An Intrusion Detection Approach Using An Adaptative Parameter-Free Algorithm
Mourad Daoudi, Mohamed Ahmed‐Nacer · 2014
Abstract—In intrusion detection from audit security, the volume of data generated by the auditing mechanisms of current systems is very large. It is important to provide security officers with methods and tools to determine predefined attack scenarios in the audit trails. The problem is Non-deterministic Polynomial-time hard (NP-hard). Metaheuristics offer an alternative to solve this type of problems. Unfortunately, many parameters have to be tuned for any metaheuristic, and their values may have a great influence on the efficiency and effectiveness of the search of good solutions. The exploration of an optimal combination of such values may be difficult and big time consuming. Clerc et al. have defined an adaptative parameter-free algorithm, called TRIBES, issued from Particle Swarm Optimization. It is developed to solve continuous problems. In this paper, we propose to adapt TRIBES to solve our combinatorial optimization intrusion detection problem. Modifications in different mechanisms and formulae adaptations in original TRIBES are made, like in the generation process of the particles and in the displacement strategies. The experimentations results show the good behavior of our approach. Comparisons with a basic genetic algorithm are provided.