Using genetic algorithm to improve the performance of multi-host vulnerability checkers

This work is based on the reconfiguration of the NetKuang to improve its performance. It operates on computer networks employing the UNIX OS. It detects vulnerabilities in poor system configurations. So this work is not only capable of searching a large number of hosts in parallel, but it also considers potential configuration vulnerabilities present in the network. The main disadvantage of NetKuang is that it can only develop one vulnerability at a time on a given system. Furthermore, there is a leak in memory when running a task. Our work aims at developing more than one vulnerability at a time. Vulnerabilities are discovered using a backward goal-based technique. That is, inducing a different search technique based on a genetic algorithm to discover the vulnerabilities. We aim at using genetic algorithms to point out several vulnerabilities simultaneously. Moreover, our technique overcomes most of the previously mentioned disadvantages produced by the standard technique. Considering the genetic algorithm, there are two different ways that the search would apply: the simple genetic algorithms, and the classifier genetic algorithms. The one chosen in this paper is the classifier genetic algorithm, that would produce the best result. The time and space complexities are computed in order to compare the proposed technique with the standard one for the purpose of getting the best results.