SCT Based Adaptive Data Aggregation for Wireless Sensor Networks

Abstract

Communication overhead is a major concern in wireless sensor networks because of inherent behavior of resource constrained sensors. To degrade the communication overhead, a technique called data aggregation is employed. The data aggregation results are used to make crucial decisions. Certain applications apply approximate data aggregation in order to reduce communication overhead and energy levels. Specifically, we propose a technique called semantic correlation tree, which divides a sensor network into ring-like structure. Each ring in sensor network is divided into sectors, and each sector consists of collection of sensor nodes. For each sector, there will be a sector head that is aggregator node, the aggregation will be performed at sector head and determines data association on each sector head to approximate data on sink node. We propose a doorway algorithm to approximate the sensor node readings in sector head instead of sending all sensed data. The main idea of doorway algorithm is to reduce the congestion and also the communication cost among sensor nodes and sector head. This novel approach will avoid congestion by controlling the size of the queue and marking packets. Specifically, we propose a local estimation model to generate a new sensor reading from historic data. The sensor node sends each one of its parameter to sector head, instead of raw data. The doorway algorithm is utilized to approximate data with minimum and maximum bound value. This novel approach, aggregate the data approximately and efficiently with limited energy. The results demonstrate accuracy and efficiency improvement in data aggregation.