Research On Distributed Security Mechanism Of Multi-agent System

Consensus problem is one of the key issues in multi-agent systems.Which means that each agent in the system updates the status value by the information of surrounding nodes,so that the certain state value of all agents in the system can be reached a consensus within a limited time.However,since multi-agent systems may be deployed in open or even harsh environments,a sin-gle agent is vulnerable to malicious interference and attacks.When an agent node is attacked,it affects nearby nodes through the network and then affects the entire system.The network topol-ogy required for secure consensus algorithms in many studies has been too harsh or specific to a particular attack.Based on the research of previous work,this paper further explores the secure consensus algorithm in multi-agent system based on the shortcomings of existing algorithms in applicability and convergence speed.Firstly,a new secure consensus scheme is proposed for the problem that the network topol-ogy required by the existing average-subsequence-reduce consensus algorithm is too complicated.In this scheme,a specific mobile node is employed as a motion detector and a network robustness builder for identifying regions in the system where the state values differ greatly and acting as a communication link between them.The simulation results show that the consistency algorithm running under f-local malicious behavior can adapt to networks below 2f+1-robustness.In addition,an resilient maximum consensus algorithm is proposed for the weak consensus problem in multi-agent systems.Each node in the system selects the maximum state value con-sidered by the node from the collected information through the set rules to perform the update iteration.The theoretical derivation proves that the algorithm needs to meet the necessary and sufficient conditions for the network topology to meet the elastic consistency.That is,the net-work still needs to satisfy 2f+1-robustness to resist f-local attacks.The experimental simulation results show that compared with the traditional average-subsequence-reduce consensus algoritm,this algorithm has a faster convergence speed under malicious attacks.Finally,the above research work is extended to the application level,and an resilient max-imum clock synchronization algorithm is proposed.Each node in the system receives its own information and uses it to calculate the relative logical deviation and store it.Then,when updat-ing,it refers to the relative logical deviation parameter to select relatively safe storage information for updating.The theoretical derivation in this paper proves that the synchronization algorithm achieves convergence within a finite time under the influence of malicious nodes.Simulation ex-periments show that the system quickly reaches a consistent situation in a small number of time steps after running the algorithm.