Reconfigurable Intelligent Surface-assisted Classification of Modulations using Deep Learning

TL;DR

This paper introduces a RIS-assisted deep learning approach for classifying digital modulations in 5G networks, demonstrating high accuracy especially at low SNR levels.

Contribution

It presents a novel RIS-assisted AI method that classifies modulations directly from received signals without feature extraction, enhancing robustness in dynamic 5G environments.

Findings

High classification accuracy at low SNR levels

Features learned directly from signals improve robustness

RIS assistance enhances modulation identification

Abstract

The fifth generating (5G) of wireless networks will be more adaptive and heterogeneous. Reconfigurable intelligent surface technology enables the 5G to work on multistrand waveforms. However, in such a dynamic network, the identification of specific modulation types is of paramount importance. We present a RIS-assisted digital classification method based on artificial intelligence. We train a convolutional neural network to classify digital modulations. The proposed method operates and learns features directly on the received signal without feature extraction. The features learned by the convolutional neural network are presented and analyzed. Furthermore, the robust features of the received signals at a specific SNR range are studied. The accuracy of the proposed classification method is found to be remarkable, particularly for low levels of SNR.