NSF Award
1715409
Heavy vehicles (e.g., trucks and busses) are a critical element of U.S. and worldwide logistics, often carrying cargo of high value or high risk (e.g., explosive liquids and gasses). Heavy vehicles often have hundreds of Electronic Control Units (ECUs) that communicate over an internal network to carry commands (such as "engage the brakes") or share sensor data (such as the temperature of pressurized cargo unit carrying petroleum). ECUs with access to the communication network can send any message they want. If the network or an ECU is compromised by an attack, the truck or a cargo container safety mechanism could malfunction. This project is gathering data from operational trucks to better understand communication among components of heavy vehicles and developing techniques to detect attacks in this environment.<br/><br/>The project is working to accomplish three main objectives: (1) Collect representative Controller Area Network (CAN) bus data from operational heavy vehicles, (2) Develop detection systems that can distinguish anomalous CAN bus network traffic, and (3) Test and verify the detection systems to reduce the number of false positives. The team is developing a log algebra to efficiently assess live CAN traffic using embedded devices with limited resources. Data is being gathered from truck traffic during highway operation, enabling the application of machine learning algorithms for anomaly detection. The team is evaluating the effectiveness of their intrusion detection techniques in their heavy vehicle testbed, using synthetic attacks against testbed ECUs and real-world CAN traffic data.
