The Industry 4.0 revolution envisions fully interconnected scenarios in the manufacturing industry to improve the efficiency, quality, and performance of the manufacturing processes. In parallel, the consolidation of 5G technology is providing substantial advances in the world of communication and information technologies. Furthermore, 5G also presents itself as a key enabler to fulfill Industry 4.0 requirements. In this article, the authors first propose a 5G-enabled architecture for Industry 4.0. Smart Networks for Industry (SN4I) is introduced, an experimental facility based on two 5G key-enabling technologies—Network Functions Virtualization (NFV) and Software-Defined Networking (SDN)—which connects the University of the Basque Country’s Aeronautics Advanced Manufacturing Center and Faculty of Engineering in Bilbao. Then, the authors present the deployment of a Wireless Sensor Network (WSN) with strong access control mechanisms into such architecture, enabling secure and flexible Industrial Internet of Things (IIoT) applications. Additionally, the authors demonstrate the implementation of a use case consisting in the monitoring of a broaching process that makes use of machine tools located in the manufacturing center, and of services from the proposed architecture. The authors finally highlight the benefits achieved regarding flexibility, efficiency, and security within the presented scenario and to the manufacturing industry overall.
The rapid emergence of new applications and services, and their increased demand for Quality of Service (QoS), have a significant impact on the development of todays communication networks. As a result, communication networks are constantly evolving towards new architectures, such as the 6th Generation (6G) of communication systems, currently being studied in academic and research environments. One of the most critical aspects of designing communication networks is meeting the restricted delay and packet loss requirements. In this context, although link failure recovery has been widely addressed in the literature, it remains one of the main causes of packet losses and delays in the network. The failure recovery time in currently deployed technologies is still far from the sub-millisecond delay required in 6G networks. The time required for distributed network architectures to converge to a common network state after a link failure is excessive. In contrast, centralized architectures such as Software-Defined Networking (SDN) solve this problem but still need to notify the failure to a centralized controller, which increases the recovery time. This paper proposes a very Fast Failure Recovery (vFFR) strategy that can recover from link failures in sub-millisecond timescales by reacting directly from the data plane of the network devices while maintaining a synchronized state with the centralized controller. We first analyze current failure recovery strategies and classify them according to the techniques used to optimize failure recovery time. Afterward, we describe the design of a vFFR strategy that combines three data plane recovery algorithms to reduce latency and packet loss under varying network conditions. Our vFFR strategy has been modeled in P4 language and tested on an emulation platform to validate the three data plane recovery algorithms under different conditions. The results show that latency varies according to the alternate path selected in the recovery algorithm, and the packet loss rate remains constant even when the background traffic reaches 90% of the link capacity. In addition, the vFFR strategy has been implemented on Intel Tofino devices, achieving a failure recovery time lower than 500 ls and a total frame loss rate below 0.005% in all cases, including those with a 35 Gbps load.
At present, there is growing interest in automated biometric identification applications. For these, it is crucial to have a system capable of accurately identifying a specific group of people while also detecting individuals who do not belong to that group. In face identification models that use Deep Learning (DL) techniques, this context is referred to as Open-Set Recognition (OSR), which is the focus of this work. This scenario presents a substantial challenge for this type of system, as it involves the need to effectively identify unknown individuals who were not part of the system’s training data. In this context, where the accuracy of this type of system is considered crucial, selecting the model to be used in each scenario becomes key. It is within this context that our work arises. Here, we present the results of a rigorous comparative analysis examining the precision of some of the most widely used models today for face identification, specifically some Convolutional Neural Network (CNN) models compared with a Vision Transformer (ViT) model. All models were pre-trained on the same large dataset and evaluated in an OSR scenario. The results show that ViT achieves the highest precision, outperforming CNN baselines and demonstrating better generalization for unknown identities. These findings support recent evidence that ViT is a promising alternative to CNN for this type of application.
With the advent of quantum computing, the increasing threats to security poses a great challenge to communication networks. Recent innovations in this field resulted in promising technologies such as Quantum Key Distribution (QKD), which enables the generation of unconditionally secure keys, establishing secure communications between remote nodes. Additionally, QKD networks enable the interconnection of multi-node architectures, extending the point-to-point nature of QKD. However, due to the limitations of the current state of technology, the scalability of QKD networks remains a challenge toward feasible implementations. When it comes to long-distance implementations, trusted relay nodes partially solve the distance issue through the forwarding of the distributed keys, allowing applications that do not have a direct QKD link to securely share key material. Even though the relay procedure itself has been extensively studied, the establishment of the relaying node path still lacks a solution. This paper proposes an innovative network architecture that solves the challenges of Key Management System (KMS) identification, relay path discovery, and scalability of QKD networks by integrating Software-Defined Networking (SDN) principles, and establishing high-level virtual KMSs (vKMS) in each node and creating a new entity called the Quantum Security Controller (QuSeC). The vKMS serves the end-user key requests, managing the multiple KMSs within the node and abstracting the user from discovering the correct KMS. Additionally, based on the high-level view of the network topology and status, the QuSeC serves the path discovery requests from vKMSs, computing the end-to-end (E2E) relay path and applying security policies. The paper also provides a security analysis of the proposal, identifying the security levels of the architecture and analyzing the core networking security properties, and a performance evaluation that confirms the feasibility and scalability of the proposal.