Data-Driven Phase Noise Mitigation in OFDM Systems

The performance of an orthogonal frequency division multiplexing (OFDM) system can be largely restricted without taking precautions against phase noise (PN) in receivers. In this paper, we focus on a time-domain PN estimation and compensation problem in OFDM systems. Despite the existing works devoted to the estimation of PN with the aid of pilot signals, they mostly select a set of orthogonal basis to assemble PN, which may lead to inaccurate estimation and compensation of PN. To alleviate the adverse impacts of PN in OFDM systems, we propose a data-driven approach that leverages historical PN samples to extract a set of non-orthogonal frequency-oriented (NOFO) basis for PN mitigation. To show the practicability of the proposed data-driven approach, we employ a 5G new radio (NR) like simulation setting and our numerical results show that the proposed approach can mitigate PN effectively, thereby attaining improved bit error rate (BER) performance.