Near Optimal Defense Resource Allocation and Information Dividing-and-Allocation Strategies to Minimize Information Leakage Considering both Random Errors and Malicious Attacks

Information technology has been increasingly progressing, and the storage cost has been reducing. Thus, individuals, enterprises and government organizations are likely to store secret data through electronic way. Moreover, along with the rise of the use of network and the prevalence of e-commerce, the crime of information theft through network has grown in high-speed. Cyber crimes, like phishing or installing Trojan horse in victims’ computers to steal information, will cause serious damage to individuals or enterprises. From the above reasons, to protect secret information stored on networks becomes an essential issue.In this thesis, we formulate the attack-defense scenario as a min-max mathematical programming problem, which is a two-level mathematical problem. In the inner problem, the attacker wants to maximize the total damage by stealing information under limited attack power. In the outer problem, the defender wants to minimize the total damage by defense resource allocation and information-dividing under limited budget. In addition, the defender also has to consider QoS requirements of authorized users. In order to solve the considered problem, we use the Lagrangean Relaxation method and the subgradient method [14][15]. We solve the inner problem under a given defense strategy first, and then propose a subgrandient-based heuristic to adjust the defender’s strategy according to attacker’s attack strategy.