From self-aware building blocks to self-organizing systems with hierarchical agent-based adaptation

How to develop a self-aware system from modularized design components is a major challenge on this theme. The authors introduce a hierarchical agent-based system architecture, which enables self-organization and self-adaptation upon parallel embedded systems. Self-organization is achieved by dynamic clusterization, which groups self-aware components into a cluster and continuously updates the organization to account for application changes and internal errors. Self-adaptation is performed by hierarchical agents, based on the run-time organization, to monitor corresponding levels of components and reconfigure the system to improve ture achieves functional scalability via partitioning of agent ware/hardware co-design of agents.