V2X communication Technology Identification Using Residual Neural Network

Abstract

Signal identification is a critical topic in new communication systems, especially for cognitive radio to have an optimal sharing of radio resources. Some existing techniques, used to identify and classify wireless communication signals, show strong performance, but their sensitivity to noise levels or high computational complexity pose challenges. In this paper, we propose to use a Deep Learning technique, based on Residual Neural Networks (ResNet) to detect and classify V2X (vehicle-to-everything) signals. Three V2X communication technologies are studied and evaluated: ITS-G5, 4G, and 5G. The proposed model offers robust performance compared to a classical CNN-1D model even for the low SNR value.