Deep Neural Networks Enhanced Phase Asynchronous Physical-Layer Network Coding

Physical-layer network coding (PNC) is a promising technique in 6th-generation (6G) that can enhance the throughput of wireless communication systems, but it suffers from the performance loss because of relative phase offset that is brought by the asynchrony between different uplinks. Traditional way such as belief propagation (BP) that is used to solve the asynchrony needs estimated phases in uplinks as prior knowledge, and the computation complexity is high. In this paper, we propose a deep neural network (DNN) based PNC model that can deal with the phase asynchrony automatically and effectively without prior knowledge, and the system has small architecture that is easy to implement. Simulation results verify that our system has advantage of dealing with relative phase offset in PNC system under various modulation types.