An Edge-Based Intelligent IoT Control System: Achieving Energy Efficiency with Secure Real-Time Incident Detection

In this study, we present an edge-based intelligent IoT control system designed to optimize energy efficiency while ensuring secure and real-time incident detection. The system leverages deep reinforcement learning (DRL) to dynamically adjust the transmission periods of IoT devices, effectively balancing energy consumption with high-quality data monitoring. Our approach employs sophisticated state representation techniques, including piecewise aggregate approximation and gramian angular field matrices, for efficient time series data processing and dimensionality reduction. The system's adaptive transmission period control mechanism adjusts data collection intervals in response to environmental volatility, enhancing both energy conservation and anomaly detection accuracy. A multi-faceted reward function, integrating data monitoring quality, energy efficiency, and incident response time, guides the DRL agent toward optimal decision-making in complex, dynamic environments. Extensive evaluations using open-source indoor air pollution datasets demonstrate that our proposed method significantly outperforms both traditional and state-of-the-art approaches. The system achieves substantial energy savings while maintaining superior anomaly detection capabilities, as evidenced by improved response times and reduced root mean square errors during anomalous events. This system can provide a scalable, robust, and adaptive solution for diverse environmental monitoring applications.