An Event-Based Data Aggregation Scheme Using PCA and SVR for WSNs
Xiaojing Zhang, Hao Wu, Qingyuan Li, Bin Pan · 2017
5G and Internet of things (IOT) develop rapidly, but the major applications of IOT-wireless sensor networks(WSNs) have numerous data, resulting in serious transmission load. In order to reduce the number of transmitted packets, this paper focuses on data aggregation for WSNs and proposes a novel event-based data aggregation mechanism using both principle component analysis(PCA) and support vector regression(SVR). The proposed method first uses correlation to achieve event checker at data aggregation node. Then when the state is normal, PCA is performed for data aggregation to reduce data dimensionality. When the state is urgent, data aggregation node receives changing data and transmits the sensing data to base station instantly, meanwhile, the data aggregation node performs SVR-based prediction. According to the prediction accuracy, data aggregation node adjusts the data transmission time interval adaptively. The simulation results show that the proposed scheme reduces the amount of transmission data among base station and one or more data aggregation nodes and decreases energy consumption compared with Adaptive-PCA.