A Distributed Processing Technique for Sensor Data Applied to Underwater Sensor Networks

International audienceWireless sensor networks (WSN) present a low cost solution to enhance our lives. They allow a large variety of applications. One of the major challenges faced by WSN is that of energy saving. A well known efficient way to reduce energy consumption is data reduction. It consists in reducing the amount of data sensed and transmitted to the sink. Consequently, sensor data communication should be minimized in order to increase network lifetime. In this paper, we propose an energy-efficient data reduction technique based on a clustering architecture. Our objective is to identify and study data similarities at both sensor and cluster-head (CH) levels. At the first level, each sensor sends a set of representative points to the CH at each period, instead of sending the raw data. When data points are received by the CH, it uses the Euclidean distance in order to eliminate redundant data generated by neighboring sensor nodes, before sending them to the sink. To validate our approach, we applied our techniques on real underwater sensor data and we compared them with other existing data reduction methods. The results show the effectiveness of our technique in terms of improving the energy consumption and the network lifetime, without loss in data fidelity