Technologies for the Development of Agent-based Distributed Applications[6]

Interoperability

The most important attempt to standardize agent system interactions so far is Mobile Agent System Interoperability Facilities (MASIF) by the Object Management Group [12]. It specifies actions like agent and class/object transfer as well as agent management and location services. Unfortunately, almost none of the available agent systems currently are MASIF-compliant, although for the predominant systems, their developers have announced MASIF-conformance for the future. The standard's scope is remarkably limited. For example, it does not address issues like cross-platform migration of agents.

Conclusions

With only basic technologies available for migration, execution, and security issues (with the exception of the problem of malicious hosts, which seems intractable if only software solutions are attempted), much work remains to be done to support agent system interoperability. Additionally, as for the practical use of mobile agent technology as a means for distributed applications development, the question of availability and applicability of design methods arises. Despite being such a new field, agent-based development of applications has already seen quite a number of proposals for programming or design methods, for example Gaia, MaSE, and AOR [14]. One restriction that all of the mentioned methods share is the assumption that agent abilities (functions) are static at runtime. Therefore, they are not suitable for the development of systems whose interactions should be knowledge-based. While there are several analysis and design methods available, one crucial question that is waiting to be answered is performance measurement [5]. Without good models to compare conventional and mobile agent designs in terms of efficiency, the entire field will remain of academic interest only.

Outlook

Mobile agent computing is still a young and immature technology. Several issues have not been resolved satisfactorily to date, among them the problem of malicious hosts, systems interoperability, and the question of performance measurement. Yet the technology has the potential to facilitate improvements in distributed systems development.

One of the strengths of mobile agent computing is the possibility of moving complex processing functions to the location where huge amounts of data are to be processed. While on the Internet, it is unlikely that an infrastructure facilitating the widespread use of mobile agent computing will be in place in the near future, mobile agent could become a key technology for knowledge-management solutions in large corporations' intranets. Similarly to the situation on the Internet, there are huge amounts of information in numerous servers, stored in files which cannot simply be scanned like plain text, that nobody can really make use of because they cannot find it. A distributed information retrieval/knowledge management solution built with mobile agent technology could help improve this situation.

Another area where agent technology could play a central role is 3G cellular networking. Consider, for example, the so-called Virtual Home Environment (VHE), which is part of the Universal Mobile Telecommunications System (UMTS) specification. VHE is intended to serve the purpose of universal access to personalized services (like e-mail, banking, etc.) from any network. This is exactly the heterogeneous, unreliable computing environment that mobile agent computing is made for.

There are numerous similar scenarios for the future, for example in the context of ongoing rapid growth of the Internet. However, it seems that the "solution in search of a problem," as some scientists call mobile agents, could, from today's perspective, indeed eventually find several problems to solve.

References

1

Baumann, J., et al. Mole—concepts of a mobile agent system. In Mobility: Processes, Computers, and Agents, edited by D. Milojicic, F. Douglis, and R. Wheeler. Stuttgart, Germany: Universit?t Stuttgart, Fakult?t Informatik, 1999.

2

Braun, P. über die Migration bei mobilen Agenten. In Jenaer Schriften zur Mathematik und Informatik Nr. 99/13, Jena, Germany: Friedrich-Schiller-Universit?t Jena, Institut f黵 Informatik, 1999.

3

Broos, R., et al. Mobile Agent Platform Assessment Report. Contribution to the EU Advanced Communications Technology and Services (ACTS) Programme, 2000. < http://www.fokus.gmd.de/research/cc/ecco/climate/ap-documents/miami-agplatf.pdf> (29 October 2001).

4

Chen, H. Developing a Distributed Dynamic Intelligent Agent Framework Based on the JINI Architecture. PowerPoint presentation, < http://gentoo.cs.umbc.edu/ronin/doc/msthesis.ppt> (28 October 2001).

[1]  [2]  [3]  [4]  [5]  [6]  [7]