Content Modification Attacks on Networked Robotic Systems

Thumbnail Image
Publication or External Link
Dong, Yimeng
Chopra, Nikhil
With the advent of communication networks in robotic systems, distributed networked robotic systems can be deployed to perform certain tasks collaboratively. However, this makes the networked robotic systems vulnerable to cyber attacks. Thus, the rigorous study of the impact of cyber attacks and the development of corresponding defense mechanisms are necessary. In this dissertation, the cyber-physical security issue of networked robotic systems is studied under a specific type of cyber attack called content modification attack, which can modify the data content transmitted in the communication networks among the robots. Specifically, algorithms for attack design and detection for content modification attacks are studied. The physics of the robotic system is utilized to design and detect the cyber attacks for networked robotic systems. Content modification attacks are studied for the synchronization problem in networked robotic systems. The considered systems include multi-robot systems, bilateral teleoperation systems and bilateral tele-driving systems. To demonstrate the potential severity of the attack, a constructive methodology for attack design is also developed. Specifically, a destabilizing content modification attack referred to as a malignant content modification attack (MCoMA) is designed based on the system storage function, which can lead to system instability and even physical system damage. To protect the system, a physics-based attack detection scheme with an encoding-decoding structure is proposed for general content modification attacks. As part of the tele-driving system study, a novel passivity-based adaptive bilateral tele-driving control scheme is also proposed in the presence of network delays and dynamics parametric uncertainties. Simulations and experiments have also been conducted to validate the proposed algorithms. This study demonstrates the potential of utilizing the physics of the robotic system to better understand and strengthen the security of the networked robotic systems.