Sampled-Data Flocking of Multi-Agent Systems Under the Cyber-Attack Problem
Subject Areas : Intelligent Multimedia Processing and Communication Systems (IMPCS)
سحر
یزدانی
1
()
Keywords: Flocking, sampled-data, cyber-attack problem, multi-agent systems,
Abstract :
Introduction: Flocking is a type of collective behavior which is observed in the nature. In the design of a flocking algorithm, it should be ensured connectivity of agents’ network and the collision avoidance, and velocities convergence of agents to that of virtual leader. In practice due to the limitations in the measurement and control units, it is often impossible to ensure the continuity of information. Thus, the study of the flocking problem under the sampled data frameworks is indispensable. However, to the best of the authors’ knowledge, there are very few works on the sampled-data flocking. On the other hand, in many practical applications, the multi-agent systems are controlled through some communication networks. The transmitted data among agents could be easily exploited by adversaries due to the open network links among sensors, controllers and actuators. Since in practice often the attacks are capable to destroy a number of edges within the network or cause to collide among agents, the study of networked system under the cyber-attacks is very important. In the cyber-attacks, successful but recoverable attacks have attracted more attention. Successful attacks refer to a class of attacks by which the network is broken dow n into a group of isolated clusters. Recoverable attacks refer to a class of attacks that the network can recover from after a period of time. In this paper, we study the sampled-data flocking of multi-agent systems under the successful but recoverable network attacks. Method: Here, defining a new discrete-time energy function we prove the asymptotic velocity convergence of agents to the velocity of virtual leader. Then, through the upper bound of the energy function, we find an upper bound for the sampling period such that the connectivity of network is preserved and collision is avoided, and also, the velocity convergence is ensured. After that, we modify the algorithm for application in cyber-attacks. Results: We show that under our proposed sampled-data algorithm, no link is lost from initial network, no collision is occurred among agents, and the velocity convergence of agents to that of virtual leader is ensured. Also, demonstrate the proposed algorithm is applicable for the flocking under the attack problem.