If artificial intelligence is regarded as a journey from A to B, cloud computing service is an airport or high-speed railway station, and edge computing is a taxi or a shared bicycle. Edge computing is close to the side of people, things, or data sources. It adopts an open platform that integrates storage, computation, network access, and application core capabilities to provide services for users in the vicinity. Compared with centrally deployed cloud computing services, edge computing resolves problems such as long latency and high convergence traffic, providing better support for real-time and bandwidth-demanding services.
The fire of ChatGPT has set off a new wave of AI development, accelerating the sinking of AI into more application areas such as industry, retail, smart homes, smart cities, etc. A large amount of data needs to be stored and computed at the application end, and relying on the cloud alone is no longer able to meet the actual demand, edge computing improves the last kilometer of AI applications. Under the national policy of vigorously developing the digital economy, China's cloud computing has entered a period of inclusive development, edge computing demand has surged, and the integration of cloud edge and end has become an important evolutionary direction in the future.
Edge computing market to grow 36.1% CAGR over the next five-year
The edge computing industry has entered a stage of steady development, as evidenced by the gradual diversification of its service providers, the expanding market size, and the further expansion of application areas. In terms of market size, data from IDC's tracking report shows that the overall market size of edge computing servers in China reached US$3.31 billion in 2021, and the overall market size of edge computing servers in China is expected to grow at a compound annual growth rate of 22.2% from 2020 to 2025. Sullivan forecasts the market size of edge computing in China is expected to reach RMB 250.9 billion in 2027, with a CAGR of 36.1% from 2023 to 2027.
Edge computing eco-industry thrives
Edge computing is currently in the early stage of the outbreak, and the business boundaries in the industry chain are relatively fuzzy. For individual vendors, it is necessary to consider the integration with business scenarios, and it is also necessary to have the ability to adapt to changes in business scenarios from the technical level, and it is also necessary to ensure that there is a high degree of compatibility with hardware equipment, as well as the engineering ability to land projects.
The edge computing industry chain is divided into chip vendors, algorithm vendors, hardware device manufacturers, and solution providers. Chip vendors mostly develop arithmetic chips from end-side to edge-side to cloud-side, and in addition to edge-side chips, they also develop acceleration cards and support software development platforms. Algorithm vendors take computer vision algorithms as the core to build general or customized algorithms, and there are also enterprises that build algorithm malls or training and push platforms. Equipment vendors are actively investing in edge computing products, and the form of edge computing products is constantly enriched, gradually forming a full stack of edge computing products from the chip to the whole machine. Solution providers provide software or software-hardware-integrated solutions for specific industries.
Edge computing industry applications accelerate
In the field of smart city
A comprehensive inspection of urban property is currently commonly used in the mode of manual inspection, and the manual inspection mode has the problems of high time-consuming and labor-intensive costs, process dependence on individuals, poor coverage and inspection frequency, and poor quality control. At the same time the inspection process recorded a huge amount of data, but these data resources have not been transformed into data assets for business empowerment. By applying AI technology to mobile inspection scenarios, the enterprise has created an urban governance AI intelligent inspection vehicle, which adopts technologies such as the Internet of Things, cloud computing, AI algorithms, and carries professional equipment such as high-definition cameras, on-board displays, and AI side servers, and combines the inspection mechanism of "intelligent system + intelligent machine + staff assistance". It promotes the transformation of urban governance from personnel-intensive to mechanical intelligence, from empirical judgment to data analysis, and from passive response to active discovery.
In the field of intelligent construction site
Edge computing-based intelligent construction site solutions apply the deep integration of AI technology to the traditional construction industry safety monitoring work, by placing an edge AI analysis terminal at the construction site, completing the independent research and development of visual AI algorithms based on intelligent video analytics technology, full-time detection of events to be detected (e.g., detecting whether or not to wear a helmet), providing personnel, environment, security and other safety risk point identification and alarm reminder services, and taking the initiative to Identification of unsafe factors, AI intelligent guarding, saving manpower costs, to meet the personnel and property safety management needs of construction sites.
In the field of intelligent transport
Cloud-side-end architecture has become the basic paradigm for the deployment of applications in the intelligent transport industry, with the cloud side responsible for centralized management and part of the data processing, the edge side mainly providing edge-side data analysis and computation decision-making processing, and the end side mainly responsible for the collection of business data.
In specific scenarios such as vehicle-road coordination, holographic intersections, automatic driving, and rail traffic, there are a large number of heterogeneous devices accessed, and these devices require access management, exit management, alarm processing, and operation and maintenance processing. Edge computing can divide and conquer, turn big into small, provide cross-layer protocol conversion functions, achieve unified and stable access, and even collaborative control of heterogeneous data.
In the field of industrial manufacturing
Production Process Optimisation Scenario: Currently, a large number of discrete manufacturing systems are limited by the incompleteness of data, and the overall equipment efficiency and other index data calculations are relatively sloppy, making it difficult to use for efficiency optimization. Edge computing platform based on equipment information model to achieve semantic level manufacturing system horizontal communication and vertical communication, based on real-time data flow processing mechanism to aggregate and analyze a large number of field real-time data, to achieve model-based production line multi-data source information fusion, to provide powerful data support for decision-making in the discrete manufacturing system.
Equipment Predictive Maintenance Scenario: Maintenance of industrial equipment is divided into three types: reparative maintenance, preventive maintenance, and predictive maintenance. Restorative maintenance belongs to ex post facto maintenance, preventive maintenance, and predictive maintenance belong to ex-ante maintenance, the former is based on time, equipment performance, site conditions, and other factors for regular maintenance of equipment, more or less based on human experience, the latter through the collection of sensor data, real-time monitoring of the operating state of the equipment, based on the industrial model of data analysis, and accurately predict when the failure occurs.
Industrial quality inspection scenario: industrial vision inspection field is the first traditional automatic optical inspection (AOI) form into the quality inspection field, but the development of AOI so far, in many defect detection and other complex scenarios, due to the defects of a variety of types, feature extraction is incomplete, adaptive algorithms poor extensibility, the production line is updated frequently, the algorithm migration is not flexible, and other factors, the traditional AOI system has been difficult to meet the development of the production line needs. Therefore, the AI industrial quality inspection algorithm platform represented by deep learning + small sample learning is gradually replacing the traditional visual inspection scheme, and the AI industrial quality inspection platform has gone through two stages of classical machine learning algorithms and deep learning inspection algorithms.
Post time: Oct-08-2023