Reinforcement Learning-Assisted Virtualized Security Framework for CAVs
Connected and autonomous vehicle (CAV) technology has brought a major transformation in the transportation sector by significantly improving the mobility of people and goods through advanced communication, sensing, and computing capabilities. However, CAVs can be hacked due to vulnerabilities in the in-vehicle software, resulting in physical damage and jeopardizing the safety of drivers and passengers. By exploiting the vulnerabilities, hackers can perform malicious actions ranging from draining batteries and taking control of the steering wheel to disabling the alarm system. The existing security solutions implemented in CAVs are static and cannot withstand evolving security threats such as Advanced persistent threats (APT) and ransomware attacks. Moreover, costly update procedures leave the CAV software unpatched for a long time, making the CAVs vulnerable to new exploits. This project aims to develop a virtualized security framework to improve the resiliency of CAV software. The framework will allow the execution of different code variants of CAV software to introduce uncertainty in the attack surface. The proposed framework will integrate the Network Functions Virtualization paradigm to implement the code variants of CAV software as virtual network functions. The proposed framework will offer the ability to optimally deploy the appropriate virtual network functions using a reinforcement learning agent. The reinforcement learning agent perceives the threat environment of CAVs and provides the optimal code variant that maximizes the resiliency of CAV software while ensuring their Quality of Service (QoS) requirements. This project aims to accomplish the following goals: (1) develop a virtualized security framework that allows fast and dynamic provisioning of different code variants of CAV software, (2) design novel and efficient algorithms designed based on game theory and Artificial Intelligence (AI) techniques including Deep Learning and Generative Adversarial Networks (GANs) to determine the optimal code variant, (3) evaluate the performance of reinforcement learning algorithm using simulations, and (4) build a proof-of-concept of the proposed security framework and evaluate its performance using real-world experiments.
Language
- English
Project
- Status: Active
- Funding: $174658
-
Contract Numbers:
69A3552344812
69A3552348317
-
Sponsor Organizations:
Office of the Assistant Secretary for Research and Technology
University Transportation Centers Program
Department of Transportation
Washington, DC United States 20590South Carolina State University
300 College Street NE
Orangeburg, South Carolina United States 29117 1600 Harden Street
Columbia, South Carolina United States 29204 -
Managing Organizations:
National Center for Transportation Cybersecurity and Resiliency
1 Research Dr
Greenville, South Carolina United States 29607South Carolina State University
300 College Street NE
Orangeburg, South Carolina United States 29117 -
Project Managers:
Chowdhury, Mashrur
-
Performing Organizations:
South Carolina State University
300 College Street NE
Orangeburg, South Carolina United States 29117 1600 Harden Street
Columbia, South Carolina United States 29204 -
Principal Investigators:
Sahoo, Jagruti
Mwakalonge, Judith
Swain, Nikunja
Biswal, Biswajit
Comert, Gurcan
- Start Date: 20240101
- Expected Completion Date: 20241231
- Actual Completion Date: 0
- USDOT Program: University Transportation Centers
Subject/Index Terms
- TRT Terms: Artificial intelligence; Autonomous vehicles; Computer security; Connected vehicles; Game theory; Machine learning; Software; Visualization
- Subject Areas: Data and Information Technology; Security and Emergencies; Transportation (General); Vehicles and Equipment;
Filing Info
- Accession Number: 01907736
- Record Type: Research project
- Source Agency: National Center for Transportation Cybersecurity and Resiliency (TraCR)
- Contract Numbers: 69A3552344812, 69A3552348317
- Files: UTC, RIP
- Created Date: Feb 9 2024 7:36PM