Secure auction for mobile agents

In this work, a secure auction place for mobile agents has been developed and implemented. In the implemented auction application, software agents are able to bid on different products independently without user intervention in a secure manner. For implementation, quality requirements for mobile agent systems were studied. By defining the possible threats that a mobile agent system may face, a set of protection mechanisms were selected to build a security architecture that was used in protecting the agent system. The agent platform and the agents were designed based on the security architecture and the implementation was carried out using Java. Finally, in order to provide proof about the robustness of the implemented system, it was tested using a mini-simulation method. As a result of the implementation, the communication security for transferring a mobile agent between nodes could be provided effectively by using traditional security mechanisms. However, the distribution of keys and certificates is required to be done manually so that the counterparts of the auction events can trust each other, which on the other hand, causes scalability problems. As a difference to most other agent platforms, the type of agent mobility was more restricted and code mobility was not used. For this reason, the security of the agent platform could be provided better. However, restrictions that had to be made to the system were quite big resulting in a decrease in the agent system's ability to adapt dynamically. In summary, there are still a few problems that have to be overcome before the time is ready for large-scale mobile agent auctions. In particular, the security of the mobile agent residing on a remote platform has to be guaranteed without the assumption of trusted platforms.