Adaptive Agents and Selforganisation

Today's software systems are constantly subjected to change. In particular, unexpected events within the software but also the environment can make a subsequent adaptation of the software necessary. These adaptations are currently mainly done manually, through changes of configuration options or changes within the software itself. Adaptive systems are capable of executing these changes autonomously.

Multi-agent systems are ideal candidates to model and implement complex software systems. However, the increasing complexity makes targeted changes more and more expensive and error prone. The goal of the CC ACT therefore is to enhance the agent framework JIAC with self-adaptive capabilities. To this end, different methods from the areas of self-organization, evolutionary algorithms, and other biologically inspired approaches are examined. Currently, the focus lies on using these methods to implement autonomous management and self-healing of software agents.