Learning-based Intelligent Attack against Formation Control with Obstacle-avoidance
Yushan Li, Jianping He, Cailian Chen, Xinping Guan · 2019
Formation control has attracted considerable attention for its wide applications, e.g, military reconnaissance, environment exploration. However, the formation suffers additional security vulnerabilities due to its distributed fashion, networked communication and openness to outside environments. Existing works focus on detection and countermeasures for some classic attacks, e.g., Denial of Service (DoS), replay and deception attacks. Nevertheless, those attacks are generally from cyberspace and the methods are based on an assumption that the attacker has some knowledge or access to the formation system, like the system dynamics is known or internal nodes are compromised. It remains an open issue given how to design a feasible attack or under what conditions an attack can be implemented. In this paper, we aim to design a feasible and intelligent attack scheme against the obstacle-avoidance of formation control. We describe it as “intelligent” for the following: i) Without any prior information of the system dynamics, the attacker can learn the detection area and goal position of an agent by trial and observation; ii) The obstacle-avoidance mechanism is regressed using support vector regress (SVR) method; iii) The strategy exhibits attack efficiency. Furthermore, a sufficient condition is obtained to guarantee the success of the intelligent attack. Simulations illustrate the effectiveness of the proposed attack scheme.