An Integrated Unsupervised and Supervised Learning Technique for Interference Analysis in Vehicular Communications

In vehicular communications, various sensors and technologies such as radars, cameras, and LiDARs coexist and often share the same radio frequency spectrum. This results in a complex and dynamic radio environment with significant mutual interference, which poses an immense challenge to achieve reliable communication and performance. Intelligent techniques based on machine learning and artificial intelligence have been found as promising solutions to tackle this challenge. In this paper, we propose a hybrid technique that integrates an unsupervised clustering method with a Convolutional Neural Network (CNN) based system that detects the interfering packets by processing the received signal spectrogram. We then compare the prediction accuracy performance of the proposed system under different Signal-to-Noise-Ratio conditions and with different training mechanisms. Our results suggest that the proposed system can perform reliably under a wide range of SNR conditions.

Automated summary

This paper proposes a hybrid technique that combines unsupervised clustering and a CNN-based system to address the interference problem in vehicular communications. Experimental results show that the proposed system performs reliably under various Signal-to-Noise-Ratio conditions.