Agent dependability as an architectural issue

Layered architectures are a proven principle for the design of software systems and components. The paper introduces a layered reference architecture for software agents which assigns each agent property to select layers. It demonstrates how the same reference architecture provides a framework for a dependability model that locates the sources of failures and the ensuing error handling with a specific layer, thus integrating dependability directly into the design of agents.

• Type: Conference
• Year: 2006
• Where: ATAL
• Authors: Peter C. Lockemann, Jens Nimis
• Keywords: Architecture, ATAL 2006, Intelligent Agents, Layered Reference Architecture, Reference Architecture