A CNN-based End-to-End Learning for RIS-assisted Communication System

TL;DR

The paper tackles end-to-end optimization of RIS-assisted communications by introducing a CNN-based autoencoder (CNN-AE) that jointly configures the transmitter, RIS, and receiver. It presents architectures for perfect CSI and channel-estimation scenarios, including a RIS block that learns optimal phase shifts and a cross-entropy loss for symbol decoding. Numerical results show BER improvements over theoretical RIS-assisted benchmarks and demonstrate fast convergence, with performance depending on RIS size and modulation. This work offers a practical, low-complexity DL-based framework for optimizing RIS-enabled links in beyond-5G networks, adaptable to CSI availability.

Abstract

Reconfigurable intelligent surface (RIS) is an emerging technology that is used to improve the system performance in beyond 5G systems. In this letter, we propose a novel convolutional neural network (CNN)-based autoencoder to jointly optimize the transmitter, the receiver, and the RIS of a RIS-assisted communication system. The proposed system jointly optimizes the sub-tasks of the transmitter, the receiver, and the RIS such as encoding/decoding, channel estimation, phase optimization, and modulation/demodulation. Numerically we have shown that the bit error rate (BER) performance of the CNN-based autoencoder system is better than the theoretical BER performance of the RIS-assisted communication systems.

Figures (10)

• Figure 1: Illustration of the system model.
• Figure 2: Illustration of the system architecture.
• Figure 3: CNN - based end-to-end learning system.
• Figure 4: RIS block.
• Figure 5: Illustration of the system architecture with channel estimation.