a convex function that correlates the energy of the estimated channel with the residual CFO. Subsequently, we develop a CFO estimation technique by maximizing the estimated channel energy. To efficiently solve the optimization problem, we propose a joint CFO and channel estimation algorithm based on the ternary search method. Additionally, we propose a model-driven DL-based receiver based on the highly accurate CFO and channel estimation. Despite slightly higher complexity, our proposed receiver outperforms existing methods and approaches the performance limit attainable in the absence of CFO, thereby highlighting its superiority.