Research Projects

Principal Investigator(s): Yongkai Wu (Clemson University)

Project Partners: Feng Luo (Clemson University), Latifur Khan (The University of Texas at Dallas), M. Sabbir Salek (Clemson University), Bhavani Thuraisingham (The University of Texas at Dallas).

Research Project Funding: Federal $113,948; Cost-share $129,722

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This research develops robust and secure artificial intelligence (AI) systems for smart transportation that could defend against novel adversarial attacks with high performance. The project addresses critical vulnerabilities in Test-Time Adaptation (TTA) mechanisms and Multimodal Large Language Models deployed in transportation systems. Through new attack discovery, effective defense framework development, and deployable prototype systems, this work ensures that AI technologies can be safely deployed in safety-critical transportation applications. The research delivers practical solutions including natural scene adversarial attack frameworks, sharpness-aware minimization based TTA defenses, event-conditioned representation compression for efficient multimodal AI, and adversarially robust multimodal fusion architectures.

US DOT Priorities: This project directly advances USDOT's Safety Priority by developing AI systems that maintain performance while resisting adversarial manipulation targeting the compromise of safety-critical transportation decisions. It addresses the Cybersecurity Priority by discovering and defending against novel run-time vulnerabilities in AI systems that are increasingly deployed in connected and autonomous vehicles. The research engages in transformative research by being the first comprehensive study of adversarial attacks specifically targeting TTA in transportation contexts. The work advances RD&T strategic goals through development of deployable prototype systems with open-source implementations.

Outputs: This research will generate several significant outputs that advance both theoretical knowledge and practical implementation in transportation AI security. First, the project will deliver software prototypes as an open-source Python package. Second, we will develop benchmark datasets for security evaluation across diverse transportation scenarios. Third, the project will release open-source tools providing complete implementations of attack and defense frameworks. Fourth, the research will produce peer-reviewed publications in top-tier conferences and journals to advance AI security research in transportation systems. Finally, we will develop educational materials, including course modules suitable for undergraduate and graduate programs in computer science, civil engineering, and automotive engineering disciplines.

Outcomes/Impacts: This research will significantly transform transportation AI security by establishing new resilient AI systems that can be safely deployed in safety-critical applications. The immediate impact includes enabling transportation agencies and autonomous vehicle developers to deploy AI systems while resisting novel adversarial attacks. The proposed framework will reduce vulnerability to run-time attacks by promoting robust optimization during adaptation, directly improving safety of the perception systems of connected and autonomous vehicles. Our resilient multimodal AI system will enable real-time processing on resource-constrained edge devices in vehicles and infrastructure while maintaining security guarantees. Long-term impacts include informing regulatory frameworks for AI deployment in connected transportation system in the real-world context. The open-source deliverables will accelerate adoption across the transportation research community and industry. Overall, the proposed work will help reduce transportation cybersecurity risks against adversarial threats and improve public safety.

Principal Investigator(s): Satish Ukkusuri (Purdue University)

Project Partners: Alvaro Cardenas (The University of California, Santa Cruz), Daniel Fremont (The University of California, Santa Cruz), Z. Berkay Celik (Purdue University), Amjad Ali (Morgan State University).

Research Project Funding: Federal $129,911; Cost-share $148,567

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This project (ProFAD) will develop a simulation-based framework that moves beyond collecting isolated failure cases, toward systematically mapping the boundaries of unsafe operation with probabilistic guarantees. Current stress testing and adversarial methods can surface rare crashes, but they typically return only a handful of failing scenarios and do not characterize how large the unsafe region is, how it evolves under noise and domain shifts, or whether nearby conditions share the same vulnerability. ProFAD addresses this gap by combining formal, structured scenario specifications (e.g., Scenic) with an adaptive, partition-based falsification engine that concentrates sampling near suspected safe/unsafe boundaries and uses surrogate models to improve sample efficiency while maintaining calibrated confidence statements. A key component is the use of Generative Flow Networks (GFlowNets) to guide diverse, reward-driven sampling of rare but safety-critical scenarios: rather than optimizing for a single “worst case”, GFlowNets enable ProFAD to discover and maintain multiple distinct failure modes and to cover the unsafe boundary more comprehensively. The end product is a set of certified unsafe regions (not just traces), integrated with high-fidelity simulators such as CARLA/AWSIM, producing interpretable “vulnerability maps” that support reproducible, scalable safety assessment of autonomous driving systems.

US DOT Priorities: The project supports US DOT priorities and the RD&T strategic goals by:

• Safety: ProFAD advances autonomous vehicle (AV) safety assurance by producing region-level evidence (“unsafe boundaries”) rather than anecdotal failures, and by using GFlowNet-driven exploration to uncover multiple rare hazard families that would be missed by single-mode adversarial search.
• Economic Strength and Global Competitiveness: By reducing dependence on costly real-world road testing and improving the efficiency and rigor of simulation-based validation, especially through sample-efficient generative sampling and surrogate-guided simulation, the project helps lower assurance costs while supporting the trustworthy deployment of AV technologies.
• Transformation: ProFAD provides a new paradigm for safety evaluation, probabilistic, boundary-focused falsification with statistical guarantees, augmented by generative models for scalable multi-modal discovery, supporting more transparent certification workflows and accelerating adoption of next-generation validation tools.

The project engages in breakthrough, advanced, and transformative research by:

• Systematic boundary discovery (not just point failures): Adaptive partitioning produces maps of unsafe operational regions rather than individual counterexamples.
• Probabilistic guarantees for safety claims: The framework attaches calibrated confidence statements to unsafe regions, enabling auditable safety evidence.
• Rare-event exploration with GFlowNets: GFlowNets enable diverse sampling across many distinct unsafe modes, improving mode coverage and avoiding overfitting to a single adversarial pattern.
• Sample-efficient falsification in high dimensions: Surrogate-guided exploration maximizes efficiency by focusing computational effort on critical boundaries, improving scalability without sacrificing rigor.

Outputs: The project is expected to produce:

• ProFAD software prototype (ready-to-use toolkit): An integrated falsification engine connecting formal scenario generation (e.g., Scenic) with high-fidelity simulators (CARLA/AWSIM), including simulation orchestration, adaptive partitioning, confidence estimation, and a GFlowNet module for diverse rare-event sampling.
• GFlowNet-based scenario generators: Learned generative policies that sample scenario parameters (e.g., traffic composition, behaviors, sensor/noise conditions, and environment factors) to efficiently discover multiple high-risk modes, with replay buffers and logging to support reproducibility.
• Scenario specification library: Reusable, structured scenario definitions capturing road geometry, traffic participants, environment, and uncertainty sources to support reproducible evaluation.
• Benchmark datasets and evaluation report: Publicly shareable benchmark datasets and standardized protocols measuring sampling efficiency, boundary coverage, and guarantee reliability against baselines (e.g., random sampling, grid search, Bayesian optimization, and adversarial RL), including comparisons to GFlowNet-free variants.
• Documentation: Open-source documentation (e.g., tutorials, examples) and research publications to enable adoption by researchers, developers, and agencies.
• New partnerships beyond the UTC consortium: The project is positioned to engage public agencies and industry stakeholders involved in AV safety validation, simulation tooling, and assurance workflows.

Outcomes/Impacts: The project is expected to produce the following outcomes/impacts:

• Improved AV safety assurance and certification readiness: Certified unsafe-region maps provide actionable evidence for developers and a reproducible basis for regulators to assess safety, supporting clearer certification and audit pathways than isolated failure reports.
• Reduced cost and time of validation: By concentrating simulation effort near safety-critical boundaries and leveraging surrogates, the framework reduces the number of expensive simulator runs and can lessen dependence on large-scale real-world mileage accumulation.
• More robust fixes and fewer blind spots: Boundary-focused outputs help teams avoid patching a single crash while leaving nearby vulnerabilities untested; instead, they can target mitigations that shrink unsafe regions across variations in traffic, environment, and noise.
• Greater transparency and public trust: Probabilistic guarantees and reproducible scenario specifications support clearer communication of safety evidence, strengthening confidence in AV deployment decisions.
• Long-term standardization potential: The project establishes a foundation for standardized probabilistic testing workflows that can be reused across systems, simulators, and stakeholders, improving consistency in how safety evidence is generated and interpreted.

Principal Investigator(s): Z. Berkay Celik (Purdue University)

Project Partners: Alvaro Cardenas (The University of California, Santa Cruz), Cihang Xie (The University of California, Santa Cruz), Satish Ukkusuri (Purdue University), Leilani Gilpin (The University of California, Santa Cruz).

Research Project Funding: Federal $107,900; Cost-share 110,223

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: The architecture of autonomous driving systems is shifting from modular pipelines to end-to-end systems powered by advanced Artificial Intelligence (AI) and Vision-Language Models (VLMs). This project develops a comprehensive framework for the compositional modeling, security analysis, and resilience testing of these next-generation systems. The research team will create formal models to abstract the behavior of AI components and build an AI-powered vulnerability analysis engine to identify semantic attacks. The project culminates in a high-fidelity, open-source software testbed that integrates these models to simulate attacks and evaluate the resilience of autonomous vehicles and drones.

US DOT Priorities: This project directly supports the US DOT priority of reducing transportation cybersecurity risks by addressing the critical security gaps in AI-driven autonomous systems. It aligns with TraCR's focus on Security and Resiliency. The project engages in breakthrough research by developing the first security modeling framework specifically designed for end-to-end AI architectures, moving beyond traditional component-level analyses. It addresses the novel attack surface introduced by Large Language Models (LLMs) and VLMs, focusing on semantic reasoning vulnerabilities that existing validation techniques fail to detect.

Outputs: The primary output is an AI-driven autonomous system security testbed, which will be developed as an open-source software tool for automated resilience analysis and attack simulation. Additional outputs include formal mathematical models for transformer and VLM components and algorithms for generating semantic attacks. The project will produce at least two publications in top-tier security or AI conferences and patent applications for the new attack generation techniques. The project leverages partnerships with industry leaders Qualcomm and Denso to validate the tools and facilitate technology transfer.

Outcomes/Impacts: This research will improve the safety and security of next-generation autonomous transportation systems by providing the industry with a standardized methodology for testing AI-driven architectures against sophisticated threats. The collaboration with industry partners will enable the direct transfer of these research findings into practice, allowing developers to secure products before they reach the market. By identifying and mitigating new classes of semantic vulnerabilities, the project will lead to more resilient autonomous agents and inform future security standards for automated vehicles.

Principal Investigator(s): Zulqarnain Khattak (Morgan State University)

Project Partners: Alvaro Cardenas (The University of California, Santa Cruz), Amjad Ali (Morgan State University).

Research Project Funding: Federal $81,152; Cost-share 85,637

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: Cooperative driving automation or Connected and Automated Vehicles (CAVs) are rapidly taking over modern intelligent transportation systems. The proliferation of CAVs has also intensified concerns around cybersecurity and data privacy. The added communication involved in these driving maneuvers serves as a vulnerable attack surface. The data communicated in Basic Safety Messages (BSMs) of CAVs is highly safety-critical, thus requires secure processing and sharing. Traditional security strategies are mostly machine learning-based that rely on centralized data processing and storage. The centralized servers act as a single point of trust, which is vulnerable to failure, compromising data privacy, and adding overhead to communication. To address these challenges, Federated Learning (FL) has emerged as a distributed learning paradigm that enables CAV agents to locally train models and only share model parameters with a global server for updates. This eliminates the need for raw data sharing, which preserves the privacy of sensitive data transfer during CAV communication and reduces the risk of single-point failure.

Despite the benefits of FL, it is still susceptible to threats like poisoning attacks, inference-based adversaries and model manipulation. The model parameters are not secured while shared iteratively between local and global agents. It is possible for adversaries to deliberately inject anomalies into the local model updates, thereby degrading the accuracy of the global model or compromising the individual local agents. To mitigate this inherent problem of FL, Blockchain serves as the apt solution. Blockchain technology is a lightweight, fully decentralized data storage framework that replaces conventional centralized databases by providing immutability and tamper-proofing to the stored data. The Secure Hashing Algorithm (SHA) and smart contracts employed by Blockchains facilitate trust and accountability in this storage solution. This research will integrate blockchain with FL to secure the training data shared between FL’s distributed agents. Due to the distributed nature of both frameworks, they complement each other well and are completely compatible for integration.

US DOT Priorities: The project fits under the USDOT theme of cybersecurity to promote resilient transportation networks and also aligns with the TraCR vision of promoting secure systems and falls under the Thrust 1 Security and Resilience of TraCR since the project will analyze the cybersecurity risks and their impact on cooperative driving using real-world experimental data and develop an anomaly detection model for secure operation of these systems.

Outputs: The major output of this project includes:

• Creation of data library for cyberattacks using real-world experiments of cooperative driving.
• Assessment of the impacts of cyberattacks and sensor anomalies on safety, stability and efficiency.
• Development of decentralized anomalous behavior detection algorithms to detect cyberattacks in lead and following vehicles for the resilient operation of cooperative driving.
• Fostering of a relationship between the Virginia Department of Transportation (VDOT), who serves as a project partner and major stakeholder, and TraCR.

Outcomes/Impacts: The VDOT has shown keen interest in this project due to real-world implications for transportation infrastructure and securing Intelligent Transportation System applications. VDOT provided a cash match of $63,558 for the first phase of the project and is continuing to be one of the major partners in this second phase with an in-kind contribution. It is essential for state agencies to find innovative approaches for the detection of anomalous behavior so that transportation systems can perform resiliently in the face of cyberattacks. The research team will work closely with the project partner, VDOT, to assist with the implementation of developed models from this research in their real-world intelligent transportation systems for traffic operations and management. The study findings will help guide understanding of the safety impacts of cyberattacks and protect critical infrastructure. The PI also plans to work on technology transfer by disseminating the findings of this research through journal publications and presentations in multiple forums, including the Transportation Research Board Annual Meeting, Automated Road Transportation Symposium and Intelligent Vehicles Symposium.

Principal Investigator(s): Amjad Ali (Morgan State University)

Project Partners: Satish Ukkusuri (Purdue University), Zulqarnain Khattak (Morgan State University).

Research Project Funding: Federal $135,154; Cost-share $135,552

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: The current state of practice in security research on connected automated vehicles (CAVs) does not consider how adversaries may evolve over time and adapt to defense strategies. This proposed research takes the state of practice well beyond the current focus on anomaly detection towards strategic response in defense strategies by developing attack defender models with game-theoretical models. Game theory provides a framework to study the strategic interactions between defenders and adversaries with conflicting objectives. Given the above background, this study will design strategic games to study attacker and defender strategies for cyber deception, as well as algorithms to compute equilibrium or optimal defense strategies in a CAV environment. Real-world data from CAV experiments conducted by PIs will be used to design game theory models in a CAV environment. The study will design two strategic games, namely a zero-sum game and a Stackelberg security game, to formalize the interactions between attackers and defenders by devising a strategic comparison between a zero-sum game and a Stackelberg security game. The proposed game models define payoff functions that capture the trade-offs between model accuracy and the success rates of attacker and defender. The dynamic attacker-defender strategies mimic real-world applications and provide the ability to provide alerts to traffic management center operators for performing cyber incident response in a timely manner, which has attracted the Virginia Department of Transportation’s (VDOT’s) interest. VDOT will serve as a partner to help with real-world implementation.

US DOT Priorities: The project aligns with the USDOT theme of cybersecurity to promote resilient transportation networks and supports the TraCR’s vision of promoting secure systems, which falls under Thrust 1 Security and Resilience of TraCR, since the project will analyze cybersecurity risks using a game theoretical approach and assess their impact on cooperative driving using real-world experimental data. The game-theoretic model will enable attacker and defender models to mimic real-world settings, with a payoff for both attackers and defenders to attack or defend the CAV environment.

Outputs: The major output of this project includes:

• Enhancement and expansion of the library of data for cyberattacks through real cooperative driving.
• A novel zero-sum game model for cyberattacks versus defense mechanisms to rigorously assess CAV resilience in various attack and defender strategies.
• A novel Stackelberg security game model to perform sequential play of the leader detection strategy, followed by attack strategies to assess CAV resilience against cyberattacks.
• Utilization several AI algorithms within defender models to test the resilience of CAV systems.
• A thorough assessment to evaluate CAV resilience against strategically optimized cyberattacks. Multiple attackers and defender strategies facilitate insightful findings.
• A partnership with the Virginia Department of Transportation (VDOT) for implementing the algorithms in their traffic management systems.

Outcomes/Impacts: VDOT has shown keen interest in this project due to its real-world implications for transportation infrastructure and for securing intelligent transportation system applications. VDOT is a major partner in this project, providing in-kind contribution support. It is essential for state transportation agencies to develop innovative approaches for detecting anomalous behavior, enabling transportation systems to remain resilient in the face of evolving cyberattacks. The research team will work closely with the project partner, VDOT, to assist with implementing game-theoretical and attacker-defender models derived from this research in their real-world intelligent transportation systems for traffic operations and management, thereby improving their resilience. Furthermore, the study findings will help guide understanding of the safety impacts of cyberattacks and protect critical infrastructure against evolving cyber threats. The proposed attacker-defender models have implications for the security and reliability of critical infrastructure.

Principal Investigator(s): Chao Fan (Clemson University)

Project Partners: Lingxi Li (Purdue University), Satish Ukkusuri (Purdue University), Latifur Khan (The University of Texas at Dallas), Bhavani Thuraisingham (The University of Texas at Dallas), Imtiaz Karim (The University of Texas at Dallas).

Research Project Funding: Federal $227,369; Cost-share $220,000

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This project develops SentinelLab, a closed-loop defender testbed designed to transform how Connected and Automated Vehicle (CAV) cybersecurity is validated. Currently, most research stops at detecting anomalies; this project bridges the gap to active defense by integrating the METS-R traffic simulator with the CARLA photo-realistic sensor simulator. The testbed utilizes a “Recognize, Decide, Act” framework. The project employs a multimodal Large Language Model (LLM) to recognize specific attack families (e.g., message replay, breaking provocation) from noisy streaming signals. It then uses an online Defender Workbench to decide on mitigation strategies via plug-in policies and automatically executes these actions in the simulation. This system enables researchers and public agencies to prepare defenses against realistic threats and quantify their impacts on safety and mobility.

US DOT Priorities: This project supports the US DOT’s strategic goal of Safety by creating a mechanism to audit and validate defenses against cyber-physical attacks before they are deployed on public roads. It advances transformation by shifting the industry focus from passive intrusion detection to active, automated resilience, thereby reducing transportation cybersecurity risks. The research is transformative because it introduces the first interoperable benchmark that measures the success of a cyber-defense not just by detection rates, but by physical safety metrics like time-to-collision and hard-braking events. This directly contributes to a more reliable, resilient transportation system that can withstand emerging threats without compromising mobility.

Outputs: The primary output is the SentinelLab Online Defender Testbed, a software prototype featuring three core components.

• Multimodal Attack Recognition Model: An AI model that aligns asynchronous Vehicle to Everything (V2X) signals and camera feeds to classify attack types in real-time with explainable reasoning.
• Defender Workbench API: A standardized interface and strategy playbook that allows researchers to plug in different detection algorithms and automated defense policies.
• Benchmark Dataset: A curated, multimodal streaming dataset containing labeled attack scenarios (e.g., replay, route manipulation) to serve as a community standard for evaluation. An onboarding playbook and tutorials will be generated to facilitate technology transfer to agencies and partners.

Outcomes/Impacts: The proposed SentinelLab testbed is positioned to fundamentally change the practice of CAV cybersecurity from reactive, ad-hoc patching to proactive, validated resilience. Currently, agencies lack the tools to verify if a cyber-defense works without disrupting traffic. This output changes that paradigm by providing agencies and original equipment manufacturers with an audit-ready environment where they can rehearse defense playbooks against realistic threats and generate empirical evidence of effectiveness before deployment. This capability enables a significant shift in procurement and practice, allowing agencies to require validated performance metrics from vendors rather than relying on theoretical claims. Furthermore, the standardized event schemas and safety metrics generated by this research will directly inform regulatory frameworks and rulemaking on V2X validation. By establishing a shared definition of safe mitigation, such as defining acceptable thresholds for vehicle deceleration during a defense action, the project provides the necessary data for regulators to set minimum standards for cyber-resilient operations.

In terms of system performance, the research will deliver measurable improvements in safety, reliability, and cost-efficiency. By automating the Recognize-Decide-Act loop, the system significantly reduces the time-to-mitigation, allowing traffic networks to recover from disruptions faster and minimizing trip-time losses. The testbed enhances safety by reducing the rate of hard-brake events and near-misses caused by cyber-attacks, ensuring that defensive actions do not inadvertently create new physical risks. Economically, the creation of an open-core benchmark reduces vendor lock-in, fostering a competitive market where defense plug-ins are interoperable, thereby lowering long-term integration costs. The project is expected to result in copyrightable software assets for the Defender Workbench and licensing opportunities for the advanced attack models, facilitating commercial scale-up and broader industry adoption.

Principal Investigator(s): Yunyi Jia (Clemson University)

Project Partners: Ardalan Vahidi (Clemson University), Judith Mwakalonge (South Carolina State University), and Jagruti Sahoo (South Carolina State University).

Research Project Funding: Federal $102,000; Cost-share $102,010

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: The need for research on experimental prototyping and demonstration of cyberattack mitigation strategies in Connected and Automated Traffic (CAT) is critical as modern transportation systems increasingly depend on interconnected and automated vehicle operations. With the growing adoption of Vehicle-to-Everything (V2X) communication, the interactions and cascading effects that arise within real connected traffic systems caused by cyberattacks need to be investigated in a timely manner.

The proposed project will deliver an integrated experimental framework and multi-layer mitigation strategy to strengthen cybersecurity in CAT systems. Key highlights include: Cyber-CAT System Development: Establish a real and virtual hybrid testbed integrating CAVs, infrastructure, and C-V2X communication to enable realistic testing of cyber threats and countermeasures; Multi-layer Mitigation: Prototype and validate Zero-Trust-LLM-assisted mitigation and AI-based intention sharing mitigation to maintain safety under attack conditions; and Guidance and Knowledge Transfer: Produce data and design guidelines to inform US DOT and industry on C-V2X security standards and resilient corridor deployment.

US DOT Priorities: This research supports US DOT priorities and the RD&T strategic goals by advancing secure, data-protected, and resilient transportation systems that maintain safe, efficient, and sustainable operations even under cyberattack conditions. The outcomes will directly enhance the safety, resiliency, and efficiency of automated transportation systems. Anticipated impacts include (a) Improved Safety: reducing vulnerability to cyber threats and maintaining cooperative awareness; (b) Traffic Resilience: ensuring smooth, stable flow under degraded communications; and (c) Operational Efficiency: minimizing downtime, disruptions, and recovery costs for agencies and industry stakeholders.

Outputs: This project will develop and experimentally demonstrate a new cybersecurity-enabled Connected and Automated Traffic (Cyber-CAT) system that enhances the safety, resilience, and trustworthiness under cyber threats. The expected outputs from the proposed research tasks are listed below:

• Development of Cyber-CAT: A hybrid testbed integrating real connected vehicles, roadside units, and a digital-twin traffic simulation environment, with built-in capability to inject and study cyberattacks.
• High-Level Mitigation (Zero-Trust-LLM-Assisted Mitigation): An integrated software module that evaluates the trustworthiness of V2X messages using Zero-Trust principles and LLM-based reasoning to maintain cooperative control under attack.
• Low-Level Mitigation (AI-based Intention Prediction and Sharing): A low-level mitigation module that predicts and shares vehicle intentions to preserve safety and traffic stability when communications are degraded or disrupted.
• System-Level Evaluations: An integrated experimental demonstration and evaluation of the Cyber-CAT system, including safety, resilience, and traffic-efficiency performance under cyberattack scenarios.

Furthermore, to demonstrate the proposed research, this project will develop an integrated Cyber-CAT prototype that combines real connected vehicles, roadside units, and a digital-twin traffic environment for testing cybersecurity and mitigation strategies. The prototype will include an automated Ford Mach-E with C-V2X communication, edge-computing hardware, and software modules for AI-assisted mitigation and AI-based intention prediction and mitigation sharing. These components will enable safe and repeatable testing of cyberattack scenarios and resilience strategies. The modular architecture and benchmark data developed through this effort will allow other researchers and agencies to adapt the Cyber-CAT prototype for future studies in secure cooperative automation and connected corridor operations.

Outcomes/Impacts: The proposed project will deliver both tangible research outputs and measurable societal impacts toward building secure and resilient Connected and Automated Traffic (CAT) systems. Expected outputs include a prototyped Cyber-CAT platform with new mitigation strategies and real-world demonstrations, a patent/invention disclosure, experimental data, and peer-reviewed publications. These innovations will provide a foundation for developing new industry standards, policy guidelines, and technical best practices in CAT cybersecurity. The outcomes will directly enhance transportation system safety and reliability by enabling proactive detection and mitigation of cyber threats, preserving network stability, and maintaining efficient traffic flow under adverse conditions. In the long term, this research will inform USDOT’s connected corridor initiatives, support secure deployment of cooperative automation technologies, and promote public trust in intelligent transportation systems through demonstrable improvements in system performance, resilience, and operational integrity.

Principal Investigator(s): Trayce Hockstad (The University of Alabama, Tuscaloosa)

Project Partners: Kun Lu (The University of Alabama, Tuscaloosa), Steven Jones (The University of Alabama, Tuscaloosa), Latifur Khan (The University of Texas at Dallas), Bhavani Thuraisingham (The University of Texas at Dallas), Eric Morris (Clemson University), M. Sabbir Salek (Clemson University).

Research Project Funding: Federal $123,198; Cost-share $210,184

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This project is designed to strengthen transportation cybersecurity in an era where AI-enabled and AI-enhanced cybercrime is increasingly capable of scaling deception, automation, and attack sophistication against cyber-physical systems. The project builds directly on the team’s prior legislative gap analysis and on the development of TraCR AI, a retrieval-augmented large language model that helps transportation officials and policymakers identify applicable legal obligations and compare regulatory approaches across jurisdictions. The central technical contribution is the development and testing of a modular defensive wrapper for transportation-focused large language models (LLMs) and retrieval-augmented generation (RAG) tools, intended to detect and mitigate adversarial attacks that exploit legal reasoning systems and to support a testbed for LLM-targeted cybercrime scenarios. In parallel, the project includes legal and policy research that assesses gaps in US frameworks for addressing AI-enabled cybercrime and draws on international examples to inform best practices for governance, enforcement, and secure deployment. The overall objective is to produce deployable defenses and practitioner-facing guidance that improve trustworthiness in AI-assisted compliance and policy analysis for critical transportation infrastructure.

US DOT Priorities: This project supports US DOT priorities by addressing an emerging threat to transportation safety and system reliability: cyberattacks that exploit AI tools and AI-shaped information environments. It advances DOT RD&T strategic goals by strengthening infrastructure resiliency, improving cybersecurity governance, and giving public agencies practical methods to assess legal obligations and policy gaps in a fragmented regulatory landscape. The project is also transformative because it treats LLMs as both decision-support tools and new attack surfaces, then builds a transportation-specific testbed to model adversarial behavior and evaluate defenses under realistic conditions. By pairing a modular defensive wrapper with focused legal research on AI-enabled and AI-enhanced cybercrime, the work establishes a scalable framework for trustworthy AI deployment in critical infrastructure contexts.

Outputs: This project is expected to produce a set of technical and legal outputs that strengthen transportation cybersecurity and improve the trustworthiness of AI-assisted policy analysis. The primary technology output will be a modular defensive wrapper for transportation-focused LLM and RAG systems that screens adversarial prompts, reduces hallucination-driven legal error, enforces source-grounded responses, and records suspicious interactions for auditing and refinement. The project will also deliver an adversarial test suite and a repeatable evaluation process to benchmark LLM vulnerabilities and defense performance in transportation compliance and policy workflows. On the research side, the team will produce a testbed framework for simulating LLM-targeted cybercrime scenarios and a legal gap analysis of US law addressing AI-enabled and AI-enhanced cybercrime, supplemented by comparative international examples and policy recommendations. These outputs will support more consistent public agency processes for identifying applicable legal requirements and deploying AI tools responsibly in critical infrastructure contexts. The project is also expected to establish or deepen partnerships outside the UTC consortium, particularly with state DOTs and other public agencies engaged in demonstrations, survey collaboration, and applied feedback on the prototype and guidance.

Outcomes/Impacts: The project outputs will be applied as practical tools and guidance to reduce cybersecurity risk in transportation systems and improve the safe deployment of AI in public-sector decision workflows. The modular defensive wrapper and adversarial test suite will help agencies and vendors harden transportation-facing LLM and RAG applications before use, reducing hallucination-driven legal error, preventing adversarial manipulation, and improving source-grounded reliability in compliance and incident-response contexts. This strengthens safety and operational reliability by lowering the chance that cybersecurity planning or regulatory decisions are based on inaccurate or compromised outputs, and it can reduce long-run costs by avoiding preventable missteps and reactive remediation. The legal gap analysis and comparative international review will translate technical findings into policy recommendations that can inform DOT cybersecurity strategies, legislative priorities, and agency guidance on responsible generative AI use. Over time, these outputs support more durable governance, stronger resilience against AI-enabled cybercrime, and more consistent practices across jurisdictions.

Principal Investigator(s): Mansoureh Jeihani (Morgan State University)

Project Partners: Mansha Swami (Morgan State University), Ehsan Mehryaar (Morgan State University), Shubham Agrawal (Clemson University), Dustin Souders (Clemson University).

Research Project Funding: Federal $80,000; Cost-share $88,244

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: Phase II will examine driver behavior and decision-making under cyberattacks in connected-vehicle contexts using high-fidelity, human-in-the-loop driving simulators at Morgan State University (urban) and Clemson University (suburban). The team will develop reusable Cyberattack Injection Modules (CIMs) for UCWinRoads and SimCreator, and validate one vehicle-centered attack (false blind-spot warning) and two infrastructure-centered attacks (falsified signal phase-and-timing and “phantom” signal-ahead information). The study will leverage IRB-approved human-subject testing, conduct a Maryland MVA pilot demonstration, and curate a Standardized Attack Scenario Library (ASL) for replication and training use.

US DOT Priorities: This project supports US DOT’s priorities, Safety, Emerging Technologies and focus area of Cybersecurity by reducing transportation cybersecurity risks through human-centered evaluation of manipulated connected-vehicle information and its safety-critical impacts. It advances RD&T goals by producing reusable, open-access modules (products) and standardized scenarios (tech transfer) that enable repeatable evaluation and faster translation from isolated studies to scalable research and practice.

Outputs: The specific outputs from this project include the following:

• Prototype urban and suburban simulator environments with connected-infrastructure features.
• Cyberattack Injection Modules (CIMs) for UCWinRoads and SimCreator, covering false blind spot warnings, falsified signal phase and timing, and “phantom” traffic signal ahead.
• CIM technical specifications (configurable parameters), documented code, and guided examples, shared via open-access platforms.
• Maryland MVA pilot demonstration deliverables, including practitioner feedback report and policy-oriented recommendations for driver-education relevance.
• Attack Scenario Library (ASL) with scenario specifications, configurations, message standards, participant instructions, and implementation recommendations; published with open access.
• Human-subject simulator dataset and comparative urban–suburban analysis, peer-reviewed manuscript(s).

Outcomes/Impacts: The key outcomes from this project are:

• Clear evidence on how drivers attribute anomalies (cyberattack vs. system fault), their confidence, trust/usability shifts, and safety-critical behavioral adaptation under attack conditions.
• Reusable CIMs + ASL lower adoption barriers, enabling other organizations to replicate and extend validated scenarios for research and workforce training.
• Comparative findings (urban vs. suburban) informing context-specific guidance for safer, more resilient connected-vehicle applications.
• MVA engagement strengthens real-world relevance and supports future driver education and policy preparedness initiatives.
• ASL documentation supporting transition planning toward evaluation in connected-vehicle testbeds and V2X/digital-twin workflows.

Principal Investigator(s): Mert Pesé (Clemson University)

Project Partners: Z. Berkay Celik (Purdue University).

Research Project Funding: Federal $87,000; Cost-share $92,813

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: Modern vehicles use software-defined vehicle (SDV) platforms that integrate functions via virtualization, but current designs lack resiliency against security incidents or hardware obsolescence. This project aims to enhance vehicle security by developing and evaluating secure live-migration techniques for virtual-machine (VM)-based workloads. By allowing actively running services to move between electronic control units (ECUs) without interruption, the project enables real-time upgrades and mitigation of cyberattacks within next-generation zonal architectures.

US DOT Priorities: This project supports the USDOT strategic goal of Safety by ensuring that critical vehicle software remains available even during hardware failures or security breaches. It engages in transformative research by adapting cloud-native live migration for the unique, real-time constraints of automotive cyber-physical systems. The work also addresses infrastructure resilience by providing a pathway to upgrade aging hardware in long-lived fleets.

Outputs:

• A novel, open-source secure live migration protocol specifically designed for automotive zonal ECU architectures.
• An implementation-ready design for a Zonal Architecture Testbed (ZAT).
• New datasets for training intrusion detection systems and performing network analysis.
• Publications detailing secure migration protocol specifications and testbed design.
• Collaboration with DENSO, a leading automotive Tier 1 company.

Outcomes/Impacts: The research will provide original equipment manufacturers (OEMs) and Tier 1 suppliers with a framework to maintain service quality and security over a vehicle’s lifespan. By establishing formal security guarantees for VM migration, the project reduces the risk of unauthorized access or state tampering during software transitions. Ultimately, this improves the reliability and durability of transportation systems by allowing for seamless security patching and hardware modernization without service interruptions.

Principal Investigator(s): Mizanur Rahman (The University of Alabama, Tuscaloosa)

Project Partners: Ahmad Alsharif (The University of Alabama, Tuscaloosa), Sagar Dasgupta (The University of Alabama, Tuscaloosa), Mashrur Ronnie Chowdhury (Clemson University), Mohammadhadi Amini (Florida International University), Naphtali Rishe (Florida International University).

Research Project Funding: Federal $152,235; Cost-share $152,273

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: Cellular Vehicle-to-Everything (C-V2X) communication, standardized in 3GPP Releases 14 and 15, PC5 sidelink mode, is the US DOT-approved technology for direct V2V (vehicle-to-vehicle)/ V2I (vehicle-to-infrastructure) communications in the 5.9 GHz band. Current standards and specifications (e.g., SAE J3161 and USDOT/ITE RSU requirements) mandate PC5 Mode 4 operation to enable interoperable safety messaging using conventional cryptographic methods, such as Elliptic Curve Cryptography (ECC). However, existing cryptographic methods are vulnerable to quantum-computing-based attacks. Thus, integrating Post-Quantum Cryptography (PQC) into C-V2X communication is imperative to ensure future resilience. However, National Institute of Standards and Technology (NIST)-standardized PQC algorithms introduce large key sizes and computational complexity, resulting in significant latency and bandwidth overhead. These effects risk violating the 100-ms end-to-end delay requirement for 10 Hz Basic Safety Messages (BSMs) and can congest the 5.9-GHz safety channel. Moreover, the direct integration of PQC into current communication standards, such as IEEE 1609.2 and ETSI, poses challenges because these frameworks were originally designed for lightweight ECC-based operations. Similarly, post-quantum Homomorphic Encryption (HE) offers robust privacy protection by allowing computation directly on encrypted data without decryption; however, its high computational cost and ciphertext expansion currently limit its use in latency-critical V2X and infrastructure-to-infrastructure (I2I) scenarios. Therefore, deploying PQC and HE within operational testbeds demands optimized scheduling, resource allocation, and adaptive algorithm management to balance cryptographic strength with real-time constraints. To address these challenges, this project aims to develop and evaluate AI-enabled PQC through real-world prototype implementation and testbed integration, thereby enabling the real-world deployment of secure and resilient communication in intelligent transportation systems. Specifically, the objectives of this project are: (i) implementation and real-world evaluation of an AI-enabled PQC integration and dynamic switching framework for C-V2X communication; (ii) real-world evaluation of a privacy-preserving roadside unit (RSU)-Cloud (I2C) communication pipeline using post-quantum homomorphic encryption; and (iii) development of a federated learning framework for collaborative PQC selection policies. To address the USDOT and TraCR 2025–2026 priorities, this project emphasizes field-tested prototypes and operational validation, rather than simulation-only evaluation, to ensure deployment relevance. This project will directly contribute to the deployment of PQC-enabled V2X communication for a secure and reliable connected transportation system.

US DOT Priorities: This project will directly focus on the TraCR’s “Thrust 4: Evolving Quantum Computing Threats and Opportunities.” In addition, this project will address US DOT’s strategic goals related to securing transportation systems as well as help make the US DOT the worldwide leader in transportation cybersecurity, and help ensure that American firms stay at the forefront of the global economy, and help keep inflation low by fostering the safe, efficient, and bottleneck-free movement of goods and workers (“Economic Strength and Global Competitiveness,” “Organizational Excellence”). To address the US DOT and TraCR 2025–2026 priorities, this project focuses directly on prototype development and real-world testbed integration.

Outputs: The outcomes of this research will be a real-world prototype of AI-enabled PQC and privacy-preserving encryption schemes on operational C-V2X testbeds using Cohda Wireless MK6 on-board units and roadside units. The prototype will demonstrate quantum-resilient, low-latency V2X communication in realistic field environments and provide a deployment framework that is adaptable to DOTs, OEMs, and transportation agencies. This experimental validation will ensure that the proposed innovations go beyond simulation and contribute measurable, deployable outcomes consistent with the US DOT’s emphasis on tangible, field-tested cybersecurity solutions. Supporting outputs include curated datasets, trained AI and federated learning models, software artifacts, configuration files, and deployment guidelines suitable for adoption by DOTs and industry stakeholders.

Outcomes/Impacts: The evolution of connected and automated vehicles (CAVs) into safe, efficient, and reliable components of the mainstream transportation system largely depends on rapid, innovative technological advancement. With time, more CAVs are expected to join the regular vehicle fleet. A deployable PQC solution for vehicular networks that can ensure security against quantum-based attacks is needed, according to the National Institute of Standards and Technology (NIST) framework. This research will directly contribute to this direction. The outcomes will inform future intelligent transportation systems (ITS) security architectures, the evolution of standards, and DOT deployment strategies by providing experimentally validated guidance on balancing quantum security, latency, and system scalability.

Principal Investigator(s): Mizanur Rahman (The University of Alabama, Tuscaloosa)

Project Partners: Thejesh N. Bandi (The University of Alabama, Tuscaloosa), Sagar Dasgupta (The University of Alabama, Tuscaloosa), Long Cheng (Clemson University).

Research Project Funding: Federal $145,866; Cost-share $151,097

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: Global Navigation Satellite Systems (GNSS), such as the Global Positioning System (GPS), form the backbone of modern positioning, navigation, and timing (PNT) services. However, these space-based systems are inherently vulnerable to cyberattacks such as jamming, spoofing, and unintentional interference, including signal blockage, particularly in dense urban areas, indoor environments, and adversarial environments. The growing dependence on GNSS, driven by the rapid adoption of autonomous and connected systems, has exposed a single point of failure in the global PNT infrastructure. GPS signals are extremely weak at the Earth’s surface, enabling low-cost jammers or spoofers to easily disrupt receivers. In response to the 2020 Executive Order on strengthening national resilience through responsible use of PNT services signed by President Donald J. Trump, US DOT, the Department of War (DoW), and the Department of Homeland Security (DHS) have jointly emphasized the need for complementary and backup PNT capabilities that are interoperable and independently capable of sustaining precision timing and navigation for critical infrastructure during GNSS outages or cyberattacks. The research goal is to develop and demonstrate a prototype ground-based, GPS-compatible, cyber-secure PNT architecture that can generate, synchronize, and broadcast authenticatable GPS-like signals from a network of ground-based nodes, allowing existing GPS receivers to obtain valid PNT solutions without hardware modification. This goal will be achieved through the following specific research objectives:

• Design and generate authenticable GPS-compatible terrestrial signals that replicate the L1 C/A (coarse/acquisition) waveform while embedding virtual ephemeris and adjusted clock-offset parameters to enable accurate and PNT computation from ground transmitters.
• Develop intelligent terrestrial nodes (at least four nodes) equipped with chip-scale atomic clocks, edge computer, and transmitters to establish a distributed ground-based PNT architecture.
• Synchronize terrestrial nodes with a master clock using precision timing distribution techniques to maintain consistent and reliable time alignment across the network. Real-Time Kinematic (RTK) positioning and differential methods will also be explored using the GEODNET hub within the UA network. Objective 4: Demonstrate that an off-the-shelf GPS receiver can deliver a valid PNT solution using terrestrial signals through software-only modifications, thereby validating the practicality, backward compatibility, and deployment readiness of the proposed system.

Outputs: This project will have the following outputs:

• A lattice reduction algorithm using the Singular Value Decomposition (SVD) approach, tailored to address security implications for q-ary lattices in PQC schemes,
• Fault-tolerant certificate segmentation algorithms designed for integrating PQC into the IEEE 1609.2 standard, enabling reliable, low-latency, quantum-resilient C-V2X communication, and
• Evaluation of PQC scheme performance utilizing federated learning-based applications in connected transportation systems.

US DOT Priorities: This project directly aligns with the vision of TraCR to enable a secure and resilient transportation cyber-physical system (TCPS) and the strategic objectives of US DOT by advancing secure, resilient, and trustworthy PNT technologies that underpin modern transportation systems. The research aligns most closely with TraCR’s Thrust 1: Security and Resiliency, as it addresses vulnerabilities in PNT systems by developing ground-based alternatives that enable real-time authentication and resilience against spoofing and jamming attacks. Through collaboration with GEODNET and controlled pilot deployment and validation, the project supports TraCR’s 2025–2026 emphasis on practical, deployable solutions that demonstrate measurable improvements in transportation cybersecurity and resilience.

Outputs:

• A ground-based PNT architecture focused on the design of GPS-compatible terrestrial signals that replicate the L1 C/A waveform while implementing virtual ephemeris and controlled clock-offset methods, representing a major innovation in signal design for resilient navigation;
• A network of four ground-based nodes synchronized with a master atomic clock, GNSS receivers, and secure transceivers to form a distributed terrestrial PNT testbed;
• A prototype ground-based PNT architecture comprising terrestrial signal nodes, which enable any off-the-shelf GPS receiver, modified through software only, to compute a valid PNT solution, thereby validating backward compatibility and deployment feasibility; and
• Experimental validation datasets demonstrating system performance under live-sky legitimate GNSS signals and controlled spoofing scenarios generated using a GNSS simulator and software-defined radios.

Outcomes/Impacts: The proposed project will deliver a complete and validated terrestrial PNT framework that demonstrates secure, GPS-compatible navigation and timing without requiring hardware modifications to existing receivers. The key outcomes include a software toolchain for generating GPS-like terrestrial signals with embedded virtual ephemeris and controlled clock offsets, a synchronized terrestrial PNT testbed with network-based timing control, and a master clock synchronization framework integrating White Rabbit and IEEE-1588 PTP protocols. The project will also produce a pilot-deployed terrestrial PNT system validated through off-the-shelf GPS receivers, confirming backward compatibility and deployment feasibility. Non-sensitive software modules, datasets, configuration files, and validation scripts will be openly shared to promote independent benchmarking, reproducibility, and future research. The results will directly advance national efforts toward resilient, authenticated PNT infrastructure, strengthen the cybersecurity posture of transportation systems, and provide a foundational architecture for next-generation positioning technologies. The research findings and methodologies will be disseminated through peer-reviewed publications, conference presentations, and collaboration with industry and government stakeholders to accelerate adoption and transition to large-scale deployment.

Principal Investigator(s): Zhenkai Zhang (Clemson University)

Project Partners: Balaji Iyangar (Benedict College).

Research Project Funding: Federal $119,626; Cost-share $127,458

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This project develops a novel electromagnetic (EM) side-channel-guided fuzzing framework for automotive Electronic Control Unit (ECU) firmware security testing. The approach addresses key challenges in ECU security research, namely that firmware is often encrypted, proprietary, and tightly coupled to hardware, making traditional instrumentation-based fuzzing impractical. By capturing and analyzing EM emanations from ECUs during execution, the framework estimates code coverage without requiring firmware modification, instrumentation, or rehosting. The system integrates this EM-based coverage feedback into a fuzzer to guide test case generation via Controller Area Network (CAN) bus communication. The project will conduct extensive fuzzing campaigns on real automotive ECUs from various manufacturers to discover zero-day vulnerabilities and enhance vehicle cybersecurity.

US DOT Priorities: This project directly supports US DOT priorities for transportation safety and security, specifically aligning with the DOT's emphasis on cybersecurity for connected and automated vehicles. The work addresses the RD&T strategic goal of improving transportation safety by proactively identifying and mitigating security vulnerabilities in critical vehicle systems before they can be exploited by malicious actors. The project supports TraCR’s Thrust 1: Security and Resiliency and Thrust 2: User and Data Privacy by developing tools to uncover firmware vulnerabilities that could be exploited to compromise vehicle control systems or user data. This research is a breakthrough and transformative as it pioneers the use of physical side-channel information (EM emanations) to guide firmware fuzzing, which is a fundamentally new approach that bypasses the need for source code access or firmware instrumentation/rehosting to perform security testing on encrypted, proprietary ECU firmware that cannot be analyzed using conventional methods.

Outputs: The project will produce the following outputs:

• A working prototype of an ECU fuzzing tool that uses EM-based feedback (software, scripts, and hardware test setup guide).
• Evaluation results on multiple ECUs, including reproducer message sequences and technical reports for any confirmed issues, were shared through coordinated vulnerability disclosure.
• Outreach materials for technology transfer, such as a demo package and a TraCR webinar presentation.

Outcomes/Impacts: The project will lead to the following outcomes and impacts:

• A more practical way to test modern ECU firmware when source code, debug access, or instrumentation are unavailable.
• Earlier discovery of firmware flaws that could lead to unsafe behavior, service disruption, or privacy loss.
• Reduced time and cost for finding deep bugs compared to purely black-box testing.
• Tools and setup guidance that can be adopted by research labs and, with partner engagement, by industry testing teams.
• Improved overall security for connected vehicles by supporting pre-deployment testing and faster remediation.

Principal Investigator(s): Rong Ge (Clemson University)

Project Partners: M. Sabbir Salek (Clemson University), Jagruti Sahoo (South Carolina State University).

Research Project Funding: Federal $84,247; Cost-share $85,836

Project Start and End Date: April 1, 2026, to March 31, 2027

Project Status: Report in Progress

Project Description: This project strengthens the cybersecurity and resiliency of camera-based perception for autonomous vehicles by addressing two fast-growing attack classes: universal adversarial perturbations (UAPs) and generative/deepfake-style scene manipulations that can add, alter, or remove objects in the camera feed. The team will first build and validate a novel hybrid attack that combines image-agnostic UAP noise with generative object-disappearance attacks (ODAs), using real-time inpainting to create “hallucination”- driven scenes in which objects are misclassified or vanish entirely. The project will also develop a quantum-enhanced hybrid defense that fuses parameterized quantum circuits with classical deepfake/manipulation detection, leveraging quantum–classical disagreement and out-of-distribution signals to robustly detect both pixel-level perturbations and semantic object edits. The project will produce deployable prototypes: (1) a real-time hybrid “malware” attack pipeline and (2) a quantum-infused hybrid detector, which will be evaluated in realistic AV scenarios and deployed for testing on connected-vehicle testbeds (e.g., Clemson University Connected Vehicle Testbed or CU-CVT and Morgan State).

US DOT Priorities: This project directly advances US DOT’s Safety and Transformation priorities by reducing transportation cybersecurity risk in a safety-critical autonomy function, i.e., camera-vision perception, where successful attacks can cause misclassification (e.g., traffic signs) or make critical objects (e.g., pedestrians) appear/disappear. These objectives directly align with US DOT’s priority, Cybersecurity: Secure and Resilient by Design, and TraCR’s Thrust 1: Security and Resiliency, by hardening AV perception under real-time adversarial conditions, and with TraCR’s Thrust 4: Emerging Quantum Computing Threats and Opportunities, by using quantum architectures to strengthen integrity and detectability in manipulated scenes. This project advances the state of the art in several respects. First, it develops a novel end-to-end “hallucination scenario” attack that combines universal adversarial perturbations (UAPs) with real-time generative object-disappearance techniques, moving beyond conventional pixel-level corruption to semantic manipulations that can make safety-critical objects vanish or appear, and delivering a malware-grade prototype plus an open benchmarking dataset for reproducible evaluation. Second, it creates an advanced quantum-enhanced hybrid defense that fuses a parameterized quantum-circuit detection path with a classical manipulation detector, leveraging quantum–classical disagreement signals alongside theory-guided robustness controls (e.g., Lipschitz-constrained learning) to strengthen detection and provide certified resilience against both pixel-level adversarial perturbations and generative/deepfake manipulations. Third, it emphasizes transformative deployability by deploying and evaluating both the hybrid attack and defense in connected-vehicle testbeds and packaging the technology for broad adoption through modular components, integration with widely used ML/quantum toolchains (e.g., PyTorch, TorchQuantum, PennyLane), containerized releases, an evaluation harness with regression gates, and operational support artifacts including provenance tracking, a model registry, and rollback procedures.

Outputs: The main outputs include:

Open-source software prototypes:

• A real-time hybrid attack “malware” prototype that hooks into an autonomous vehicle camera stream and produces a UAP-ODA manipulated video feed in real time (universal adversarial perturbations + generative object disappearance via inpainting).
• A quantum-infused hybrid detector prototype that detects both pixel-level adversarial perturbations and semantic/generative manipulations in real time, intended for adoption by AV researchers and developers.

New detection methods:

• A quantum–classical fusion detection method that uses parallel quantum and classical paths and leverages quantum–classical disagreement / OOD signals to improve detection of semantic edits.
• A theory-and-practice package for certifiable robustness, including parameter-dependent Lipschitz bounds and theory-guided Lipschitz controls implemented as training losses/constraints.

Benchmarking and datasets:

• An open hybrid-attack dataset of manipulated driving scenes derived from established AV benchmarks (e.g., KITTI/Cityscapes/nuScenes), enabling reproducible evaluation of defenses against advanced attacks.
• Benchmarked evaluation results and performance reports quantifying attack effectiveness and detection/latency tradeoffs, including planned testing on standard driving datasets.

Outcomes/Impacts: The project will positively impact the transportation system by improving safety, reliability, durability, and cost-effectiveness of camera-based autonomy: it reduces the risk of catastrophic perception failures caused by adversarial patches/UAPs or deepfake-like generative edits that can mislead detection/segmentation or make critical obstacles “disappear”; it increases reliability and resilience across varied operating conditions through testbed validation and continuous regression gating that reduces brittleness as models and environments change; it strengthens durability and maintainability via adoption-ready release engineering (containerization, stable APIs, provenance tracking, model registry, and rollback procedures) so defenses remain usable and updatable as new attacks emerge; and it supports cost and scalability by hardening low-cost camera perception, preserving its affordability relative to more expensive sensing modalities while lowering cybersecurity risk and enabling safer deployment at scale.

Principal Investigator: Mashrur Ronnie Chowdhury (Clemson University)

Project Partners: M Sabbir Salek (Clemson), Balaji Iyangar (Benedict), Hadi Amini (FIU), Selcuk Uluagac (FIU), Naphtali Rishe (FIU), Mansoureh Jeihani (MSU), Ehsan Mehryaar (MSU), Satish Ukkusuri (Purdue), Biswajit Biswal (SCSU), Jagruti Sahoo (SCSU), Judith Mwakalonge (SCSU), Mizanur Rahman (UA), Latifur Khan (UTD), Bhavani Thuraisingham (UTD), Alvaro Cardenas (UCSC)

Research Project Funding: Federal $1,422,841; Cost-share $1,595,593

Project Start and End Date: June 1, 2025, to May 31, 2026

Project Status: Report in Progress

Project Description: TraCR’s foundational project aims to develop technological tools, prototypes, testing platforms, and facilities to ensure the cybersecurity and cyber-resilience of multimodal transportation systems and related infrastructure. The project is led by Clemson University (Clemson) under the strategic direction of Dr. Ronnie Chowdhury (Lead PI), with coordination support from Dr. Sabbir Salek (Co-PI), and involves all eight other TraCR partner institutions organized into four subgroups. A structured project governance framework, including biweekly subgroup meetings, monthly full-team coordination meetings, quarterly progress reporting and advisory board engagement, ensures alignment with project milestones, integration across teams, and effective monitoring of technical progress and deliverables.

Clemson University collaborates with Benedict College (Benedict), South Carolina State University (SCSU), and the University of Texas at Dallas (UTD) to advance a comprehensive, automated threat modeling capability for multimodal transportation systems. Building on the Transportation Cybersecurity and Resiliency Threat Modeling Framework (TraCR-TMF), the team conducts testbed-in-the-loop evaluations within Clemson’s real-world cybersecurity testbed, implementing digital-twin-based cybersecurity analysis of in-vehicle networks, and engaging state transportation agencies to assess operational transferability. Additionally, the team will work to integrate graph-based reasoning models into threat modeling, deploy supervised ModernBERT classifiers, and align with the MITRE Embedded Systems Threat Matrix to strengthen structured system-to-vulnerability mapping and improve threat coverage across transportation cyber-physical systems.

The other partner institutions will develop additional real-world and virtual testing platforms to support cybersecurity experimentation for multimodal transportation. Florida International University (FIU) and the University of Alabama at Tuscaloosa (UA) are jointly advancing the Open-Source Connected and Automated Mobility Co-Simulation (OpenCAMS) environment and related simulation platforms, integrating SUMO, CARLA, and network simulation tools, to evaluate privacy-aware multimodal large language models and post-quantum-secure C-V2X communications. Their efforts further include the development and validation of spoofing attack models targeting Basic Safety Message transmissions and multi-frequency GPS receivers, as well as investigations into backdoor-resilient perception systems and the security of vision-language models for intelligent transportation applications.

Purdue University (Purdue) and the University of California, Santa Cruz (UCSC) are advancing adversarial testing methodologies through integrated physical-virtual experimentation frameworks that combine miniature autonomous vehicle testbeds, CARLA/METS-R simulation coupling, and scenario-based vulnerability discovery. These activities include simulation-to-real validation of perception and traffic signal spoofing attacks, evaluation of V2X safety message vulnerabilities, cybersecurity analysis of shared micromobility Bluetooth pairing protocols, implementation of lightweight post-quantum cryptographic protections for vulnerable road user beacons, and closed-loop security assessments of traffic signal controller infrastructures, along with investigations of secure multimodal AI agents and memory-augmented reasoning architectures for autonomous robotic transportation systems.

In addition, Morgan State University (MSU) is enhancing its connected vehicle cybersecurity experimentation capabilities by developing replay-attack models targeting C-V2X onboard units and evaluating mitigation strategies in its real-world testbed environment, in collaboration with Clemson University. These efforts quantify communication-level impacts on safety-critical applications and support the development of deployable countermeasures to strengthen resilience against wireless attack vectors affecting connected transportation infrastructure.

US DOT Priorities: This project aligns with the U.S. DOT priority Cybersecurity: Secure and Resilient by Design through a two-pronged approach. By developing a comprehensive, dedicated threat modeling tool for multimodal transportation systems, the project supports detailed assessments of system- and communication-level security vulnerabilities and identification of corresponding mitigation strategies informed by widely used industry tools, databases, and frameworks. Concurrently, the project advances transportation cybersecurity testing capacity through multiple real-world, miniature, virtual, and hybrid testbeds.

Outputs: The outputs of TraCR foundational projects include tools, frameworks, datasets, and testing platforms shared publicly through open-source libraries, prototypes, and peer-reviewed publications. TraCR-TMF, developed by Clemson, Benedict, SCSU, and UTD researchers, is an LLM-supported threat assessment tool publicly available for adaptation to multimodal transportation systems. Planned Year 3 outputs include extending the published tool to leverage additional cybersecurity frameworks, guidelines, and best practices, and integrating human-in-the-loop feedback. The OpenCAMS platform, developed by FIU and UA, is fully open source, with its first version publicly available, providing the research community with an accessible, flexible, and collaborative environment for advancing cybersecurity research of next-generation transportation systems. Planned outputs include new security testing capabilities, such as post-quantum digital signature algorithms for C-V2X communication, and demonstrations of cybersecurity mitigation strategies using the platform. Work led by Purdue and UCSC advances the Scenic open-source ecosystem by introducing D4+ and METS-R security APIs to support safer validation of autonomous transportation systems. By expanding Scenic with worst-case, optimization-guided scenarios, the work improves the realism, coverage, and security relevance of scenario-based testing. This physical-virtual testing platform also supports configurable disturbances, including replayed BSMs and falsified travel-time or charging-station price/availability signals, allowing researchers to emulate a broader range of malicious or unsafe interactions previously difficult to model. Expected outputs of MSU’s real-world cybersecurity testing facility include a dataset for resilience evaluation and performance benchmarks, a testbed-ready attack-emulation capability, and safety application performance assessment under cyber stress.

Outcomes/Impacts: TraCR-TMF, developed and maintained by Clemson, Benedict, SCSU, and UTD, improves threat detection by identifying 23 previously unrecognized risks in pilots, reduces assessment time by 98%, and enables proactive mitigation strategies that enhance transportation system safety, security, and reliability. The tool fosters open-source collaboration, establishes AI benchmarks for transportation cybersecurity, and supports workforce development through integration into cybersecurity-related courses and hands-on demonstrations, making advanced practices accessible to professionals and students. To support broader collaboration and impact, the FIU and UA team has made OpenCAMS available to all TraCR-affiliated research groups through a GitHub repository and will continue refining the testbed’s usability and scalability to expand research and educational use across TraCR institutions and beyond. Purdue and UCSC’s research enables the detection of unsafe interactions between autonomous systems and malicious agents, directly informing design improvements and safety assurance practices. By revealing potential attack scenarios and defenses, it supports the safer deployment of connected vehicles, influences validation standards, reduces cybersecurity risk in transportation systems, and provides actionable insights that can shape testing protocols and regulatory guidance to improve overall safety, reliability, and resilience. Project outcomes will further strengthen transportation systems by enabling safer, cyber-resilient connected vehicle deployments supported by validated cybersecurity testing and mitigation approaches. Expected outcomes and impacts of MSU’s effort include improved safety performance under cyber stress, resilience against stealthy, standards-compliant attacks, and guidance for future policy and technical development.