Towards Aware, Adaptive and Autonomic Sensor-Actuator Networks

Abstract

We propose A3SAN, a framework for context-aware, resource-aware, autonomic, and adaptive management of Sensor-Actuator Networks (SANSs). We introduce new techniques for autonomic network configuration and management in reaction to context and resource dynamics. We propose a novel approach for quantitative context representation and management based on Potential Fields that allows us to quantify interesting events spatiotemporally, and simplifies the fusionand grouping of concurrent contexts. Adaptability is achieved by associating each node in the network with a dynamic Node Affinity Profile, which determines its suitability to serve each event type. Different configuration and management tasks such as clustering, task allocation, and role assignment are carried out using a distributed variant of the Affinity Propagation algorithm. A Fuzzy Logic based decision-making engine provides effective context analysis and conflict resolution between competing tasks, enabling swift adaptation to context and resource dynamics. Using simulation, we evaluate the efficacy of these techniques, and their ability to achieve our goal of efficient and autonomous management of SANs.