Provisioning of secure Multi-Agent Systems (MAS) based on trust contracts

Abstract

This research work focuses on the development of an Interaction Security Contract (ISC) mechanism that enhances the capability of agents to protect themselves against compromised entities in Multi-Agent system. To realize the framework, two tasks were carried out namely ® a simulation of an agent based system was created and trust accuracy levels based on the proposed mechanism were evaluated to establish the significance of the proposed security scheme; and (if) a security-aware agent-oriented shopping mall system was designed and implemented to demonstrate the proposed contract based security model. The results obtained are threefold; (i) This study was able to establish that trust models based on reputation do not obtain accurate trust values for agents to make correct trust decisions in agent systems; (ii) it was established that it was even riskier to form collaborations without any trust model in place as agents collaborated with agents whose security status was not known to them. This clearly shows that agents were at risk of being compromised; and (iii) it was established that if agents publicize their security properties truthfully, trust in the overall agent systems will be gready improved as agents themselves will only be collaborating with agents who meet their specified level of trust In conclusion the study advocates the employment of the proposed Interaction Security Contract (ISC) mechanism, since it was demonstrated how knowledgeable trust relationships can be formed and improved in agent based systems.