Basic Concepts and Development of Edge Computing

Edge computing refers to the use of network, computing, storage, and application core capabilities as an open platform close to the source or data source, providing nearest-end services nearby. Its applications are launched on the edge, resulting in faster network service response and meeting the industry's basic needs in real-time business, application intelligence, security and privacy protection. Edge calculations are between physical entities and industrial connections or at the top of physical entities. The cloud computing can still access the historical data of edge calculations.

Edge computing is not a new word. As a content distribution network CDN and provider of cloud services AKAMAI, as early as 2003 in cooperation with IBM "edge computing." As one of the largest distributed computing service providers in the world, it was responsible for 15-30% of global network traffic. In an internal research project, he proposed the purpose of "edge computing" and solved problems, and provided Edge Edge-based services through AKAMAI and IBM on its WebSphere.

For the Internet of Things, a breakthrough in edge computing technology means that many controls will be implemented through local devices without having to be delivered to the cloud, and the processing will be completed at the local edge computing layer. This will undoubtedly greatly increase processing efficiency and reduce the load on the cloud. Because it is closer to the user, it can also provide faster response for the user and solve the demand at the edge.

First, the technical architecture

In China, the Edge Computing Alliance ECC is working hard to promote the convergence of three technologies, namely the integration of OICT (Operational, Information, and Communication Technology). The object of calculation is mainly defined in four areas. The first is the problem of the equipment domain. The pure IoT devices that appear in [1] are different but overlap with the automated I/O collection. section. Those data that can be directly used to optimize at the top level and do not participate in the control itself can be processed directly on the edge side; the second is the network domain. At the transmission level, the direct end IoT data and the data from the automated production line will have different transmission methods, mechanisms, and protocols. Therefore, the data standard problem of transmission must be solved here. Of course, it can be directly accessed under the OPC UA architecture. The underlying automation data, but for the interaction of Web data, there will be coordination between IT and OT problems, although there are some leading automation companies have provided mechanisms for Web-based data transmission, but most of the live The data still has these problems. The third is the problem of data fields, data storage, format, and other data fields that need to be solved after data transmission, as well as data query and data exchange mechanisms and tactical issues that are considered in this field.

The last one is also the most difficult application domain. This may be the most difficult problem to solve. The application model for this field has not yet had more practical applications.

The definition of the reference architecture of edge computing alliance ECC for edge computing includes equipment, network, data and application four domains. The platform provider mainly provides hardware and software foundations for network interconnection (including bus), computing capability, data storage and application. facility.

From the perspective of industrial value chain integration, ECC proposed CROSS, which is based on agile connection, real-time, data optimization, application intelligence, security and privacy. Security brings value and opportunities to users at the edge of the network, which is the focus of the alliance members.

Second, the nature of the calculation

Automation is in fact a "control" as the core. Control is based on "signals," while "calculation" is based on data. More meaning means "strategy" and "planning." Therefore, it focuses more on "scheduling, optimization, routing." Just as with the nationwide high-speed rail dispatching system, every additional trip reduction will trigger the adjustment of the dispatching system. It is based on time and node scheduling and planning issues. The application of edge computing in the industrial field is more of this type of "calculation."

Simply put, traditional automatic control is based on signal control, while edge computing can be understood as "information-based control."

It is noteworthy that although edge calculations and fog calculations are low-latency, the 50mS and 100mS cycles are still used for "control tasks" such as 100μs for high-precision machine tools, robots, and high-speed graphic printing systems. It is a very large delay, and edge computing is called "real-time", and from the perspective of the automation industry - unfortunately, it is still categorized as "non real-time" applications.

Third, the division of labor

In the industrial field, edge application scenarios include energy analysis, logistics planning, and process optimization analysis. In terms of production task assignments, optimal equipment scheduling is required for production based on production orders. This is the basic task unit for APS or generalized MES and requires a lot of calculations. These calculations depend on the software platform of a specific MES vendor or the "edge computing" platform - an analysis platform built on the basis of Web technology. There will not be much difference in the future. In a sense, the MES system itself is a traditional architecture, and its core can be either in a dedicated software system or in the cloud, fog, or edge. In this application scenario, the overall division of labor in the entire smart manufacturing and industrial IoT applications is as follows.

Automation vendors provide "acquisition," which includes the role of data sources, which utilizes native "information" such as machine production, status, and quality that automation has produced in distributed I/O acquisition, bus interconnect, and control machines.

ICT vendors provide "transmission" to achieve industrial connectivity. Because in terms of how to provide data transmission, storage, and computing, ICT vendors have their traditional advantages, including cost, and have the advantage of a cloud platform.

The business experience and knowledge of traditional industrial companies provide the basis for analysis of software (independent or internal) manufacturers. Understanding these business processes is still essential. The coordination of the industry chain and the ultimate goal are still the core issues of "quality, cost, and delivery."

Four, three types

1) Personal Edge (PersonaLEDge)

This edge computing is around us personally, sometimes at our side, at our home; it includes home robots, smart glasses, smart tablets, and sensors under your skin, watches, home automation systems, yours Amazon Echo (echo) and smart phones.

The Personal Edge is generally mobile. As we move between the home and the workplace, Personal Edge computing devices enter the Business Edge area.

With the proliferation of smart home devices, digital health and other personal devices, we will hear more about personal edge computing in the next five years.

2) Business Edge (Business Edge)

This is the type of edge computing that is most concerned. Machines and people connected at the business edge (Business Edge) converge here. Business Edge can be found in our carpeted offices, in areas without carpets, or even in open areas where we work and play.

Many discussions on the Internet of Things seem to assume that this is the only margin, and each IoT discussion expresses the benefits of this edge computing. Mission-critical SPA ("sensing-processing-acting") has a strong tendency to focus on development in this area, especially in the industrial IoT field.

Many vendors are providing development environments for such applications to help customers develop edge applications and analytics. Amazon LambdaGreengrass (https://aws.amazon.com/lambda/) and AzureIoTHub are examples of such software.

Note 1: Amazon LambdaGreengrass is Amazon's cloud computing service, through AWSLambda, you can run the code without configuring or managing the server. You only have to pay for the amount of time you spend - no charges are generated when the code is not running. With Lambda, you can run code for almost any type of application or back-end service, and you don't need to manage it all. Just upload your code and Lambda will handle everything you need to run and extend high availability code. You can set your code to be triggered automatically from other AWS services or directly from any web or mobile application.

Note 2: AzureIoTHub is Microsoft's cloud computing for Internet of Things (IoT) applications. AzureIoTHub is an IoT center and is a cloud service that provides registration, management, and communication for IoT devices. It is an important part of Microsoft Azure IoTSuite and an important foundation of Microsoft's Internet of Things strategy. Microsoft Azure IoTHub can be used to manage billions of IoT devices, provide two-way communication support between cloud and devices, handle trillions of information per month, and simplify integration with other Azure services, including Azure Machine Learning And Azure flow analysis.

3) Cloudy Edge

This is the least discussed type of edge computing, but it is the oldest type of edge. Cloudy Edge is a topological term for the edge of a service provider or enterprise network where traffic is first accessed from a dial-up modem into a home or remote branch office.

The CloudyEdge was once a network edge with no computing power. They are called PoP (points-of-presence).

The need for application performance and content delivery requires the addition of application and data processing capabilities at the network edge. Modern edge data centers (EdgeDataCenters) can meet this need. The Content Delivery Network (CDN) is using them so that we can get better page and video loading effects. Mobile Edge Computing (Mobile Edge Computing) enhances this edge advantage because people need better mobile app performance.

So the old PoP has no future in content and computing. The SP edge, mobile edge, and enterprise edge together form the cloud. This kind of edge is still related to ensuring application performance and smooth content delivery.