Detecting Fake Position in Smart Vehicle Networks Using Advanced Machine Learning

What is it about?

Vehicular Ad-hoc Networks (VANETs) allow cars to communicate with each other and with roadside systems to improve road safety and traffic flow. However, some vehicles can send fake location data, causing confusion or accidents. Our research developed a system to detect these fake messages using machine learning, a technology that helps computers learn from data. We tested four methods—Decision Trees, Random Forests, K-Nearest Neighbors (KNN), and Multilayer Perceptron (MLP)—combined with a technique called bagging, which improves accuracy by combining multiple predictions. Using a simulated dataset called VeReMi, we found that KNN with bagging was the best at spotting fake location data, achieving nearly perfect accuracy for most scenarios. This system runs on roadside units, making it practical for real-world use without overloading car systems. Our work helps ensure safer and more reliable smart vehicle networks.

Featured Image

Photo by Nicki Eliza Schinow on Unsplash

Photo by Nicki Eliza Schinow on Unsplash

Why is it important?

This research is timely because smart vehicle networks (VANETs) are becoming critical for modern transportation systems, enabling features like collision warnings and traffic management. However, fake location data from malicious vehicles can disrupt these systems, risking accidents or traffic chaos. Our study is unique because it uses advanced machine learning (KNN with bagging) to detect these threats with near-perfect accuracy, even in complex scenarios. By running detection on roadside units, our approach is practical and scalable, reducing the burden on vehicles. This work strengthens the security of smart transportation, paving the way for safer roads and more reliable traffic systems. As self-driving cars and connected vehicles grow, our findings are crucial for protecting these technologies from cyber threats.

Perspectives