EEDCS: Energy Efficient Data Collection Schemes for IoT Enabled Wireless Sensor Network

In the data collection of WSN, the main challenge is that the communication has to be energy-efficient to overcome the limited battery constraint along with increasing the lifetime of the networks. A number of research articles are available in the literature for energy-efficient communication in WSN which is mostly based on the Auto regressive integrated moving average (ARIMA) model along with Principal component analysis (PCA) model. Although the previously proposed schemes have certain limitations like they assume in terms of the relationship between data elements is linear and the standard deviation is constant which is not practical solution. Whereas the PCA model may be only applicable in the complete data set. It means any lost element in the data set can degrade the accuracy of the data compression (DC) scheme. To consider the above-mentioned limitations in this study, we have propose an energy-efficient model for clustered WSN by minimizing transmission overheads. The proposed model is a sequence of two consecutive schemes for data prediction (DP) along with DC. The DP scheme is based on Bi-directional long short term memory (B-LSTM) model and the DC scheme is based on Probablistic principal component analysis (P-PCA) model. We have conducted an extensive simulation for the proposed scheme. Furthermore, we have compared our proposed scheme with the existing related schemes. The results show that the proposed scheme outperforms compared with other relater schemes.