Neural Networks-Based Phase Estimation and Symbol Detection for RIS-Assisted Wireless Communications

Abstract

In this letter, two state-of-the-art Artificial Neural Networks (ANN)-based models are proposed for multiple Reconfigurable Intelligent Surfaces (RISs)-assisted wireless communications. The first model is the ANN-RIS model, where the potential of RIS phase prediction is investigated. Subsequently, for the receiver (RX), an ANN-based symbol detection model is proposed, where the ANN at the receiver is termed as ANN-RX. The ANN-RIS is trained using the channel coefficients between the transmitter TX-RIS and RIS-RX as input to generate the phase shift required at the RIS as output. Further, the ANN-RX is trained using the baseband received signal as the input and the corresponding transmitted symbol as the output. The training performance of both the models is examined in terms of Mean Square Error (MSE) and cross-entropy for ANN-RIS and ANN-RX, respectively. In addition, considering multiple RISs, the Bit Error Rate (BER) performance of the ANN-RX model is evaluated for different sizes of the RIS panel and different quantization levels of RIS element phases. Finally, the performance of the proposed methods is compared with existing methods in the literature. The obtained results demonstrate that the proposed methods perform better in terms of BER when compared to other existing methods. © 1997-2012 IEEE.