The digitization and intellectualization have been envisioned as the fundamental basis for future Industrial Internet of things, which integrates sensor technology, industrial control technology, communication technology, and artificial intelligence (AI). Specifically, the collaboration among these above techniques is crucial for the successful implementation of intelligent applications. This article develops an end-to-end visual control framework to accomplish multi-crane collaborative sorting in wireless time sensitive networking (TSN) networks. The design primarily incorporates field devices, data transmission, artificial intelligence (AI), and industrial control. An advanced binocular stereo visual recognition model based on deep learning is investigated to accurately obtain the world coordinates and types. A cooperative control scheduling model that combines a scheduling strategy with an anti-collision strategy is presented to effectively control multiple cranes for sorting tasks. The device data and commands are transmitted through industrial 5G-TSN integrated network for ultra-reliable, low-latency, and deterministic transmission. The proposed visual sorting system is further validated through the establishment of an experimental prototype, demonstrating its exceptional real-time performance while enabling flexible intelligent manufacturing.

Green network aims to promote the sustainable development of communication systems, and base station (BS) and cells sleeping has been proven effective in reducing the power consumption of these systems. However, the current Reinforcement Learning (RL) based methods for multi-cells collaborative sleeping face significant challenges in real-world applications due to the complex users-to-cells connection relationships, and have been rarely researched in city-scale deployments. In this article, a robust RL-based multi-cells sleeping model called Graph Deep Deterministic Policy Gradient (GDDPG) is developed for handling highly complex communication scenarios. Besides, we first propose a framework for deploying multi-cells sleeping models at the city scale. Then two algorithms are put forward for determining the essential cells needed to maintain basic radio coverage and for effectively grouping these cells, which are two crucial works in the framework. Additionally, to address the temporal variation of traffic patterns, transfer learning is employed to fine-tune the pre-trained RL model periodically. Finally, we validate the feasibility of city-scale deployment algorithms and demonstrate the effectiveness of GDDPG by leveraging a computational platform and real-collected cells data from a telecom operator in China. Experimental results show that GDDPG effectively manages the sleeping states of up to 72 cells in a real-world environment. The experimental scenario is much more complex than those in other studies.

Location-based service (LBS) is the core of internet of things (IoTs), which serves tracking, navigation and monitoring. The ubiquitous magnetic signals are temporally stable and spatially distinguishable, and can achieve high-precision and ubiquitous positioning results without additional infrastructure, which is favored by researchers and has become a major research hotspot. Although there has been extensive research in the field of indoor magnetic positioning, there is still room for optimization in terms of positioning accuracy and robustness. Aiming at the problem that the magnetometer is offset and susceptible to environmental interference, we propose an online magnetometer calibration algorithm without user perception. Aiming at the inconsistency of magnetic data spatial scale problem caused by differences in device sampling frequency and user walking speed, we leverage different scales to segment the magnetic data, extract the magnetic sequence features of the corresponding scales through Transformer, utilize the attention mechanism to score the weights of the different scale features, and finally fuse the multiple scale features for positioning. We conduct extensive and well-designed experiments on public datasets and self-collected datasets. The experimental results indicate that the proposed method effectively solves the magnetic spatial scale problem and improves indoor magnetic positioning accuracy.

This paper investigates the joint scheduling of cross-channel traffic resources in 5G-Time-Sensitive Networks (TSN) and proposes an adaptive prediction method suitable for cross-domain Channel State Information (CSI). Firstly, we analyze the 5G-TSN cross-domain data forwarding mechanism by leveraging the architecture of the 5G-TSN bridging network and combining the functions of 5G and TSN network elements. Secondly, we propose a representation method for the 5G-TSN cross-network wireless CSI, specifically the data transmission delay information, as a dataset for channel quality prediction. This serves as a data foundation for subsequent intelligent prediction. Next, to make better use of the local information of channel state and achieve fast convergence, we employ an Extreme Learning Machine Auto-Encoder (ELM-AE) prediction logic based on Deep Extreme Learning Machine (DELM) and introduce the Dung Beetle Optimizer (DBO) algorithm to improve the DELM regression prediction. We perform prediction and analysis of the 5G channel delay and TSN domain data transmission delay. Then, we use the 5G-TSN CSI, collected in practice, as the data source to train and test the wireless channel delay indicators, which helps form the 5G-TSN channel model. Finally, we build a laboratory transmission prototype test bed for 5G-TSN cross-network transmission and conduct end-to-end transmission delay testing based on the proposed offline-generated channel model. The results demonstrate that the channel prediction model enables the end-to-end delay to decrease to less than 5 ms, the cross-network time synchronization accuracy to reduce to less than 100 ns, and the relevant performance indicators to reach industry-leading levels.

We develop a deep-learning-based multicrane visual sorting system with virtualized programmable logic controllers (PLCs) in intelligent manufacturing, which enables the accurate location and suction of the materials on the conveyor belt. First, virtualized PLCs are deployed in the field and the cloud to break data islands for efficient communication between low-level devices. Second, artificial intelligence algorithms are integrated into the physical industrial control system in which cooperation between virtualized PLCs and the visual recognition model is developed to complete the industrial control closed loop. Third, we establish a visual recognition model in which object detection algorithms are used to process the original image and then obtain the position and type of the object in the pixel coordinate system. In addition, a new linear interpolation-based backpropagation neural network is presented to provide the transform relation between the pixel coordinate system and the world coordinate system that the crane needs to precisely suck the material. The whole system is applied in a time-sensitive network environment in a highly reliable and stable manner. The experimental prototype system demonstrates that high recognition accuracy can be achieved for the visual sorting system within an acceptable time frame. The accuracy of the sorting task reaches 96.5% and the average consumption time of each object is approximately 2.317 s when the speed of the conveyor belt is 5.2 m/min.

Abstract Next‐generation 6G networks will fully drive the development of the industrial Internet of Things. Steel surface defect detection as an important application in industrial Internet of Things has recently received increasing attention from the military industry, the aviation industry and other fields, which is closely related to the quality of industrial production products. However, many typical convolutional neural networks‐based methods are insensitive to the problem of unclear boundaries. In this article, the authors develop a region‐based fully convolutional networks with deformable convolution and attention fusion to adaptively learn salient features for steel surface defect detection. Specifically, deformable convolution is applied into selectively replace the standard convolution in the backbone of the region‐based fully convolutional networks, which performs significantly in scenarios with unclear defect boundaries. Moreover, convolutional block attention module is utilised in region proposal network to further enhance detection accuracy. The proposed architecture is demonstrated on two popular steel defect detection benchmarks, including NEU‐DET and GC10‐DET, which can effectively present the performance of steel surface defect detection by abundant experiments. The mean average precision on two datasets reaches 80.9% and 66.2%. The average precision of defect crazing, inclusion, patches, pitted‐surface, rolled‐in scale and scratches on NEU‐DET is 58.2%, 82.3%, 95.7%, 85.6%, 75.9%, and 87.9% respectively.

The space–air–ground-integrated network (SAGIN) will greatly promote the development of the Internet of Things (IoT). Green IoT will be an important part of SAGIN. Ambient backscatter communication (AmBC) is a potential solution for green SAGIN IoT. To improve the achievable sum rate (ASR) of the AmBC system, we propose a reconfigurable intelligent surface (RIS)-assisted AmBC system. In the single-backscatter device (BD) AmBC scenario, we first give the phase shifts that maximize the gain of the reflection link of the AmBC system, and then give the optimal reflection coefficient. The proposed scheme does not need to solve the convex semidefinite program (SDP) problem and has the characteristics of low computational complexity. In the multi-BD AmBC scenario, we first propose a multi-BD phase shifts initialization strategy to ensure the stability of the proposed scheme. Then, we give the optimal reflection coefficient and phase shifts based on the iterative method. Simulations show that the RIS-assisted AmBC scheme is superior to the non-RIS-assisted AmBC scheme.

Person reidentification (re-ID), which is a significant and potential application in the Internet of Things (IoT), aims to retrieve pedestrians of interest given a labeled image in a camera network. Now, it is still existing many challenges that severely influence feature representation in practical scenarios. Many methods adopt the attention mechanism in convolutional neural network (CNN) to improve the ability of feature learning. Although they only apply 1-D attention block in the popular deep learning architecture, the learned features are not discriminative for the feature representation. In this work, we investigate a self-focusing network (SFNet) that considers both the channel-dimensional attention and spatial-dimensional attention to adaptively learn more discriminative features. Namely, we embed the new attention module into the common backbone network, which can focus on the salient region by inhibiting the redundant features. Specifically, we design eight variants of the channel-dimensional attention and spatial-dimensional attention throughout the entire network and explore the most powerful feature representation. The heatmaps of different layers are visualized to intuitively present the performance of SFNet. Furthermore, we compare SFNet with the prior work on three popular person re-ID benchmarks by abundant experiments.

Person reidentification (re-ID), which aims at recognizing the pedestrians captured by multiple nonoverlapping cameras, has attracted more interest due to its significant and potential application in the Internet of Things like intelligent visual surveillance. However, person reID is still a challenging problem in the situations of various pose, similar appearances, partial occlusion, etc. To handle these obstacles, in this article, we investigate an innovative exciting-inhibition network (EINet) that is a two-branch network composed of the exciting branch and the inhibition branch. The channel-spatial attention block that recalibrates the relationship between channels and highlights features at different spatial positions is used in the exciting branch. A novel Soft Batch DropBlock that randomly selects a continuous region of the intermediate feature maps at the same location is applied in the inhibition branch to inhibit the trivial by an inhibitive mask and reinforce learning the remaining regions. We integrate the comprehensive features from both branches for evaluation and show the performance of EINet intuitively using the visualization method. Abundant experiments demonstrate the state-of-the-art performance by comparing with the previous methods on three popular person re-ID benchmarks. For example, our method obtains 95.64% Rank-1 and 88.75% mean average precision (mAP) on Market-1501, and 77.00% Rank-1 and 74.51% mAP on CUHK03-Detect in the single query mode, respectively.