![]() |
Abstract for Nelson Minar
|
Mobile Agent Research
My main research interest is in distributed computing systems with no central controller or designer. I would like to do for distributed computation what the Web has done for distributed data. In my work I apply ideas from complex systems to computer processes. The phenomenon of self-organization in a variety of natural systems points to a new way to design computational systems. Metaphors from biology (artificial life) and economics (agoric computation) are particularly useful.
Mobile agents are a natural, powerful paradigm for distributing computation across networks. Frustratingly, mobile agents have not yet really been proven effective in many practical applications. At the Dartmouth workshop I intend to work with other people to understand better what mobile agents really are good for, what application domains they are most effective in.
In my paper "Computational Media for Mobile Agents" (December 1996) [3], I talk about one application - using mobile agents to facilitate the use of idle CPU time on the Internet. In this paper I also establish a framework for evaluating mobile agent systems and survey existing implementations for their practicality. My research now is applying these ideas to building a real system for distributed resource coordination.
Practical research topics include the design of languages and libraries to make mobility natural, security (protecting mobile agent servers and the mobile agents themselves), resource-use accounting, and description of resources and services.
1: http://agents.www.media.mit.edu/groups/agents/ MIT Media Lab Software Agents Group 2: http://www.santafe.edu/projects/swarm/ Swarm Simulation System 3: http://nelson.www.media.mit.edu/people/nelson/research/dc/ Computational Media for Mobile Agents