Fault tolerant data transmission reduction method for wireless sensor networks

Abstract

Several theoretical studies have clearly demonstrated that the Dual Prediction Mechanism (DPM) remains the most efficient technique for data reduction in Wireless Sensor Networks (WSNs). In real world, the deployed sensor nodes suffers from packet loss and even failures which renders the DPM unreliable, since it requires flawless synchronization between the source (sensor node) and the destination (Sink). In this paper, we introduce a Fault Tolerant Data Transmission Reduction (FTDTR) technique consisting of three main components: DPM-based transmission reduction, synchronization and packet loss detection, and finally reconstruction of missing data. Our method was evaluated on real-world data sets collected at our laboratory and compared to three recent prediction-based data reduction approaches. The results were promising in quality of the replicated measurements and transmission reduction.