Ambient Intelligence to Detect Misuse of Electricity Consumption Based on IoT Using Blockchain Technology

Abstract

Electricity abuse and energy inefficiencies are still open issues in smart grid systems, demanding high-performance anomaly detection mechanisms. In this paper, we propose an IoT-enabled electricity monitoring system that combines machine learning (LightGBM) and blockchain (Polygon network) for real-time anomaly detection, secure data storage, and transparent energy tracking. IoT smart meters are utilized to monitor real-time electricity usage data, whereas LightGBM classifies anomalies efficiently with high precision. The key innovation is the use of blockchain for decentralized anomaly logging with tamper-proof records and enhanced trustworthiness. Unlike centralized approaches, Polygon blockchain immutably stores electricity data, giving verifiable anomaly logs. Using an interactive IoT dashboard and real-time notifications, users can monitor consumption patterns and respond to anomalies efficiently. The proposed system achieves 96.77% accuracy, an AUC-ROC of 0.99, and an F1-score of 96.69%, outperforming CNN-LSTM and CNN-XGBoost in both accuracy and use of computational resources. Despite these advantages, blockchain transaction costs (0.001996 POL per transaction) and IoT-wallet integration complexity pose challenges. Future work will explore cost-reducing blockchain optimizations such as meta-transactions and relayers, along with model enhancements for IoT scalability. This approach provides a low-cost, scalable, and secure solution for smart energy management, enhancing grid security and sustainability.