The Mobile Agent Technology

Within the last five years the paradigm of Intelligent and Mobile Agents has gained momentum in the field of software engineering. Mobile Agents (MAs) are autonomous, asynchronous and intelligent software entities which, in order to fulfil their tasks, can migrate to and reside in a number of networked nodes. A variety of agent languages, architectures and platforms have been developed within the last three years in academia and industry, including AgentTCL, Telescript, Tabriz, Active X, Java and a variety of Java enhancements, such as Aglets, Java-To-Go, etc. However, all of these developments are incompatible with each other and designed for specific environments and application domains, due to the lack of agent standards. The application of the agent technology, however, has already a history and spreads over many areas, including user interface/personal assistance, mobile computing, information retrieval & filtering, data mining, smart massaging, the electronic marketplace, and telecommunication services control and service / network management. The last two areas represent still new application fields for agent technology, now rapidly gaining importance in the problem area addressed by the emerging global Information Infrastructure. The heterogeneity, distribution, scale and dynamic nature of the emerging Information Infrastructure, referred to as NII, EII, or GII, is calling for new paradigms for the control and management of the open resources. In this context, the agent-based technology seems to offer a promising solution to cope with the complexity of this environment. Within such an open environment, an agent-based solution can: - reduce the requirement of traffic load and the availability on the underlying networks (via the autonomy and asynchronous operations of the agents); - reduce the requirement on customer intelligence during the installation, operation and maintenance of the resources (via the intelligence and autonomy of the agents); - enable on demand provision of customized services (via dymanic agent downloading from the provider system to the customer system and further on back to the provider system or directly to the resources); - increase the flexibility, reusability and effectiveness of the software-based problem solutions; - allows for a more decentralized realisation of service control and management software, by means of bringing the control or management agent as close as possible or even onto the resources. Agent platforms could be considered as enhancements of todays distributed object-oriented platforms, e.g. DPEs. Hence the long term target should be to head for an appropriate enhancement of standard object-oriented service and management platforms, such as CORBA and the TINA DPE in order to allow for more flexible service realisations. However, in face of existing legacy environments, such as IN, TMN and Internet based platforms, the short to medium term integration of agent technology into these systems could enhance the flexibility of these systems to a large extend and thus offer new market opportunities. Key to the success for the integration of agent technology in particular environments is the adoption of standardised solutions. The standardization of agent technologies has started in 1996 within three fora, namely the Object Management Group (OMG), the Foundation of Physical Intelligent Agents (FIPA), and the Agent Society.