Optimal Policy Of Voting System In Malicious Attack Environment

System operation is not only affected by internal faults,but also by external malicious attacks,which directly affects the reliability of the system.The voting system has high fault tolerance and strong decision-making power,and it has been widely used in many fields.This paper mainly studies the reliability of voting system that have suffered multiple malicious attacks and gives the corresponding optimal strategies.The reliability here refers to the probability that the system can make correct decisions for the input propositions.This paper discusses the reliability of a majority voting system composed of various components(normal components,upgraded components and camouflaged components)and focuses on the optimal strategy for the quantitative combination of different types of components.Assuming that the resource value of the defender and the attacker is given,the attacker randomly selects k components from all components to attack each time.After being attacked many times,the state of each component will become uncertain.This paper gives the numerical theory derivation process for calculating the reliability of the system when k = 1 and provides the corresponding algorithm(algorithm 1).Under the assumption that all attacks are initiated by Poisson processes and the time of random proposition input obeys a truncated normal distribution,for the case of k ? 1,this paper provides an algorithm(algorithm 2)to simulate the expected reliability of the system.Algorithm 2 is suitable for the calculation of system reliability given the reliability function of each type of component.This paper provides corresponding optimal strategies for attackers and defenders,respectively.Under certain conditions,our research shows that:(1)The worst strategy for the attacker to choose is to attack all components each time,and it is better for the attacker to randomly choose k = 1 or k = 2 components to attack;(2)For the defender,the more the number of upgraded components,the greater the system reliability.When the number of upgraded components is certain,appropriately increasing the number of camouflaged components can also improve the system reliability.In addition,in the face of different attack strategies of the attacker(choose different k),this paper use the genetic algorithm(algorithm 3)to provide a solution for the defender to find the optimal combination of components.Experiments show that this method has good performance accuracy and stability.