A QoS-driven Self-Adaptive Architecture for Wireless Sensor Networks

Recently, Wireless Sensor Networks (WSN) have become increasingly used to perform distributed sensing and convey useful information. These kinds of environments are complex, heterogeneous and often affected by unpredictable behavior and poor management. This fostered considerable research on designs and techniques that enhance these systems with an adaptation behavior. In this paper, we focus on the self-adaptation branch of the research and give an overview of the current existing approaches. We also analyze the collected approaches and we summarize their common and individual characteristics. Then, we describe our proposed approach to adapt running WSN applications while adopting the autonomic control loop [1]; MAPE: Monitoring, Analysis, Planning, and Execution. Differently from other approaches, where adaptation is generally performed by simply re-deploying another version of application, we focus on the distinction between three different levels of adaptation. We define a sensor level (level1) composed of terminal leaf nodes, a cluster head level (level2) that is an elected node with collection capability and a base station level (level3) which is an enhanced capabilities node that can be a computer or a mobile smart phone. This makes our system able to provide quick adaptation to multiple context parameter changes and to deal with multiple users requirements changes in order to preserve energy consumption efficiency, and maintain system lifetime durability. To illustrate our approach, we study the Smart Home Health Care (SHHC) system over the AZEM simulator which is an enhanced version we developed of AvroraZ. This case study enables us to show the feasibility and the efficiency of our approach for self-adapting WSNs.