IIDS

Interactive Intelligent Distributed Systems

The IIDS group's research focuses on management of large-scale interactive distributed systems.

Management of large-scale distributed "open" systems is a challenge, especially if autonomous processes are part of the model. In closed systems, with their own sets of rules and obligations, some form of central control may be possible. In open systems this is not the case. Trust and security clearly play an important role inhibiting centralized management.

Local management becomes increasingly more important. Self-management is the next step: self-configuration, self-healing, self-optimization, and self-protection of autonomous systems. Together they make it possible for large-scale systems to fend for themselves. As good understanding of configuration and reconfiguration processes is mandatory for the first two aspects.

Agents are examples of processes that may roam the Internet. They are autonomous, can communicate with each other and interact with services and objects. As a result they may, for example, negotiate contracts, perform transactions, but also decide to spawn off new agents to perform specific tasks. This clearly does not decrease the complexity of managing large-scale distributed open systems on the Internet: it emphasises the need for new technology. New types of directory services, for example, are needed. Which technology should be used to design these services, is, however, unclear.

The research programme of IIDS focuses on the implications of the use of mobile processes, of agents in particular: the technology needed to develop and manage large scale agent systems. Scalability, reliability, security, and heterogeneity of such systems are aspects that need to be considered if the Internet is ever to be used to its full potential. Middleware, services and applications are the main parts of this programme. Middleware, together with services, provides the support. Self-management of the middleware is a next step in this line of research. Applications provide insight in requirements and provide the means with which to evaluate the results.

The focus of the research continues to be (1) on the middleware AgentScape Operating System (AOS) and the related services, in particular configuration services, and (2) on the legal implications of the use of agent systems. The Java version of the AgentScape middleware will provide a more stable basis for application development and support.