PRESS RELEASE: 5G has ushered in a smart era of high-speed internet of everything, where the manufacturing industry is looking forward to a digital transformation to reduce costs and improve efficiency. Besides technological innovation, investment is also an important factor. To settle this problem, Dr. Zhao Fuchuan, chief engineer of ZTE 5G transport, introduced a new technology of built-in computing power of the network and computing network convergence to speed up the process with low cost.
Over the past few years, driven by digital transformation, we’ve seen good results of the innovative “5G+ Industrial Internet” model. Some pilots have developed from demonstration to replication. However, some problems become increasingly serious: Traditional small and medium-sized manufacturing enterprises in industrial parks cannot enjoy the benefits brought by 5G technologies as soon as possible due to insufficient funds, weak information infrastructure, and inadequate capabilities of IT staff. Take machine vision quality check in manufacturing for example, small and medium-sized enterprises have no ability to build a dedicated AI vision check network, maintain the system, or continuously upgrade the check models and algorithms. They have to rely on manual check, which leads to high work intensity, high labor costs and low check accuracy. As a result, these enterprises are gradually marginalized or even eliminated in the market competition.
ZTE is always committed to the applications of 5G technologies to support the digital transformation of all industries, trying to deliver benefits to all. To tackle the above-mentioned difficulties, Dr. Zhao proposed a new “Hyper-edge Engine” computing network convergence solution: The network slicing is used to change the centralized on-the-cloud computing center applications into the cloud training + network edge inference, providing a low-cost high-security computing solution for the industrial quality check of small and medium-sized enterprises. The Hype-edge Engine puts quality check samples on the cloud through the slicing network and conducts AI training with the powerful computing power of the cloud to generate a quality check model. Then, it downloads the quality check template via the slicing network to the private line access devices integrated with super-edge AI computing power and completes quality check by real-time inference and calculation, which improves the reliability and security of quality check. The slicing network is built by Ethernet hard slicing FGU (Fine-Granularity Unit) Channel. It offers not only the connection between the private line access devices and the public cloud, but also the connection between the small and medium-sized enterprises in different parks, sharing manufacturing data across parks. The FGU Channel can provide high-quality network services with hard isolation, zero packet loss, and deterministic latency for ToB applications, achieve the maintenance and update of the AI training models, and ensure high-security computing power services for the digital transformation of traditional small and medium-sized enterprises with low cost and easy deployment.
“Simply speaking, it is convergence of cloud and network. By releasing public cloud capabilities, widely deployed access network offers large quantity of edge computing nodes, and hard slicing pipes are used for flexible and secure scheduling of resources of computing nodes.” explained Zhao. The first feature of the Hyper-edge Engine solution is network investment savings. Enterprises do not need to build new computing power centers or redeploy private networks, but only add computing power boards on CPEs (Customer Premises Equipment) to quickly deploy ultra-edge computing power. The second one is high data security. All real-time services in the park are processed locally. Cross-park manufacturing services are connected by secure hard slicing channels to prevent data leakage and assure high security.
ZTE and China Mobile have jointly completed technical verification and solution deployment with industry customers since 2022. The practice results show that the enterprise quality check efficiency is raised fivefold with the investment reduction of 90% in computing power infrastructure construction and 75% in quality check equipments. The other key technology in this solution, FGU Channel hard slicing, has been put into commercial use, and awarded “the Most Innovative Network Slicing Case” at Network X 2022.
“The operator is also enthusiastic about this solution. It enables the operator to make full use of the network technologies and services which it is familiar with, and combine them with AI. It expands services based on the existing market, and has great advantages in network deployment costs and service operation. Therefore, this is undoubtedly an area worth pursuing. ” said Zhao. “Of course, how to replicate solutions to bring technological innovations to more industries is an issue that we need to further explore.”