Deep Learning Based Energy Predictive Model for Task Scheduling in Sustainable IoT

The proliferation of IoT devices has become a worldwide concern due to the rapid growth and integration of smart technologies across various industries and consumer applications. Thus, the demand for sustainable IoT has remained in the top interest of designers. Power consumption optimization is the most critical factor affecting the design of sustainable computing architectures in IoT devices, necessitating the implementation of task scheduling techniques. In this manuscript, an energy predictive framework has been presented for task scheduling operations in sustainable IoT devices. The proposed model has been developed using Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network. Additionally, Single Spectrum Analysis (SSA) technique has been incorporated to improve the prediction accuracy. The model has been trained using real-time data. It predicts harvestable solar energy possessing Mean Absolute Error (MAE) as low as 0.001664 and Mean Square Error (RMSE) of 0.002749, outperforming benchmark DL networks.