Study on Cost-Sensitive Communication Models on Large-scale Monitor Networks

Large-scale distributed monitor networks are in face of the challenge of tremendous data communication costs due to the resource restriction. Prediction models can be used to reduce communication cost over the networks. A framework is proposed which provides a mechanism to maintain adaptive prediction models that significantly reduce communication cost over the distributed environment while still guaranteeing sufficient precision of demand results. Prediction models are also proposed to process prediction queries over future data streams in this paper. Three particular models, static model, linear model and acceleration model, and the corresponding tuning schemas are given. Experimentations are performed based on the simulated data and ocean air temperature data measured by TAO (tropical atmosphere ocean). Analytical and experimental evidence show that the proposed approach significantly reduces overall communication cost and performs well over prediction queries.