Talos: An Architecture for Self-Configuration

Autonomic computing concerns the design and development of software systems with self-maintenance capabilities and behaviors that conform to high-level policies. Self-reconfiguration is a key feature of an autonomic system. It enables self-transition from the current configuration, which may have failed or be under-performing, to a new configuration that better addresses system goals. In this paper, we propose Talos, a conceptual architecture based on the Belief-Desire-Intention (BDI) paradigm that supports various forms of self-reconfiguration. We also present a first realization of Talos in Jason, a BDI infrastructure based on the AgentSpeak programming language. Our approach follows software engineering principles and is founded on goal-oriented analysis from Requirements Engineering.