A Software Architecture Approach for Structuring Autonomic Systems

Autonomic systems manage themselves given high-level objectives by their administrators. They utilise feedback from their own execution and their environment to self-adapt in order to satisfy their goals. An important consideration for such systems is a structure which is conducive to self-management. This paper presents a structuring methodology for autonomic systems which explicitly models self-adaptation while separating functionality and evolution. Our contribution is a software architecture-based framework combining an architecture description language based on pi-calculus for describing the structure and behaviour of autonomic systems, a development methodology for evolution and mechanisms for feedback and change.