Enhancing Energy Efficiency in Sensor Cloud Through Time Series Forecasting of Sensor Data

In today's interconnected world, diverse sensor types are critical for powering various applications and services.The limited energy resources of these sensors present a significant challenge in managing sensor networks efficiently.To address this, we propose an energy-saving sensor cloud that utilizes a data prediction technique.Typically, a sensor node in a Wireless Sensor Network (WSN) gathers and transmits data to the cloud every 10 minutes, consuming substantial energy.In contrast, our proposed method requires sensor nodes to communicate with the cloud every 110 minutes, as the cloud system's forecasting method is capable of predicting ten steps ahead, thus reducing transmission frequency.We have applied Wavelet-based Forecasting (WBF), Auto-Regressive Integrated Moving Average (ARIMA), and a hybrid ARIMA-WBF for these predictions.The ARIMA model demonstrates superior performance compared to the other techniques when dealing with linear sensor data.Our method results in a power consumption that is approximately 90.9% lower than that of traditional methods within the sensor cloud, owing to reduced data transmission frequency.Additionally, our approach yields notably lower Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) in predictions.