Federated learning emerges as a key enabler toward pervasive intelligence across IoT ecosystems with provable privacy guarantees. While recent efforts on client selection have been made for optimizing its communication efficiency in iterative model aggregation over resource-constrained networks, their scalability fundamentally breaks down in dense deployments. This limitation stems from the NP-hard complexity of congestion-aware scheduling, where co-channel interference creates exponentially growing solution spaces. In this context, we present a novel client selection framework with asymptotic scalability in massive IoT, which leverages the intrinsic graph topology with insights from statistical physics. First, this work formulate a universal client selection problem, capturing both positive network externalities derived from collaborative knowledge exchange and congestion effects induced by co-channel interference. This formulation is then transformed into a node classification task via Ising spin Hamiltonian mapping, establishing an explicit connection between federated learning, statistical physics, and graph optimization. Building on this foundation, we develop a lightweight graph neural solver that adaptively selects clients via recognizing node state with iterative neighbor aggregation of learnable embeddings. Comprehensive experiments validate that our approach maintains state-of-the-art scheduling performance, while scaling to network sizes orders of magnitude beyond what conventional methods can handle.

Industrial control networks still face significant challenges in grayscale-based intrusion detection, including feature degradation, dispersed attention weights, and computational redundancy, particularly under dynamic attacks and large-scale traffic streams. To address these issues, this paper proposes the Dynamic Gradient-Regulated Attention Network (DyGRA-Edge), a framework tailored for resource-constrained industrial IoT edge environments. It integrates three key mechanisms: a Dynamic Gradient Controller (DGC) guided by Lyapunov stability to regulate feature evolution, anisotropic diffusion convolution (ADC) to enhance edge responses in low-contrast grayscale images, and sparse entropy attention (SEA) to minimize conditional entropy and suppress redundant features. Experiments conducted on two real-world industrial platforms—oil-gas gathering and catalytic reforming—demonstrate that DyGRA-Edge achieves detection accuracy above 98.5%, with a false positive rate of 2.1% and inference latency between 63–98 ms. Ablation studies confirm the complementary contributions of each module, while comparisons with lightweight models such as MobileViT and FastViT show that DyGRA-Edge provides superior robustness against noisy traffic and redundant features. Edge deployment validation further indicates that in a Raspberry Pi environment using the NSL-KDD dataset, DyGRA-Edge maintains an average CPU utilization of only 5.3%. These results highlight that DyGRA-Edge achieves a strong balance of accuracy, real-time performance, and resource efficiency, making it a high-precision and low-latency IDS solution for industrial edge environments.