Distributed context-aware Affinity Propagation clustering in Wireless Sensor Networks

Abstract

We foresee the need for dynamically clustering nodes in Wireless Sensor Networks (WSNs) according to a multitude of disparate co-existing contexts. To this end, we propose a distributed, low-overhead context-aware clustering protocol for WSNs. We employ Affinity Propagation (AP) for clustering nodes based on multiple criteria including location, residual energy, and contextual data sensed from the environment. We propose a novel approach for context representation based on potential fields. We discuss the integration of our context representation model with AP and demonstrate using simulation the effectiveness and proficiency of the proposed protocol in satisfying its intended objectives.