GSMA Edge Computing Smart Factory PoCProject Roundtable

China Mobile, Hangzhou Mstar, and Huawei showcased their progress on the GSMA Edge Computing smart factory PoC project at the 5G MEC Summit in the 2019 Global Mobile Broadband Forum (GMBBF). The PoC project was initiated in Shanghai during MWC in June, 2019. This is the first time the project has been exposed to public dialogue, and many details were elaborated upon, such as the options for enterprises in the manufacturing industry and for 5G MEC application scenarios.

Project Background

GSMA is committed to promoting the development of the global mobile communications industry. In recent years, GSMA has found a decline in value for connectivity in the global market. How do we "surpass connections" and help carriers increase their revenue? This has become an important question for GSMA. With this in mind, they launched the Edge Computing PoC project using an IoT capability program and invited more than 10 carriers partners from the IoT Programme Steering Group (PSG) to participate. Finally, GSMA chose the Smart Factory project proposed by China Mobile and Huawei, as the subject for the Edge Computing PoC project.

Neill Young (GSMA): What progress has been made since the PoC project was initiated in MWC Shanghai in June?

Geng Liang (China Mobile): In the beginning it was only China Mobile and Huawei participating in the Edge Computing PoC project. After the project was initiated, we immediately started to look for industrial manufacturing enterprises as project implementation partners and corresponding Edge Computing application scenarios. We are lucky to have Haier, a world-famous enterprise, hosting the smart factory for the PoC project, and Hangzhou Mstar, who provides machine vision solutions for Haier as an application partner. Thanks to a joint effort by all four parties, we have made vital progress in the several months since July. We deployed the 5G network in Haier's refrigerator factory, deployed the machine vision application pilot on the 5G MEC platform, and verified the application in Haier's factory. At present, the PoC project is actively preparing for more application cases in its next phase.

Neill Young: What is China Mobile's strategy for MEC?

Geng Liang: China Mobile released its "5G+AICDE" plan at MWC Shanghai in June this year. The "E" in this plan represents Edge Computing, which shows that MEC plays an important role in the 5G strategy of China Mobile. In 2018, China Mobile set up an Edge Computing open lab to focus on the development of "4+1" scenarios. The "4" stands for 4 verticles: smart manufacturing, V2X, smart city, and AR/VR, and the "1" stands for campus. We believe that the campus network will become the first scenario for commercial application of 5G MEC. This is the strategy that led us to choose Haier for the implementation factory of the PoC project. The factory is a typical campus scenario. There are a variety of applications in campus scenarios that can fully utilize the value of MEC, such as manufacturing pipeline, video surveillance, and machine vision.

Neill Young: As a machine vision solution provider for Haier, what does Hangzhou Mstar see as valuable in 5G MEC?

Jay Chou (Hangzhou Mstar): We can address this based on customer requirements. In the manufacturing industry, improving production efficiency and product quality is key for enterprises. Machine vision has been used widely in the past, but each node needed to be equipped with a dedicated computer, occupying space and wasting resources. In addition, each set of equipment is basically isolated, it can cost a lot to test each device, along with data silos, complex line maintenance, complex software upgrade, slow deployment and slow commissioning. However, the machine vision solution can be deployed centrally on the 5G MEC platform with its large bandwidth, low latency, and high performance computing. Working together with the cloud, we can implement many different scenarios, such as OCR or detection of refrigerator doors. Despite this, costs are low and maintenance is simple. The 5G network can be used to collect a high volume of data at high speeds. With this, data can be collected from multiple cameras and factories and aggregated in a 5G MEC node. The node works with the cloud to implement deep learning and self-optimization, to fully "mine" the "gold mine" that is industrial data and improve the accuracy of product analysis. In summary, I believe 5G MEC is revolutionary for the manufacturing industry.

Neill Young: How do you think carriers should develop Edge Computing?

Hu Chunzhe (Huawei): We held many meetings and field visits with Haier after their factory was chosen for implementation in the PoC project. During this process, we found that customer's want several things from MEC architecture: simplicity, collaboration, and efficiency. I think the key to developing Edge Computing for carriers lies in "Connectivity + Computing". Various apps in the smart factory PoC project run on a unified platform, utilizing 5G network capabilities such as high bandwidth, low latency, and guaranteed SLA, combined with high-performance heterogeneous computing. The simplified architecture of centralized O&M and maintenance-free edge nodes provides plug-and-play deployment and simplified O&M. This solves the problems mentioned by Mr. Jay Chou, such as the complex line maintenance, complex software upgrade, and slow deployment and commissioning. In addition, the solution must have high performance computing to be able to integrate as many apps as possible in the edge nodes. Huawei's 5G MEC solution is just developed based on "Connectivity + Computing" and can well meet the requirements of vertical industry customers. I have to say that I was a little nervous at the beginning of the project, but as I was gradually coming to understand customer requirements,I got more and more confident about our MEC solution.

Neill Young: What are some of the challenges for carriers developing Edge Computing?

Geng Liang: 5G MEC is a new technology, but technology is not the most difficult obstacle in its development. The real challenge lies in the business model. Before 5G, the main revenue source for operators was consumer services, where the business model is relatively simple. However, once we enter the vertical industry market, we will see involvement from players from both the downstream and the upstream of the industry chain. The role of operators will change completely, and they will no longer simply be buying and selling. Instead, they will need to play a collaborative role together with customers and industry partners to share the benefit and value of vertical industries. We look forward to demonstrating this through the PoC project.

Neill Young: What challenges will industry partners face with 5G MEC development?

Jay Chou: As Geng Liang said, the development of 5G MEC involves a number of players in the upstream and downstream of the industry chain. As a machine vision solution provider, our applications are deployed on the 5G MEC platform, which is constructed by carriers. We cooperate with other applications on the platform to provide services for enterprises. This requires strong mutual trust and collaboration between all players, only if that can be achieved, 5G MEC will show a better level of development.

Neill Young: As an equipment vendor, what challenges do you think you face with 5G MEC?

Hu Chunzhe: For challenges, we need to look at the customer requirements. First, with 4G, the core network is only deployed at the central site. In the future, the number of 5G MEC nodes will increase by dozens or even hundreds of times, and network deployment and maintenance will increase sharply. 5G MEC must support plug-and-play deployment and maintenance-free edge nodes to minimize this workload. Second, industry partners will be deploying various applications on the 5G MEC platform. How will they quickly deploy applications? This would really help carriers and enterprise customers build agile services. I believe that code-free API integration is necessary to cope with this challenge.

After the four speakers shared the progress made in the PoC project, and the challenges faced, they discussed other 5G MEC application scenarios with great interest. Among those scenarios, machine vision is widely regarded as a highlight because of its value as a ubiquitous application. We also hope that PoC can become an example for replicating machine vision in more 5G MEC scenarios.

Editor's note: The next phase of the GSMA Edge Computing Smart Factory PoC project will focus on the deployment of application cases other than machine vision on the 5G MEC platform and further explore the business model. The result and further progress will be made public at MWC20. Let's wait and see.