WAPI-Integrated Congestion Control and Adaptive Transmission for Power IoT

Abstract

Modern power IoT systems feature massive terminal deployment and increasingly diverse service scenarios. Such conditions frequently cause network congestion, packet disorder and unstable transmission, severely undermining overall network performance. Traditional optimization methods cannot adapt to dynamic heterogeneous network environments and lack native compatibility with the widely adopted WAPI security protocol. Addressing these core defects in power IoT transmission, this study builds an integrated communication optimization framework with embedded WAPI security modules. It further proposes SymDistill-CC, a new algorithm serving as the core congestion control component of the framework. It leverages the teacher-student distillation architecture to transform deep reinforcement learning policies into readable symbolic decision trees, offering transparent decision logic, comparable transmission performance and significantly reduced computation overhead. We then put forward the ECN-FPS packet scheduling mechanism. It detects shared bottlenecks via explicit congestion signaling and use forward prediction to mitigate misordered packets, substantially easing buffer pressure on receiving terminals. Finally, we construct a security-aware dynamic retransmission mechanism for power IoT that perceives network states in real-time under the WAPI authentication framework and adaptively selects unicast or multicast retransmission modes to enhance transmission efficiency and resource utilization. Experimental results demonstrate that in simulated WAPI network environments, the proposed methods achieve a 15%–20% throughput improvement, a 42.5% reduction in out-of-order packets and a link utilization up to 88%–90%, with significant advantages also shown in retransmission efficiency and other key metrics, providing an effective solution for constructing secure, efficient, and reliable power IoT communication systems.