Increasing the Resilience of Transportation Systems under a Combination of Cybersecurity Attacks and Extreme Events

Investigator(s):

• Arlei Silva, Rice University, ORCID # 0000-0003-1792-0076 (PI)

Project Description:

This project is focused on measuring the resilience of transportation systems with respect to cyberattacks and extreme events (hurricanes and power outages). This will require developing an ATMS simulator for a given road network system to simulate potential cyberattacks and their impact on the traffic. We will propose combinatorial optimization algorithms for optimally attacking the ATMS and measure the impact of such attacks to assess the resilience of the system. We will also evaluate the impact of concurrent extreme events on the transportation system, especially hurricanes and power outages. These extreme events are expected to become more likely in the upcoming years due to climate change and are particularly relevant to the city of Houston, TX, where the PI’s institution is located. The proposed approaches will be evaluated on publicly available datasets in collaboration with other members of the center. Main findings will be summarized in at least one research paper and the final project report. Software, datasets, and metadata produced through the project will be made publicly available.

We summarize the major objectives of this project as follows :

1. Design a realistic simulator for an Advanced Traffic Management System (ATMS) based on multiple sub-systems (e.g., sensors, connected vehicles);
2. Develop combinatorial and machine learning-based optimal attacks to the ATMS;
3. Evaluate the impact of attacks on the ATMS in combination with extreme events (power outages and hurricanes).

Figure 1 Resilience of transportation systems under a combination of cybersecurity Attacks and extreme events. The figure shows road closures due to flooding and malfunctioning traffic sensors due to a cybersecurity attack.

This project will engage in transformative research by developing new methods for evaluating the resilience of transportation systems. The team will first develop a realistic simulator of a small-scale advanced traffic management system. More specifically, the simulator will combine traffic sensor data, weather data, connected vehicles, and other ATMS capabilities (e.g., smart traffic signals). The simulator will be fine-tuned using real transportation data and will support testing coordinated attacks on multiple sub-systems (e.g., traffic sensors and connected vehicles). Next, we will develop algorithms to discover optimal attacks using both combinatorial optimization (e.g., greedy algorithms, game-theoretical approaches) and also machine learning-based approaches (including Reinforcement Learning). These algorithms will maximize different metrics such fraction of the overall system disrupted and the number of users impacted. We will also consider multiple levels of knowledge to be exploited by the algorithm during the attacks, including full knowledge of the system. From a more theoretical perspective, we will investigate the hardness of discovering such attacks under multiple assumptions (i.e., the limitations of the attacker). Finally, we will apply these attacks to the simulator to identify critical components of the system. A component is critical if its attack maximizes the impact over the ATMS. Finally, we will study the extent to which protecting such critical systems reduces the effectiveness of future attacks on the ATMS.