Zhang Honghai and Hu Li: Sino -US technology is almost different in AI technology, but the application of AI in the vertical industry, that is, the industrialization capacity based on AI, has a large gap between China and the United States.
The negotiations between China and the United States are restarted, the negotiations are twisted, and the focus of the game between the two parties and the bargaining chips of the counterparts cannot be separated from 5G, data cross -border flow (AIOT raw materials), cloud computing and other high -tech fields.Recently, it is reported that the cloud computing market may open to the public through the pilot model of the free trade zone, which has also doubled the anxiety of other interests.As a direct beneficiary of 5G landing deployment, and the marginal calculation of the twin brothers who are closely related to cloud computing may also be involved in it, but it will not affect its future quickly standing out in the industry, becoming an important driving force for the digitalization of the country's industry.
With the process of negotiations between China and the United States, the A -share capital market is changing, and 5G has pulled a wave of main markets. This wave of markets quickly sweeps the concepts of flexible screens, the Internet of Things, and smart cities.The edge calculation is no exception. The average increase of 35.14%in the previous quarter (2019Q1) of the edge of the public fund market (2019Q1) was arranged into the TOP 10's most popular concept sector and the stamina was full.The Baidu search index shows that the mobile terminal of the edge computing searches in the past 90 days increased by 457%year -on -year and an increase of 178%month -on -month. Their search absolute value reached its peak in early March, or even more than 5G.
5G and edge computing technology development and deep integration
Talking about the kernel technology calculated 5G and edge computing, the phenomenon must be removed to see the essence. The so -called history is a facial mirror, so the essence of the material is often hidden in its own growth path.Looking back at the development of 1G to 5G, the underlying network infrastructure has continuously experienced revolutionary evolution.From the 1G frequency points (FDMA) simulation modulation method to the time of multiple address (TDMA) and code points (CDMA) at 2G, and then to the thorough guidance of 3G medium code division (CDMA), sign the high -speed IPIn the era of data networks, mobile phones have become a rare thing.However, evolution is still continuing, 4G research and development have been quickly put on the agenda, and the Internet speed increases to 50 times the 3G, achieving high -quality transmission of 3D image.
However, in the face of the diversification, complexity, and professionalization of service scenarios, the scene itself puts forward a three -dimensional challenge to the full dimension of the network's delay, bandwidth, computing and storage, and 5G emergence.According to the standard requirements of the United Nations International Telecommunications Alliance (ITU), the three major application scenarios of enhanced mobile broadband (EMBB), ultra -high reliability and low latency communication (URLLC), and massive machinery (MMTC) can beThe definition of high -challenge quantitative indicators: peak throughput 10Gbps, delayed 1MS, connected 1 million, high -speed mobility 500km/h.It can be seen that it takes large capacity, large bandwidth, Dalian, low latency, and low power consumption as the demand, which not only realizes the user experience of cross -scenario, but also opens the door to magic of all things.
The various applications of 5G in the interconnection scene of all things are difficult to divide with terminal computing problems, and the edge computing technology has undoubtedly carried almost all the terminal computing power.Refer to the edge computing alliance (ECC) and the Industrial Internet Alliance (AII) in the white paper computing in the white paper released by the end of 2018: connecting the bridge between the physics world and the digital world. The edge computing is connected, binding, distributed, fusedThe basic characteristics and attributes such as sexual and data first entrances, and have a significant Cross (that is, the massiveness of the connected mass and the real -time nature of the heterogeneity, the optimization of data, the intelligence of the application, and the protection of security and privacy).
From the perspective of the number of articles calculated by Google academically, the growth trajectory of edge calculation can be divided into three major stages, the stability period, growth period, and stable period.
During the storage period, the CDN network proposed by AKAMAI in 1998 (ie:: through modules such as load balancing and content distribution, the closest cache server of the closest cache server is reduced, thereby reducing network congestion and improving access response.)After experiencing the concept of Cloudlet, the transition from the cache to the function cache is emphasized. It emphasizes the downward to the edge server on the cloud server, and the edge calculation begins to have a basic form.
Since then, technical researchers have begun to pay attention to the upper proximity of the interconnection service functions. Moving edge computing (MEC) has become the most representative.
In the mobile edge computing, the terminal device is considered to have no computing power. The edge server is based on the terminal and cloud computing centers. At the same time, the wireless access network of mobile terminal users can be improved.User experience and become a key technology in 5G development.Subsequently, a internal report of the Northwest National Laboratory in the United States proposed that marginal computing includes both the downlink of cloud service functions, and also included the upward service of all things, and the edge computing terminal urgently needs computing power.
The growth period in 2015 came, marginal computing was hot in the US government, academic, and industrial circles, and domestic development speed was not lagging. Huawei, the Chinese Academy of Sciences, ARM, Intel, etc. established the edge computing industry alliance in Beijing to continuously integrate resources and strengthen cooperation to strengthen cooperation and strengthen cooperation., Guide this area into the deep water area.After 2018, it ushered in a stable period, and the edge computing was pushed to the public. Cloud computing companies, CDN companies, communications operators, equipment manufacturers, and core research institutions have begun to speak, which has also led to the capital boom and traffic we see now.Gathering.
The 5G+AIOT concepts proposed by the two sessions before were also set up a stylish bridge between 5G and edge computing, which also marked the advent of the era of super smart networks.In the future, 5G deployment, supplemented by the coordinated wings of AI and IoT, will increase the number of connected devices, and the amount of data volume of the edges will also present the exponential growth trend.If these data are handled by Yun as the core management platform, there will be problems with real -time, agility, and security privacy, but use terminal edge calculations. You can process massive data nearby.Many problems are solved.From this point of view, 5G has also promoted the development of marginal computing, and the development of the two is always complementary.
5G deployment will be calculated on the edge as an angle
5G combined with edge computing is an important driving force for a new round of scientific and technological revolution and industrial change. It is related to whether my country can seize a major opportunity for the new round of historical scientific and technological revolution and industrial reform.The national strategy has been upgraded from the Internet+to intelligent+, which is intended to further help traditional enterprises transform from Internetization to digital intelligence.5G realizes the help of the floor deployment from top to bottom. The entire underlying scientific and technological environment may be upgraded by large infrastructure and model interactions.Interaction between smart terminals).
It can be imagined that we used to use mobile phones in the past, using 4-8 hours a day, mobile terminals generate 4-8 hours of data information for physical space networks. The interaction between terminals may be 7 days and 24 hours, and 5G undoubtedly accelerates such massive data dataFor information generation and interaction efficiency, the data volume will usher in a new round of large outbreak, and the industry will also move from the Internet to the intelligentization of data -driven.
Compared with the 4G era, the marginal computing market has expanded to trillion levels. Cloud computing companies, CDN companies, communications operators, equipment manufacturers, and core research institutions have entered the venue, and there are hundreds of contention in the market.At this time, the huge assistance of the edge computing for industrial applications has been accompanied by the beginning of 5G. The successful deployment of 5G in the future may quickly promote the industry to enter a standardized high -speed growth period.From another perspective, 5G deployment and commercial use have made the infrastructure all upgraded and the communication network architecture is adjusted simultaneously. More importantly, the intelligent industry application brought by the interconnection of all things will have a blowout effect.
Under the interconnection of all things, edge computing is an important connection tentacle to realize interconnection. The basic characteristics of the fiveG echoing each other and forming a collaboration with each other. When the application of the service industry scenario, there are many advantages in the edge computing.) Distributed and low latency calculations, high efficiency, practicalityThe advantages of time -associated protection can support the autonomous driving, intelligent manufacturing, smart cities (transportation, environmental protection, energy, safety, etc.), AI industry applications and other scenarios, which can realize real -time data processing, analysis, and feedback, and make decision -making guidance in the terminalaction.
Taking intelligent manufacturing as an example, the central intelligent robot is responsible for overall control at the Cloud Computing Center. The assembly line terminal collaboration robot continuously exchanges and analyzes data to achieve real -time decision -making, guiding other terminal robots upstream and downstream of the patent line to carry out coordination in orderprocess.
The ultimate goal of intelligent manufacturing in the future is to achieve intelligent chemical plants. The machine terminal devices have self -perception capabilities, independently complete terminal computing, data processing, and data synchronization and collaboration between terminal equipment, and make optimal decisions to complete complex process operations.Minimize artificial intervention as much as possible.2) Edge computing can help alleviate the pressure of data traffic and reduce data traffic from equipment to cloud. According to statistics, the cost is only 39%of the cloud computing alone.
Its efficient energy -saving traits can support scenes similar to AR/VR, drone, panoramic live broadcast, and autonomous driving to generate massive, multi -dimensional, heterogeneous data in the terminal.Taking autonomous driving as an example, real -time monitoring of the inside and outside of the car must be monitored in real time. During the process, multi -dimensional dynamic detailed data information such as car network data, car owners' behavior data, and road traffic data is huge.Driving requires real -time decision -making on the inside and outside environment and various dimensions of the vehicle, and instruct the vehicle to drive safely or warn the risk of warning in a timely manner to ensure driving safety to the greatest extent.
From the perspective of industrial services and the actual scenario, the vertical industry and specific scenarios under the edge calculation under 5G drive are inseparable from the close collaboration and complement of the cloud computing.Specifically, cloud computing focuses on big data processing and analysis of the global, non -real -time, and long cycle, and marginal computing is suitable for the processing and analysis of locality, real -time, and short -cycle data. It can be said that we are entering a border cloud collaborationThe era.With the further promotion of 5G technology, diversified applications will promote the rapid iteration of edge computing. Traditional data centers will become more and more extended to the edges, and the computing tasks undertaken by the edges will continue to increase. At the same time, compared with the universal serverThe edge computing server can be personalized and differentiated for 5G and edge computing specific scenarios. It has lower energy consumption, wider temperature adaptability, and more convenient operation and maintenance management.
Therefore, it is foreseeable that the technological development trend will transform the centralized computing structure with cloud computing as the core to the decentralized computing architecture with edge computing.
5G and edge computing in the context of the trade war
Behind the Sino -US trade war, the root cause of resource competition is worth exploring, or the incremental market growth of resources driven by technology is weak, or the psychological crisis of the national level of stock resources has faced the exhaustion of the stock and has no effective replacement of resources.Here we might as well verify with bold assumptions.
In recent years, due to the long -term stagnation environment of basic science theory, basic science and technology have become a veritable stock market, and it is difficult for market cakes to continue to grow.Fortunately, the basic science application still maintains innovation and vitality, showing a gesture of all flowers, and faintly can see the shadow of the incremental market.The competition of basic science and technology has always faced a highly high -tech reality. Patent, R & D resources have been long -term and continuous investment, and the reserves and cultivation of professional talents have continuously put forward new challenges.
The breakthrough of China's 5G technology has begun to emerge in the field of basic science and technology in the world. It has also begun to eat this huge stock market cake. This kind of benign competition has more or less stress, anxiety and anxiety to other players in the stock market. The United StatesExcessive reactions are particularly obvious.
As we all know, in the incremental market of basic science applications, China and the United States have long been divided into autumn colors and drove side by side. Take AI applications as an example. In terms of AI technology, such as computer vision, voice technology, and basic hardware areas, the United StatesApplication, that is, there is still a large gap based on AI -based industrialization capabilities, but this gap is not due to the AI technology capabilities itself, but because the domestic industries themselves are backward and insufficient in informationization and automation.The shortcomings of these infrastructures of the industry itself, the maturity and emergence of edge computing, to some extent with 5G landing deployment can help it effectively improve.
It can be foreseeable that 5G will be successfully deployed in the future. Taking AIOT core technology as the power, the mass production data of all things interconnect as the raw material, edge calculation as the terminal tentacles, which can achieve digital transformation and change and iteration of various industries.The steady realization belongs to China's unique industrial 4.0, and ushered in the era of intelligent manufacturing leading the world.
(Zhang Honghai Department of the Ministry of Strategy of JD.com's Ministry of Strategy and Technology Strategic Consultant; Assistant Researcher of the Institute of Information Engineering, the former Chinese Academy of Sciences, and former Baidu senior algorithm engineer. This article only represents the author's own point of view.)
