Secure Connected Vehicle-based Traffic Signal Systems Against Data Spoofing Attacks

Abstract

The emerging Connected Vehicle (CV) technology is widely expected to greatly enhance traffic safety and efficiency by enabling vehicles, pedestrians, and infrastructures to communicate with one another. As a promising CV application, CV-based traffic signal control aims to improve the traffic efficiency at intersections by dynamically optimizing traffic signal control plans based on the mobility information submitted by surrounding CVs. Effective CV-based traffic control relies on accurate estimation of the queue length i.e., the number of vehicles waiting at intersections, to determine the optimal traffic signal control plans. Despite significant efforts on accurate queue length estimation, the robustness of queue length estimation has so far received very limited attention. A recent study has demonstrated that it is possible for malicious CVs to significantly manipulate the queue length estimation by reporting false mobility data, which can cause severe traffic congestion. To tackle this challenge, we introduce a robust queue length estimation mechanism that first utilizes the mobility data reported by all the CVs waiting in the queue to calculate multiple preliminary queue length estimates. Then, the robust statistical methods are adopted to derive a resulting estimated queue length whose accuracy is kept at an acceptable level even though there exist multiple malicious CVs in the queue. The simulation results confirm the effectiveness of the proposed mechanism.