Spatio Temporal Error Adaptation Model (STEAM) for Energy Efficient Wireless Data Gathering

Abstract

Redundant Data communication greatly downgrades the performance and reliability of Wireless Sensor Network's (WSN). Spatial similarity and temporal similarity are the intrinsic features of sensory data. By reducing this spatio-temporal data redundancy, large amount of residual energy and bandwidth can be conserved. Very few methods use both spatial and temporal correlation to reduce data redundancy, where the system's error threshold is divided between spatial and temporal error thresholds that are fixed throughout the network's life span. In Spatio-Temporal Error Adaptation Model (STEAM), the spatio-temporal correlation between sensory data is exploited towards data reduction. The spatial redundancy is reduced by a data aware clustering and temporal redundancy is reduced by model based prediction approach. In the proposed work, the proportion of spatial and temporal error thresholds on total error is adapted based on data dynamics and spatial pattern changes of the network. The proposed work decreases considerable amount of data communication, while maintaining the data accuracy within user defined error tolerance. The proposed work follows a distributed approach, hence it is highly scalable. Since the work adapts the threshold based on network parameters, the performance is improved on diverse network conditions. The work attains up to 89% communication reduction in wireless data gathering system with an error tolerance of 0.4°C on collected data.