Rice CS joins UH in research to improve transportation cyber security

Silva, Hu and Jermaine focus ML, AI, and graphs on DOT infrastructure

Silva, Hu and Jermaine focus ML, AI, and graphs on DOT infrastructure

One way to disrupt a country’s economy is to interfere with its transportation infrastructure, including traffic, signals, mapping, and vehicle communications. Thanks to a U.S. Department of Transportation (DOT) grant, Rice University computer scientists are collaborating with partners at the University of Houston and several other institutions to better understand and prevent these types of cyber security threats.

“Many people are familiar with traditional attacks on centralized systems, such as the rise of ransomware,” said Rice Computer Science assistant professor Arlei Silva. “But we are also looking at distributed threats, like disrupting the flow of information as multiple autonomous vehicles communicate with each other. Then there are complex attacks and failures involving the sub-systems that interact with the major components of our transportation infrastructure.”

Identifying transportation infrastructure threats

Sub-system threats might include masking an attack on signal controls as an innocuous failure: a tree falling onto a power line causes the intersection traffic lights to go dark. For individual drivers, cyclists, and pedestrians, the results of a failure or an attack are equally dangerous. But if multiple intersections are targeted in simultaneous attacks, the ensuing traffic snarls expand to impact all the traffic that would normally pass through the area — from childcare pickup and health care appointments to supply chain deliveries and emergency responses.

Silva said, “When I arrived at Rice, my interest in transportation and infrastructure prompted me to seek research collaborators with similar interests. That is how I met Jack, who invited me to visit UH, give a talk about my research, and discuss problems at the intersection between graphs, transportation, and cybersecurity.” 

Yupeng (Jack) Zhang is an associate professor of Information Security and a member of UH’s computer information systems faculty. Zhang was already involved in an ongoing cybersecurity and transportation project with the University of Hawaiʻi at Mānoa and suggested Silva join them in seeking DOT funding for a new proposal. Guohui Zhang, a civil and environmental engineering professor at UH Mānoa, had participated in a previous DOT program and felt the new proposal was a good fit for the next round of funding.

Zhang, Zhang, and Silva invited colleagues from Embry-Riddle Aeronautical University, Texas A&M Corpus Christi, and the University of Cincinnati to complete the consortium. Together, they won approval as a Tier 1 university transportation research center (UTC). The University of Houston serves as the lead university for the Transportation CyberSecurity Center for Advanced Research and Education, or CYBER-CARE.

Incorporating data science, machine learning, and artificial intelligence

“The project is exciting for me as a researcher, but it is also important to me personally because I bike to work. I even chose my home based on the best bike paths for my commute to Rice. It is important to ride, walk, or drive with confidence that our roads, intersections, and vehicles are safe and free from malicious manipulation,” said Silva.

“Making sense of complex data streams is where my graph background is helpful, but we also needed the data science, machine learning, and artificial intelligence expertise of Chris Jermaine and Ben Hu for this project.”

Xia (Ben) Hu’s research is focused on data outcomes that are both understandable and useful for decision making. He said, “We have been working on interpretable machine learning of important applications in healthcare and social sciences. Through those projects we have developed a systematic framework in understanding how AI systems work and a decision is made. This provides a solid foundation for us to carry out the proposed research in transportation cyber security.”

Hu is the director of Rice’s Data to Knowledge (D2K) Lab as well as an associate professor of Computer Science whose research has been cited more than 10,000 times. He and his team were intrigued by the opportunity to apply their work in the sphere of transportation cyber security, and he also sees an opportunity for D2K students to get involved. 

He said, “We are excited about working with domain experts in developing interdisciplinary solutions for this important application, and we expect to engage with the established D2K Capstone program to integrate the research outcome of this project into data science education.”

Silva believes Hu’s contributions in scalable machine learning will provide better user interactions with the model, and the resulting data will be more accessible and more intuitive to interpret. Jermaine’s work in machine learning and depth incursion will improve and expand the model’s solution development.

Increasing security for transportation infrastructure

The solutions Silva refers to can be used to forecast and suggest preventative measures for cyber security vulnerabilities in the transportation infrastructure. For example, the increase in autonomous cars will include more vehicle-to-vehicle communications to prevent accidents or share weather or traffic updates. How could this kind of communication be abused, and how could that abuse be prevented?

“We recognize that interactive systems can be targeted for weaponized systems attacks,” said Silva. “There are also non-malicious hacks that reveal vulnerabilities in our infrastructure. Some smart people have already learned how to manipulate traffic light systems in order to drive across town through all green lights. This isn’t just a movie scene, this is actually being done – sometimes for self-gratification purposes. But if it can be done for personal pride, it can obviously be targeted for a more serious attack.

“Or, if a mapping system attempts to route public traffic through a private road, there are processes in place to make corrections to the mapping system. Our model might identify options or processes like these that could be misused to cause delays, damage, or more serious consequences.”

Leveraging Rice University’s local impact 

Silva’s long-term interest in transportation and infrastructure pushed Rice to the top of his faculty offers. With organizations like the Kinder Institute for Urban Research and the recent research focus on Cities of the Future in the George R. Brown School of Engineering, Rice provides space for Silva to direct his research into the areas that matter most to him.

“I was excited to see how well Rice is already regarded in these areas. The CYBER-CARE work is a great way for us to interact locally and at the state and federal levels, thanks to all the roads and bridges in need of routine maintenance as well as improving the safety of how they are used,” said Silva. In addition to the space Rice provided to develop his research group in areas that support urban planning, the university has also supported his growth as a faculty member and a principal investigator. 

Silva said, “This was my first opportunity to serve as an institutional PI, and one of the lessons I have learned is how large research centers like this require teamwork across campus. I am very grateful for the support I received from the Rice Office of Research, the School of Engineering, and the CS Department. 

“The assistance I received from Jon Meyer, through the Office of Proposal Development, for months last summer really made this project possible. Ricardo Mottu, our Department's Proposal Specialist, ensured we got all the documents approved in time. Now I'm very excited to work with Ben, Chris, and the rest of the team to address these pressing challenges through the innovative research to be done here at Rice."



Carlyn Chatfield, contributing writer