An Edge-Device Coordination Approach Based on Resource-Sharing for Real-Time Load Monitoring

Faulty electrical wiring, overloading, and especially hazardous appliances are now threatening more and more peoples lives. Consumer electronics, particularly smart home devices such as smart sockets with load monitoring capabilities are gaining popularity, since they are expected to control the risks of electricity consumption based on appliance-level information. However, their timeliness and practicality remain a challenging issue. As a result, this paper presents an edge-device coordination (EDC) approach for real-time load monitoring that fully shares and exploits idle computing resources in edge-device systems. The complicated electrical safety protection tasks of smart home devices are first modeled in detail. Next, a novel edge-device coordination framework is proposed, building on a refined resource manager and a task scheduler that handles uncertainty with lightweight deep reinforcement learning. Furthermore, an electrical safety protection system that includes smart socket with communication, sensing, and control capabilities, as well as a Qt-based edge server are designed for experiments. In the simulation of a house, the task completion rate increases by 17.1%, while the efficiency improves by 3.28% with the proposed EDC approach. We further validate the scalability and robustness of EDC approach under varying device numbers and network conditions, and demonstrate its superiority over state-of-the-art methods.