Detecting and Preventing Malicious Behavior in Connected Vehicle Networks

What is it about?

In this paper, we present a novel system for detecting malicious behavior, specifically distributed denial-of-service (DDoS) attacks, in Vehicular Ad-hoc Networks (VANETs). DDoS attacks aim to flood the network with enough traffic to make legitimate communication difficulty or impossible. Our paper makes contributions to the state-of-the-art in malicious behavior detection in VANETs by proposing a multi-layered real-time method for detecting both the onset of DDoS attacks and the vehicles responsible for the attack. The datasets used in our work are generated through realistic VANET simulations in multiple cities (with different traffic flow characteristics), while varying the number of attackers in each simulation to enhance the robustness of our model. Our novel data preprocessing method helps to mitigate some of the limitations of machine learning approaches such as data quality and model accuracy. Our research offers valuable insights into VANET security, with promising results in detecting malicious behavior in various real-world VANET scenarios.

Featured Image

Photo by Ishan @seefromthesky on Unsplash

Photo by Ishan @seefromthesky on Unsplash

Why is it important?

Unlike prior work, we present a framework for detecting malicious behaviour in connected vehicle networks that more suited for deployment in real-world networks. We do this by combining attack onset detection and the malicious node detection into a single framework. Our proposed framework is also designed to work in real-time and tested on data from actual vehicular network simulations. All of these make our framework more suitable for real-world deployment than prior work in this space.

Perspectives