| The notion of mobility has always been a prime factor in human endeavor and achievement. This need to migrate by humans has been distilled into software entities, which are their representatives on distant environments. Software agents are developed to act on behalf of a user. Mobile agents were born from the understanding that many times it was much more useful to move the code (program) to where the resources are located, instead of connecting remotely.;Within the mobile agent research community, security has traditionally been the most defining issue facing the community and preventing the paradigm from gaining wide acceptance. There are still numerous difficult problems being addressed with very few practical solutions, such as the malicious host and agent problems. These problems are some of the most active areas of research within the mobile agent community.;The major principles, facets, fundamental concepts, techniques and architectures of the field are well understood within the community. This is evident by the many mobile agent systems developed in the last decade that share common core components such as agent management, communication facilities, and mobility services. In other words new mobile agent systems and frameworks do not provide any new insights into agent system architecture or mobility services, agent coordination, communication that could be useful to the agent research community, although these new mobile agent systems do in many instances validate, refine, demonstrate the reuse of many previously proposed and discussed mobile agent research elements. Since mobile agent research for the last decade has been defined by security and related issues, our research into security patterns are within this narrow arena of mobile agent research. The research presented in this thesis examines the issue of mobile agent security from the standpoint of security pattern documented from the universe of mobile agent systems. In addition, we explore how these documented security patterns can be quantitatively compared based on a unique weighting scheme. The scheme is formalized into a theory that can be used improve the development of secure mobile agents and agent-based systems. |
