ERI: Physical Layer-Enabled Cooperative Learning for Secure and Anonymity-Preserving Intelligent Transportation Systems

The advent of cooperative driving within intelligent transportation systems (ITS) presents a promising avenue to enhance safety and optimize traffic flow through seamless vehicle-to-everything (V2X) communication. This wireless interconnection enables vehicles to interact dynamically with each other and with the infrastructure, allowing for real-time coordination in collision avoidance, platooning, and adaptive traffic management. However, this connectivity brings forth significant cybersecurity challenges that fundamentally reshape the safety assumptions of modern transportation systems. Unlike traditional vehicular systems where control was largely localized, V2X-enabled environments create a highly interdependent ecosystem in which the behavior of one compromised or malfunctioning node can propagate rapidly across the network. Traditional security mechanisms in Intelligent Transportation Systems (ITS) primarily focus on individual vehicles or isolated onboard systems, often neglecting the complex interactions among cross-vehicle dependencies and the cascading effects introduced by misbehaving neighboring vehicles. As a result, attacks or faults that exploit inter-vehicle cooperation can remain undetected until they manifest as system-wide disruptions. Moreover, the evolving landscape of threats extends beyond traditional cyberattacks. Artificial intelligence (AI)-related threats pose an increasingly intricate challenge, including adversarial manipulated sensor inputs, learning-based deception strategies, and AI-generated behavioral anomalies that mimic legitimate vehicle actions. These attacks are particularly difficult to detect because they may not violate explicit protocol rules but instead exploit learned or statistical patterns in vehicle behavior. Consequently, distinguishing between misbehaving vehicles, faulty sensor behavior, and AI-driven anomalies becomes a critical and unresolved challenge. This creates a pressing need for intelligent, context-aware, and system-level security mechanisms that can reason over both cyber and physical interactions. In response to these multifaceted challenges, the proposed project seeks to develop a Physical Layer–Enabled Cooperative Learning framework for secure and anonymity-preserving Intelligent Transportation Systems. The proposed research leverages advanced data-driven models integrated with physical-layer security principles to provide a deeper understanding of the dynamic and evolving landscape of smart transportation systems. To this end, the project will develop a hierarchical security framework that strengthens resilience across intra-vehicular, inter-vehicular, and system-level operations. At the intra-vehicular level, the framework will secure communications among in-vehicle components such as sensors, controllers, and embedded modules, while addressing threats arising from misbehaving nodes, and adversarial AI behaviors that may closely resemble normal system operation. At the inter-vehicular level, the project will extend security analysis to cooperative interactions among multiple vehicles and roadside infrastructure, focusing on detecting compromised communication links and coordinated attacks across connected systems. Finally, at the system level, the project will provide a comprehensive assessment of the reliability of vehicles. This multi-layer approach will integrate physical-layer intelligence with cooperative learning to identify subtle anomalies that are not observable at higher protocol layers. Finally, to maximize impact and accessibility, the project will deliver an open-source, publicly available platform to enable researchers, practitioners, and industry stakeholders to evaluate, and extend the proposed methods. In addition, educational modules and training resources will be disseminated to support workforce development and hands-on learning in secure autonomous systems, AI-driven cybersecurity, and intelligent transportation technologies. This aw NSF Award ID: 2553310 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Wassila Lalouani | Institution: Towson University, TOWSON, MD | Award Amount: $149,457 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2553310 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2553310.html