A Novel Energy-efficient Sensor Cloud Model using Data Prediction and Forecasting Techniques

Abstract

An energy-efficient sensor cloud model is proposed based on the combination of prediction and forecasting methods. The prediction using Artificial Neural Network (ANN) with single activation function and forecasting using Autoregressive Integrated Moving Average (ARIMA) models use to reduce the communication of data. The requests of the users generate in every second. These requests must be transferred to the wireless sensor network (WSN) through the cloud system in the traditional model, which consumes extra energy. In our approach, instead of one second, the sensors generally communicate with the cloud every 24 hours, and most of the requests reply using the combination of prediction and forecasting methods in the cloud system, which results in less communication and more battery life for the sensor. In our model, we used the ANN model initially, which had predicted the temperature for a given day with an accuracy of 92%. The results of ANN, together with the earlier real temperatures, are given as input to the ARIMA forecasting model, which provides an accuracy of 96% for one day in advance. Our simulation shows that the proposed method saves more energy compared to the traditional approach.

Evaluating Hybrid Framework of VASNET and IoT in Disaster Management System

“…The leading cause of sensor node's energy waste is the radio system [4]. For that reason, several concepts and strategies has been emphasized on power saving in reducing the data sending like scheduling, aggregation, routing and clustering [8]. Generally, when selecting different types of wireless network technologies linking with DMS, a few considerations we need to take into account on the particular application like power consumption and maximum distance range [9].…”