On The Reliability Of A Voting System Under Malicious Attacks

Voting systems have been widely used in many areas due to their technical decision making abilities.However,when a voting system operates on any internet platform,it faces a number of potential malicious attacks including malware and denial of service attacks.In this work,we study the reliability of a majority voting system under malicious threats.Three multiple attack scenarios are considered including targeted,random,and dynamic attacks.We show that under some conditions,the reliability of a voting system decreases with the number of attacked voting components for the targeted attacks;however,for the random attacks,the reliability of a voting system can increase with the number of attacked voting components.For the dynamic attacks,it is discovered that the reliability of a voting system is significantly impacted by the attack rate and the random proposition time.We further develop efficient algorithms for computing the reliability of voting systems.All the above works consider the scenario that the attacker distributes attack resources equally against the target components.In this thesis,we further consider the scenario that the attacker allocates attack resources unequally against the target components.In practice,the attacker may spend some resources to exploit the common vulnerability among the targeted components.If the vulnerability is discovered,the attacker may adopt the unequal split strategy to launch attacks,i.e.,the attacker strategically distributes attack resources against target components.Similarly,we develop an effient algorithm for computing the reliability of the voting system in this case.Experiments suggests that in the targeted attack when a common vulnerability can be exploited,for the attacker,it is better to choose more voting components as targets.