Wireless industrial cyber-physical systems are increasingly popular in critical manufacturing processes. These kinds of systems, besides high performance, require strong security and are constrained by low computational capabilities. Physical layer authentication (PHY-AUC) is a promising solution to meet these requirements. However, the existing threshold-based PHY-AUC methods only perform ideally in stationary scenarios. To improve the performance of PHY-AUC in mobile scenarios, this article proposes a novel threshold-free PHY-AUC method based on machine learning (ML), which replaces the traditional threshold-based decision-making with more adaptive classification based on ML. This article adopts channel matrices estimated by the wireless nodes as the authentication input and investigates the optimal dimension of the channel matrices to further improve the authentication accuracy without increasing too much computational burden. Extensive simulations are conducted based on a real industrial dataset, with the aim of tuning the authentication performance, then further field validations are performed in an industrial factory. The results from both the simulations and validations show that the proposed method significantly improves the authentication accuracy.

In this paper, we investigate the security threats in mobile edge computing (MEC) of Internet of things, and propose a deep-learning (DL)-based physical (PHY) layer authentication scheme which exploits channel state information (CSI) to enhance the security of MEC system via detecting spoofing attacks in wireless networks. Moreover, three gradient descent algorithms are adopted to accelerate the training of deep neural networks, which enables smaller computation overheads and lower energy consumptions. In addition, the maximum likelihood function of multi-user authentication method is derived, which explains why cross entropy is chosen as the loss function. The vectorization cost function is also derived. The mini batch scheme and $\ell _{2}$ regularization are adopted to improve training accuracy and avoid over-fitting, respectively. Moreover, the simulation and experimental results show that the DL-based PHY-layer authentication approaches can distinguish multiple legitimate edge nodes from malicious nodes and attacker by CSIs, effectively. Our proposed method supports a better performance compared with the traditional hypothesis test based method.

Trealize the Industry 4.0 vision and enable mobile connectivity and flexible deployment in harsh industrial environments, wireless communication is essential. But before wireless communications technology can be widely deployed for critical control applications, first it must be assessed, and that requires a comprehensive characterization of the wireless channel. This can be done by analyzing large amounts of wireless data collected from different industrial environments. In this article, we discuss the possibilities offered by a recently published industrial wireless data set. This data set is more exhaustive than measurements previously reported. We show two cases of how those data have been applied to improve latency performance and to investigate the feasibility of physical-layer security techniques for wireless communication in industrial environments.

Introduction: Zanthoxylum bungeanum Maxim is an economically significant crop in Asia, but large-scale cultivation is often threatened by frequent diseases, leading to significant yield declines. Deep learning-based methods for crop disease recognition have emerged as a vital research area in agriculture.Methods: This paper presents a novel model, LT-DeepLab, for the semantic segmentation of leaf spot (folium macula), rust, frost damage (gelu damnum), and diseased leaves and trunks in complex field environments. The proposed model enhances DeepLabV3+ with an innovative Fission Depth Separable with CRCC Atrous Spatial Pyramid Pooling module, which reduces the structural parameters of Atrous Spatial Pyramid Pooling module and improves cross-scale extraction capability. Incorporating Criss-Cross Attention with the Convolutional Block Attention Module provides a complementary boost to channel feature extraction. Additionally, deformable convolution enhances low-dimensional features, and a Fully Convolutional Network auxiliary header is integrated to optimize the network and enhance model accuracy without increasing parameter count.Results: LT-DeepLab improves the mean Intersection over Union (mIoU) by 3.59%, the mean Pixel Accuracy (mPA) by 2.16%, and the Overall Accuracy (OA) by 0.94% compared to the baseline DeepLabV3+. It also reduces computational demands by 11.11% and decreases the parameter count by 16.82%.Discussion: These results indicate that LT-DeepLab demonstrates excellent disease segmentation capabilities in complex field environments while maintaining high computational efficiency, offering a promising solution for improving crop disease management efficiency.

Industrial wireless cyber physical systems are vulnerable to malicious node attacks, especially clone attack. The existing clone detection schemes are either based on upper layer observations or physical layer CSI. The schemes based on upper layer observations are vulnerable to defamation while the schemes based on CSI perform better against defamation, but are badly affected by channel conditions. This article applies physical layer reputation and BPNN to clone detection, aiming at improving the detection accuracy. The proposed scheme accumulates the physical layer reputations by CSI and input them to the neural network. The cloud server performs attack detection by group detection first. If a certain group is classified as attacked, the corresponding edge processor will trace the attack and identify the specific clone nodes. During the attack tracing, multiple reputations of each node is adopted for a comprehensive inspection. Extensive experiments are conducted on USRPs to show the accuracy improvement.