Anomaly Detection in Vehicular Ad-Hoc Networks (Vanets) Using a Hybrid Deep Learning Approach

Nowadays there is a rapid growth in the number of vehicles emphasizing the necessity for better Intelligent Transportation System (ITS) for Vehicular Ad-hoc networks (VANETs). This interest towards VANETs increased the necessity to develop advanced machine learning and deep learning models to improve the capabilities of anomaly detection in connected vehicles. Existing models can only identify specific type of anomalies and they fail to detect anomaly in real-time data. So it is crucial for us to implement better anomaly detection framework that can work well in both the mentioned scenarios. In this paper, VANET anomaly detection framework is discussed that uses various Deep Learning (DL) approaches. Data is extracted from the Luxembourg SUMO Traffic (LuST) Scenario. This scenario simulates traffic of 24 hours in Luxembourg city and 27 different kinds of anomalies are introduced into the dataset. 4 different DL models are being implemented and their performances are evaluated using various evaluation metrics like accuracy, precision, recall and f1-score. The hybrid model with best results is CNN-GRU-Transformers with an accuracy of over <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 6 \%}$</tex>. It addresses the difficulties commonly found in real-world applications and exhibits good performance with real-time data