Autonomic Context-Aware Wireless Sensor Networks

Abstract

Autonomic Computing allows systems like wireless sensor networks (WSN) to self-manage computing resources in order to extend their autonomy as much as possible. In addition, contextualization tasks can fuse two or more different sensor data into a more meaningful information. Since these tasks usually run in a single centralized context server (e.g., sink node), the massive volume of data generated by the wireless sensors can lead to a huge information overload in such server. Here we propose DAIM, a distributed autonomic inference machine distributed which allows the sensor nodes to do self-management and contextualization tasks based on fuzzy logic. We have evaluated DAIM in a real sensor network taking into account other inference machines. Experimental results illustrate that DAIM is an energy-efficient contextualization method for WSN, reducing 48.8% of the number of messages sent to the context servers while saving 19.5% of the total amount of energy spent in the network.