A Lorawan-Based Smart Home Energy Management System Empowered by Artificial General Intelligence

This paper presents the design and development of an intelligent, long range wide area network (LoRaWAN)-based smart home energy management system that integrates real-time analytics and artificial intelligence (AI) techniques. Utilizing a publicly available, appliance-level energy consumption dataset, the system simulates realistic household usage patterns across multiple devices. A MATLAB-based LoRaWAN transmission model, incorporating packet loss and signal attenuation, emulates low-power wide-area network (LPWAN) behavior under realistic wireless communication conditions. Key system functionalities include anomaly detection via statistical thresholding and moving averages, energy consumption forecasting through linear regression and decision tree models, i.e., fitrtree, idle device detection, and daily cost prediction in United Arab Emirates Dirhams (AED). The system demonstrates moderate predictive accuracy, with mean R2 values of approximately 0.55 per appliance. An interactive command-line interface and an artificial general intelligence (AGI)-inspired natural language chat module enhance usability, enabling non-technical users to query energy data and cost insights effectively. Visualization tools support real-time energy pattern recognition and future consumption forecasting, facilitating informed user decisions. Despite simulated packet loss and missing data, the system maintains robust performance through data interpolation and resilient model training. The proposed framework lays the foundation for scalable, intelligent home energy systems and offers pathways toward deeper learning integration, dynamic pricing models, and edge deployment for real-time autonomous energy management.