Investigation of Effectiveness of Deep Learning on OFDM and NOMA Systems

Abstract

Research on deep learning (DL) to do detection of non-orthogonal multiple access (NOMA) and OFDM is presented in this paper. The successive interference cancelation (SIC) is generally fulfilled at the receiver in NOMA systems that decode multiple users in a successively. The detection accuracy is mostly based on the true detection of previous users due to the effects of error propagation. The NOMA receiver based on DL is described with deep neural network (DNN), which implements an estimation of channel and detection of signal together. The receiver has robust characteristics on the power allocation of the user is explicit from the simulation results. DNN is suitable for both linear channels and nonlinear channels, also the receiver is getting well on detection while the number of users is increasing. DL approximation obtains better achievement than a ML detection that ignores interference effects when the interference of the inter-symbol is intense. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.