Deep learning (DL)-based physical-layer authentication (PLA) has shown promising performance in industrial Internet of Things (IIoT) scenarios. Existing methods formulate PLA as a single multiclassification task with a static training dataset and implicitly assume the training and authentication data are drawn from the same distribution. However, user mobility over time in mobile IIoT leads to a temporal distribution shift in channel impulse response (CIR), severely degrading authentication performance. To address this challenge, we propose a temporal meta-metric learning method for PLA (TMML-PLA), a temporal meta-metric learning method that reformulates multiuser PLA as a sequence of temporal few-shot classification tasks based on the intrinsic temporal correlation characteristic of CIR. When new CIR samples require authentication, the model dynamically constructs a task using a small set of the most recent CIR samples of each user as the support set to ensure distributional consistency. A temporal meta-metric learning scheme based on the prototypical network (ProtoNet) is introduced to tackle these tasks under evolving CIR distributions, and a temporal decay mechanism is designed to emphasize recent training samples. Experiments on a real-world IIoT dataset demonstrate that TMML-PLA significantly improves both accuracy and robustness. For instance, when the number of users is 15, and the authenticating CIR samples increase over time, TMML-PLA maintains stable accuracy from 88.20% to 85.75% while the DL-based ResNet drops sharply from 83.93% to 46.95%, confirming its robustness and accuracy in mobile IIoT scenarios.

Deep learning (DL)-based multi-user physical layer authentication (PLA) in the Industrial Internet of Things (IIoT) requires frequent updates as new users join. Class incremental learning (CIL) addresses this challenge, but existing generative replay approaches depend on heavy parameterized models, causing high computational overhead and limiting deployment in resource-constrained environments. To address these challenges, we propose a parameter-free statistical generator-based CIL framework, PSG-CIL, for DL-based multi-user PLA in the IIoT. The parameter-free statistical generator (PSG) produces Gaussian sampling on user-specific means and variances to generate pseudo-data without training extra models, greatly reducing computational overhead. A confidence-based pseudo-data selection ensures pseudo-data reliability, while a dynamic adjustment mechanism for the loss weight balances the retention of old users’ knowledge and the adaptation to new users. Experiments on real industrial datasets show that PSG-CIL consistently achieves superior accuracy while maintaining a lightweight scale; for example, in the AAP outer loop scenario, PSG-CIL reaches 70.68%, outperforming retraining from scratch (58.57%) and other CIL methods.

Active speaker detection is a challenging task in audio-visual scenario understanding, which aims to detect who is speaking in one or more speakers scenarios. This task has received extensive attention as it is crucial in applications such as speaker diarization, speaker tracking, and automatic video editing. The existing studies try to improve performance by inputting multiple candidate information and designing complex models. Although these methods achieved outstanding performance, their high consumption of memory and computational power make them difficult to be applied in resource-limited scenarios. Therefore, we construct a lightweight active speaker detection architecture by reducing input candidates, splitting 2D and 3D convolutions for audio-visual feature extraction, and applying gated recurrent unit (GRU) with low computational complexity for cross-modal modeling. Experimental results on the AVA-ActiveSpeaker dataset show that our framework achieves competitive mAP performance (94.1% vs. 94.2%), while the resource costs are significantly lower than the state-of-the-art method, especially in model parameters (1.0M vs. 22.5M, about 23x) and FLOPs (0.6G vs. 2.6G, about 4x). In addition, our framework also performs well on the Columbia dataset showing good robustness. The code and model weights are available at https://github.com/Junhua-Liao/Light-ASD.