Investigations on Computational Intelligence Based Techniques for Efficient Security Threat Controls

Abstract

Computer security is defined as the protection of computingsystems against threats to confidentiality, integrity and availability. Thethreats to security in computing are interception, interruption, modificationand fabrication. The controls available to address the threats to security incomputing include encryption, network controls, programming controls andoperating system controls.The traditional methods of controls to threats in security which isthe center of focus in this thesis are beset with drawbacks justifying acompelling need to look for better / efficient solutions. The work in the thesiscenters around investigating computational intelligence based approachessuch as Genetic Algorithms (GA) and the two paradigms of swarmintelligence viz Ant Colony Optimization (ACO) and Particle SwarmOptimization (PSO) to overcome the drawbacks existing in the traditionalmethods of controls available to combat threats in security in computing. Thecomputational intelligence based approaches are applied to issues in securityin computing such as Group rekeying for secure multicast, Leakeridentification in secure multicast, Text and image encryption, Messageauthentication and Packet filtering firewall.Group rekeying for secure multicast is achieved using an ACO/GAbased Boolean Function Minimization Technique (BFMT) when the usersjoin in or leave a group. Some of the group rekeying methods proposed in the ivliterature require a large number of keys to be stored and distributed to theusers in the group. Of these, the group rekeying methods employing BFMTapproaches stores and distributes minimum number of keys. However, thegroup rekeying methods employing BFMT approaches are not withoutdrawbacks too. Therefore, to overcome these drawbacks, the computationalintelligence based approaches viz, Genetic Algorithm and Ant ColonyOptimization are proposed to obtain a minimized boolean expression forgroup rekeying in secure multicast. The genetic algorithm based approachtermed Boolean Expression Evolver (BXE) is proposed to obtain a minimizedboolean expression. However the drawback of the genetic algorithm approachis the increase in the time taken to obtain a minimized boolean expression,since all the chromosomes in each generation need to be considered. Toovercome this drawback, an Ant Colony Optimization (ACO) based approachtermed Ant Colony Optimization Boolean Expression Evolver (ABXE) isproposed to obtain a minimized boolean expression. The minimized booleanexpression represents the minimum number of messages required to distributeminimum number of keys to the users in the secure multicast group.Leaker identification in secure multicast applications such ascommercial pay-per-view video multicast and pay-per view digital libraryshould prevent users from leaking any information. Termed Ant ColonyOptimization Leaker Identification Algorithm (ACOLIA), the novel techniqueserves to efficiently identify the leaker. The significant improvement inACOLIA is that the number of comparisons is less when compared to theexisting sequential search method for leaker identification.