Principal Investigator: Inseok Hwang
Over the past decade, advancements in sensor, computation, communication, and battery technologies have led to the growth of Unmanned Aerial System (UAS) operations in the defense and commercial sectors. In the near future, UAS operations in urban areas are expected to increase in their scope, level of autonomy and technological capabilities. Through the introduction of new technologies in the UAS domain, such as cloud computation, ad-hoc mesh networking and self-navigation capabilities, new vectors for cyberattacks are introduced as well. It is thus of imminent importance to secure these technologies against adversaries seeking to disrupt UAS operations or the infrastructures that rely on them.
In urban environments, the proximity of UASs to human resources means that any collisions or failure in their operation can have adverse consequences. Moreover, the various obstacles in urban environments pose a unique challenge for UAS cybersecurity, as they can interfere with the sensing capabilities of the UAS (for e.g., due to obstruction or reflection of GPS signals by the buildings). These vulnerabilities can be exploited by attackers to design stealthy cyberattacks, which can remain undetected over a period of time while maximizing their impact. Sophisticated cyberattackers may even coordinate their attack across multiple UASs in a mesh network, taking advantage of the complex UAS-UAS interactions to achieve their objectives. In addition to preventing such attacks through better encryption and authentication schemes, it is important to have failsafe mechanisms in place which can assure the safety of UAS operations, as well as that of the surrounding human life and property.
We address these distinctive issues by combining a wide range of mathematical tools to identify, analyze and patch the various security vulnerabilities in current UASs. The cybersecurity of existing UAS autopilot firmware is ensured using automated vulnerability-discovery tools, which can find and fix various logic bugs in the code that could be exploited by adversaries. To facilitate real-time cyberattack mitigation during operation, the firmware is updated with novel algorithms which enhance the robustness and resilience of UAS navigation and control. We also develop offboard cyberattack-detection algorithms which can be used to exploit external redundancies (such as camera-based surveillance and UAS traffic management systems) to detect the presence of adversarial agents in UAS operations. Furthermore, a system-of-systems perspective is used to discover the emergent vulnerabilities in large-scale UAS operations, which may otherwise remain unidentified due to their complex and unpredictable nature.
To test and validate the developed algorithms on real UASs, it is necessary to recreate these vulnerabilities and cyberattack-mitigation mechanisms in a controlled test environment which closely resembles the real-world setting. To this end, a mixed-reality experimental testbed is being developed at the Purdue UAS Research and Test Facility (PURT), which is equipped with the largest indoor motion capture system in the world and has capabilities such as realistic emulation of GPS signal characteristics in urban environments. The goal of the project is to implement the developed cyberattack-mitigation algorithms in conjunction with emulation of various cyberattacks in the testbed, such that the security of UAS firmware and physical components can be experimentally assured before their deployment in urban environments.
Other PIs: Dongyan Xu, James Goppert
Students: Shiraz Khan, Dawei Sun, Kartik Anand Pant, Zhanpeng Yang, Hyungsub Kim, Jefferson Kim and MinHyun Cho
Khan, Shiraz, Inseok Hwang, and James Goppert. "Robust State Estimation in the Presence of Stealthy Cyberattacks." 2022 American Control Conference (ACC). IEEE, 2022.
Keywords: Cyber-physical system, Cybersecurity, Unmanned Aerial System