Adaptive routing protocol for large-scale power internet of things based on edge computing and federated learning

The rapid proliferation of heterogeneous devices in Power Internet of Things (PIoT) presents significant challenges for efficient routing and energy management in large-scale deployments. This paper proposes a novel adaptive routing protocol that integrates edge computing and federated learning to address the complexities of heterogeneous device coordination and energy efficiency optimization in PIoT environments. The proposed approach employs a hierarchical architecture where edge nodes serve as distributed processing points, enabling local decision-making while maintaining global optimization through federated learning mechanisms. The adaptive routing algorithm dynamically adjusts routing parameters based on real-time network conditions, device characteristics, and energy constraints, while the federated learning framework enables collaborative optimization without centralized data sharing. Comprehensive experimental evaluation demonstrates that the proposed protocol achieves 35-50% higher network throughput, 40-60% reduction in end-to-end delay, and 45-65% energy savings compared to traditional routing protocols. The results validate the protocol's superior scalability and robustness, maintaining consistent performance across networks of up to 10,000 devices while traditional approaches exhibit significant performance degradation beyond 1000 devices.