With the continuous development of intelligent manufacturing technology, the intelligent production line for the Industrial Internet of Things occupies an important position in the field of industrial intelligence. OPC UA is a standard for communication data exchange between intelligent production line devices. OPC UA can establish a unified information model for production line devices and improve the connectivity of heterogeneous networks. However, in industrial wireless network application scenarios, there are various types of sensing information and large amounts of data to be transmitted. Each type of data has different requirements for real-time. The traditional OPC UA communication method is difficult to achieve real-time and reliable transmission in intelligent production lines and can’t meet the transmission requirements of time-sensitive data. In this paper, we propose a real-time communication model of the OPC UA wireless network for intelligent production lines (UAMPDS). IEEE 802.15.4e TSCH is used as the wireless communication infrastructure, and OPC UA protocols are fused to this model. Meanwhile, we design a load-aware time-slot scheduling algorithm to dynamically allocate and schedule time slots according to the network topology and traffic load of each node. We also implement a dynamic preemptive resource scheduling strategy based on service differentiation to ensure real-time transmission of time-sensitive data. The experimental results show that this model achieves differentiated services based on data with different latency tolerance and effectively reduces the transmission latency of real-time service data under the constrained network resources. It can avoid the starvation phenomenon of low priority queues, and improves the overall quality of service of intelligent production line communication networks.

In recent years, self-supervised monocular depth estimation has gained popularity among researchers because it uses only a single camera at a much lower cost than the direct use of laser sensors to acquire depth. Although monocular self-supervised methods can obtain dense depths, the estimation accuracy needs to be further improved for better applications in scenarios such as autonomous driving and robot perception. In this paper, we innovatively combine soft attention and hard attention with two new ideas to improve self-supervised monocular depth estimation: (1) a soft attention module and (2) a hard attention strategy. We integrate the soft attention module in the model architecture to enhance feature extraction in both spatial and channel dimensions, adding only a small number of parameters. Unlike traditional fusion approaches, we use the hard attention strategy to enhance the fusion of generated multi-scale depth predictions. Further experiments demonstrate that our method can achieve the best self-supervised performance both on the standard KITTI benchmark and the Make3D dataset.

Linear Wireless Sensor Networks (LWSNs) are often deployed in unsupervised scenarios such as railways. However, dynamic environments are prone to signal fading and packet loss, resulting in increased transmission delay and decreased throughput. To address this problem, this paper proposes a parallel transmission DCW-CSMA/CA protocol based on the linear topology. The protocol designs a dynamic competition window adjustment algorithm based on node density to reduce multi-hop transmission delay. In addition, under the constraints of single-hop transmission and double-hop interference, DCW-CSMA/CA selects nodes with the highest residual energy in topology groups to perform concurrent data transmission, thereby enhancing network throughput. To ensure fair data transmission among nodes at different depths along multi-hop paths, this paper introduces a dual-queue fair scheduling mechanism. Simulation experiments conducted on the OMNeT++ platform demonstrate that the proposed DCW-CSMA/CA protocol achieves approximately 2.3 times higher throughput than traditional CSMA protocols and exhibits lower transmission latency than three comparison protocols, thus validating its efficiency and practicality in LWSNs.

The monitoring system built on the OPC UA (object linking and embedding (OLE) for process control unified architecture) protocol cannot detect the congestion of each terminal network and the load pressure of the cluster server nodes in real time. In the case that the number of terminal accesses is increasing and the traffic volume is gradually increasing, some server nodes of the system are overloaded or crash. In response to these questions, this paper presents a dynamic control model based on state perception. This model can perceive the communication state of client and server in real time, and uses the current resource occupancy index of the server and the network congestion indicators of each client as the adjustment conditions. The model algorithm can dynamically adjust the client served by the overload node to the server node with small load. Meanwhile, the model uses the congestion estimation method based on bandwidth estimation to accurately estimate the network conditions of each client and improve the validity of state-aware information. The simulation results show that the model can sense the load of each server and make reasonable adjustments to achieve a good load balancing effect. At the same time, the experiment verifies the feasibility and efficiency of the model.