Clustered Predictive Model Based Adaptive Sampling Techniques in Wireless Sensor Networks

According to the data spatial correlation of Wireless Sensor Networks (WSNs), this study proposes a clustering mechanism based on the data gradient. In the proposed clustering mechanism, the cluster head nodes maintain Auto Regressive (AR) prediction model of the sensory data within each cluster in the time domain. Moreover, the cluster head nodes adjust the temporal sampling frequency based on the implementation of above predicted adaptive algorithm model. By adjusting the temporal sampling frequency, the redundant data transmission is reduced as well as ensuring desired sampling accuracy, so as energy efficiency is improved. The temporal sampling frequency adjustment algorithm takes into account spatial and temporal combined correlation characteristics of sensory data. As a result, the simulation results demonstrate the performance benefits of the proposed algorithm.