Energy Usage Patterns in Smart Campus With STM32 and GreenLoRaWAN Communication Protocol‐Based Temporal Attention Recurrent Graph Convolutional Neural Network on Edge Computing

ABSTRACT An effective smart campus energy management strategy requires precise monitoring and prediction of energy consumption across different facilities, which can vary under occupancy and operational conditions. In this paper, Energy Usage Patterns in Smart Campus with STM32 and GreenLoRaWAN Communication Protocol based Temporal Attention Recurrent Graph Convolutional Neural Network on Edge Computing (SC‐STM32‐LWAN‐TRCNN) is proposed. Here, the input data are gathered from the smart campus using the sensor node. To execute this, the energy usage and environmental parameters across the smart campus are monitored using STM32‐based sensor nodes in edge computing. By using GreenLoRaWAN Communication Protocol (GLoRaNCP), the sensor nodes communicate with data wirelessly to the central gateway. Then, the input data are pre‐processed using Multiple Local Particle Filter (MLPF) to reduce or remove the noise from the gathered input data. The pre‐processed data are given into the Temporal Attention Recurrent Graph Convolutional Neural Network (TRCNN) for predicting the energy usage pattern from the smart campus. Finally, Shrike Optimization Algorithm (SOA) is employed to optimize the weight parameter of the TRCNN classifier that predicts the energy usage pattern precisely from the smart campus. The proposed SC‐STM32‐LWAN‐TRCNN method is implemented and analyzed with the help of performance metrics, such as accuracy, mean absolute error (MAE), root‐mean‐squared error (RMSE), energy consumption, communication efficiency, throughput, and computation time. The performance of the SC‐STM32‐LWAN‐TRCNN approach achieves 22.74%, 20.86%, and 30.67% higher accuracy and 25.19%, 17.54%, and 33.79% lower computation time compared with existing methods, respectively.