Mobile Agents – Where are we heading?

Peter Braun
Centre for Intelligent Agents and Multi-Agent Systems
Swinburne University of Technology

Friday, 26th November 2004
0930 - 1100

Theatre 3, level 2
University of Melbourne ICT Building
111 Barry Street, Carlton.

Abstract:

The fascinating idea of mobile agents, introduced by James White about 10 years ago, has allured many researchers all over the world. A mobile agent is a program that can migrate from host to host in a network of heterogeneous computer systems and fulfil a task specified by its owner. It works autonomously and may communicate with other agents and host systems. During the self-initiated migration, the agent carries all its code and some kind of execution state with it.

Mobile agents were a hot topic in the domain of software engineering and distributed systems. Reasons are the problems that more traditionally designed distributed systems, especially client/server systems, might have in handling excessive work-load, managing direct access for large numbers of customers, and user mobility. Mobile agent technology was supposed to help designing innovative solutions in this domain by complementing other approaches by adding the concept of code mobility, machine based intelligence, and improved network- and data-management possibilities.

However, the interest in mobile agents seems to have dwindled over the last two years. The reasons for this are manyfold. It is already accepted that there is no killer application, which can be solely implemented using mobile agents; all problems can be solved using traditional design techniques. Some expectations regarding performance benefits were not fulfilled. The advantage of "code-shipping versus data-shipping" to reduce network load is easy to prove in theory but hard to achieve in practise. Severe security problems are still open and some problems are expected to be unsolvable. Although the mobile agent community has developed more than one hundred agent toolkits, most of them are only tailored to specific research issues.

The presentation will give an overview of some of the most important problems regarding mobile agents: security, communication, and mobility. We will discuss reasons for mobile agents' poor performance compared with client/server approaches and propose some ideas how to optimise the migration process of mobile agents. We will present the current state-of-the-art technology of location-transparent communication mechanisms and give an overview of approaches to solve some important security problems. Finally, we will introduce a new plugin-oriented agent toolkit, named Tracy that is suited for research on mobile agent related topics as well for application development.

Biosketch:

Peter Braun is currently a Postdoctoral Research Fellow in the Centre for Intelligent Agents and Multi-Agent Systems of the Faculty of Information and Communication Technologies of Swinburne University of Technology. He received a M.Sc. Diploma in 1998 and a Ph.D. in 2003, both in Computer Science from Friedrich Schiller University Jena, Germany. In the last years, Peter has focused his research on migration aspects of mobile agents, in particular on design issues, mathematical network models, and optimisations of the migration process. In his thesis, Peter has proposed a new mobility model, named Kalong, including a new network transmission protocol for mobile agents. He is chief-architect of the mobile agent toolkit Tracy.