As the earliest team to study edge computing in-depth, the China Construction Investment Communications team released an in-depth report "Mobile Edge Computing, Standing at the "Center" of
Objectively speaking, edge computing has indeed entered the industrialization stage, but it is still in its early stages. We believe that the core that restricts the development of edge computing is not technology, but application requirements and business models. After all, edge computing emphasizes more on computing, storage and distribution close to users. It can be achieved by "building edge computer room DCs, deploying edge servers and other hardware, and strengthening edge and cloud collaboration". But precisely because of this, additional fixed asset investment urgently needs new application scenarios and the derived model of paying extra for edge computing, otherwise edge computing will be a castle in the air. However, 5G has entered the commercial sprint stage, and future applications may bloom. Scenarios such as autonomous driving, industrial control, and live broadcasting have a strong demand for edge computing. Therefore, with the development of edge computing, investment opportunities will arise.
Core viewpoint
Mobile Edge Computing (MEC) is a "hardware + software" system that provides IT service environment and cloud computing capabilities at the edge of the mobile network to reduce the latency of network operations and service delivery. Its technical characteristics mainly include "proximity, low latency, high broadband and location awareness", and will have broad application prospects in the future, such as Internet of Vehicles (such as unmanned driving), AR, video optimization acceleration, surveillance video analysis, etc. IDC forecasts show that 40% of the data will be analyzed, computed and stored on the edge of the network in 2018.
The relationship between mobile edge computing and cloud computing can be compared to the relationship between the local offices of a group company and the group headquarters, which can complement each other. Cloud computing grasps the overall situation and focuses on the analysis of non-real-time and long-term data, and can play its strengths in fields such as periodic maintenance; while mobile edge computing focuses on the local area, focusing on the analysis of real-time and short-term data, which can better support the real-time intelligent processing and execution of local businesses.
Compared with CDN, mobile edge computing is closer to the edge of the wireless network, with a deeper sinking position, so the delay is smaller; the focus of CDN application scenarios is "distribution acceleration", while mobile edge computing not only needs to "accelerate", but also has open API capabilities and localized computing storage capabilities, which can make the network intelligent. Therefore, traditional CDN is an IO-intensive system centered on cache distribution services, and one of the future evolution directions is edge computing.
First, the three major application scenarios of 5G and the delay indicators less than 1ms determine that the end points of 5G services cannot be all on the cloud platform of the backend of the core network, so the development of mobile edge computing is necessary. Second, the core of the Internet of Things is to connect everything. With the rapid growth of the number of connections, on the one hand, it means the generation of massive data . On the other hand, IoT devices often require intelligent computing. Mobile edge computing can help the Internet of Things better realize sensing, interaction and control between things through data processing capabilities closer to the edge. Third, SDN will help the development of mobile edge computing. For example, the SDN architecture allows the network to flexibly use cloud computing and edge computing resources, meet agile and dynamic system needs, and provide users with the best service.
Mobile edge computing was born in 2013 and is still in the process of technological research and development and industrialization. Although it is still in the early stages of development, as one of the core technologies of 5G, it has broad development prospects. Data shows that by increasing the deployment of edge cloud servers, operators can reduce proprietary network deployments and save more than 35% of the backhaul line usage between wireless access networks and existing application servers. Therefore, giants have made plans, including Nokia , Intel , Huawei , ZTE , etc. From an investment perspective, we recommend focusing on four directions. First, edge computing device manufacturers, such as servers and gateways, such as Unigroup Co., Ltd. ( New H3H3 ), Inspur Information, ZTE , Rihai Intelligence, Sites, etc.; second, the opportunity for CDN service providers to deploy edge computing, such as Netsu Technology ; third, edge computer room design and construction providers, such as data ports; fourth, cloud security manufacturers, cloud computing has been sinking to the edge, and the importance of data security and network security will be self-evident. You can pay attention to Shenxinshui, China News Service, and Hengwei Technology.
We believe that mobile edge computing is expected to develop together with 5G. Superimposed edge computing emphasizes getting closer to users and paying more attention to the richness of edge nodes, so it will be mainly deployed in the operator network in the early stage. Therefore, companies with first-mover advantages and good cooperation with operators may be more likely to stand out, such as ZTE, Rihai Intelligence, NetEase Technology, etc.
Risk warning
Mobile edge computing technology development is not as expected, and the business model is unclear.
-
What is mobile edge computing?
Mobile edge computing (MEC) was originally launched in 2013 on a computing platform jointly launched by IBM and Nokia Siemens. Afterwards, major telecommunications standards organizations began to promote the standardization of mobile edge computing. According to the European Telecommunication Standards Association (ETSI), mobile edge computing focuses on providing IT service environments and cloud computing capabilities at the edge of the mobile network, emphasizing proximity to mobile users to reduce latency in network operations and service delivery.
In 2016, Huawei initiated the "Edge Computing Industry Alliance" in the country. According to the definition of the edge computing industry alliance, edge computing is an open platform that integrates the core capabilities of network, computing, storage, and application at the edge of the network near the object or data source, and provides edge intelligent services nearby to meet the key needs of industry digitalization in agile connection, real-time business, data optimization, application intelligence, security and privacy protection.
We believe that mobile edge computing provides content storage computing and distribution services near the mobile user side through in-depth cooperation with content providers and application developers, so that applications, services and content can be deployed in a highly distributed environment to better meet the needs of low latency and high bandwidth .
According to Intel's architecture, mobile edge computing is located between the wireless access point and the wired network. The traditional wireless access network has the conditions for service localization and close deployment, thus providing high bandwidth and low latency transmission capabilities. At the same time, the service surface sinks to form localized deployment, which can effectively reduce the requirements for network backhaul bandwidth and network load. Mobile edge computing provides application programming interfaces (APIs) and opens basic network capabilities to third parties, so that the network can be customized and interact on demand according to third parties' business needs. This will be the first step for 5G to move towards a flatter network.
The technical characteristics of mobile edge computing are mainly reflected in: proximity, low latency, high broadband and location cognition.
(1) Proximity: Since the mobile edge computing server is arranged very close to the information source, edge computing is particularly suitable for capturing and analyzing key information in big data. In addition, edge computing can directly access the device, so it is easy to directly derive specific commercial applications.
(2) Low latency: Because the mobile edge computing service is close to the terminal device or runs directly on the terminal device, the latency is greatly reduced.This makes feedback faster, while also improving the user experience, greatly reducing the congestion that may occur in other parts of the network.
(3) High bandwidth: Since the mobile edge computing server is close to the information source, it can simply process data locally without uploading all data or information to the cloud. This will reduce the transmission pressure of the core network, reduce network blockage, and greatly increase the network rate.
(4) Location awareness: When the network edge is part of a wireless network, whether it is WiFi or cellular, local services can use relatively little information to determine the specific location of each connected device.
The basic components of mobile edge computing include: routing subsystem, capability open subsystem, platform management subsystem and edge cloud infrastructure. The first three subsystems are deployed in mobile edge computing servers, while the edge cloud infrastructure is composed of small or micro data centers deployed at the edge of the network.
The core device of the mobile edge computing system is a MEC server built on the IT general hardware platform. mobile edge computing system provides localized public cloud services by deploying cloud computing facilities (i.e. edge cloud) at the edge of wireless base stations or wireless access networks, and can connect to private clouds within other networks (such as enterprise networks) to realize hybrid cloud services. The mobile edge computing system provides a virtualized environment based on a cloud platform, allowing third-party applications to run on virtual machines (VMs) in the edge cloud. Relevant wireless network capabilities can be opened to third-party applications through platform middleware on MEC servers.
The relationship between mobile edge computing and cloud computing can be compared to the relationship between the local offices of the group company and the group headquarters. Cloud computing grasps the overall situation and focuses on big data analysis of non-real-time and long-term data, and can play its expertise in cyclical maintenance and business decision-making support; edge computing focuses on local and focuses on the analysis of real-time and short-term data, which can better support the real-time intelligent processing and execution of local businesses.
For the timeliness of data processing, if cloud computing is completely relied on, the transmission time and feedback time will greatly reduce the data processing efficiency. If you first perform a simple and preliminary processing through mobile edge computing, and then upload complex data to the cloud, and solve it through cloud computing, this can not only solve the timeliness of data processing, but also reduce the transmission cost, but also reduce the pressure on cloud computing. Therefore, the operating mode of cloud computing and mobile edge computing is as follows: the edge end first preprocesses the data, extracts features and transmits them to the cloud and then performs calculation and analysis.
CDN is a content distribution network. Its purpose is to publish the content of the website to the "edge" closest to the user by adding a new network architecture to the existing Internet, so that users can obtain the required content nearby, so as to improve the response speed of users to access the website.
There is a close connection between CDN and mobile edge computing.
CDN has many similarities and mobile edge computing, and there are similarities in achieving goals. Both are generated against the backdrop of the continuous increase in user experience requirements and the surge in the number of users and data traffic. The meaning of network "edge" in CDN and "edge" in mobile edge computing both mean that different from previous network architectures, the server is closer to the wireless access network (RAN). However, compared with CDN, mobile edge computing is closer to the wireless access network and sinks deeper. Due to the reduction of physical distance, natural mobile edge computing further reduces latency compared to CDN.
But in terms of architecture, mobile edge computing is quite different from CDN. The typical architecture of mobile edge computing includes capability open systems and edge cloud infrastructure, which makes mobile edge computing have open API capabilities and localized computing power, which is exactly what CDN lacks.
Due to its own technical characteristics, the focus of CDN application scenarios is on "acceleration", such as website acceleration, video on demand and live video broadcast, and no intelligent scenarios appear. Mobile edge computing includes computing power, so it has low latency and intelligence characteristics. In addition to application scenarios including CDN, mobile edge computing will play a very important role in application scenarios such as Internet of Vehicles and smart medical care that require intelligence.
With the continuous advancement of technology and the increasing changes in the industrial environment, users' requirements for high frequency and high interaction are becoming more and more extreme, not only the requirements for delays, but also the ability to intelligent allocation and processing and calculating massive data. Therefore, traditional application scenarios of CDN such as video acceleration will be challenged. In this regard, CDN needs to make further upgrades based on market demand, such as intelligence, and the most important thing is intelligent provisioning and intelligent computing. In terms of application scenarios, CDN should also be continuously upgraded, from the initial image acceleration, website acceleration, and video acceleration, to carrying various high-definition video, VR/AR and other heavy applications, to the carrying of big data technology, the Internet of Things, and artificial intelligence. And these are exactly the problems that mobile edge computing needs to solve.
Therefore, traditional CDN is an IO-intensive system centered on cache services. One of the evolution directions of future CDN is to form an edge computing system.
2
Multiple factors promote the accelerated development of mobile edge computing
The core of the Internet of Things is to enable the Internet of Things to connect and operate intelligently. Edge computing can help the Internet of Things better realize sensing, interaction and control between objects through data analysis and processing capabilities closer to the edge. "Mobile edge computing" is an architecture that extends computing, network and storage capabilities from the cloud to the edge of the Internet of Things network, following the model of "business applications at the edge and management at the cloud".
Currently, various smart devices with sensors are rapidly connected to the Internet. IDC statistics show that by 2020, more than 50 billion terminals and devices will be connected to the Internet. We expect that the number of Internet of Things connections in my country will be about 840 million in 2016, and is expected to increase by 317% in 2020 to reach 3.5 billion.
The rapid growth of the number of connections means the generation of massive data on the one hand, and on the other hand, such connected devices often require intelligent calculations. According to IDC's estimates, 40% of the data will need to be analyzed, processed and stored at the edge of the network in 2018.
The problems caused by massive data are inconvenient storage and hysteresis of calculation results. Cloud computing is one of the solutions to this problem. When facing such a huge amount of data, cloud computing can provide storage and computing support for big data. However, if a large amount of data generated by the Internet of Things is completely processed by cloud computing, then all the data generated on the edge of the network need to be uploaded to the cloud through the network. Not only will the transmission time be very long, but the transmission cost is also very high. More importantly, since the data is first uploaded to the cloud and then fed back to the terminal for execution, the data processing efficiency will be greatly reduced.
Taking intelligent driving as an example, when monitoring obstacles in front of the car, if you cannot process them intelligently in a timely manner and control the direction to avoid obstacles, but first transmit them to the cloud and then feedback them back to the terminal, extremely small delays may lead to car accidents.
Such a large number of devices require intelligent computing, and it is difficult to complete by relying solely on cloud computing. Therefore, in the face of the large number of connections and large amounts of data generated in the future Internet of Things era, we need to reconsider the network layout. For example, a short video of an internet celebrity is about 10MB. If 1,000 people in a region watch this video, 10GB of network traffic will be generated. During this process, the video content was actually sent repeatedly from the Internet to the mobile network 1,000 times, and 99.9% of the network bandwidth was wasted. If the video was cached at nodes close to the edge, a lot of bandwidth would be saved.
The data characteristics of the Internet of Things are diversity, heterogeneity, massiveness and high growth. Therefore, the screening and timely processing of data poses a challenge to the current network architecture.According to the survey results of the International Telecommunication Union (ITU), in the era of the Internet of Things, data processing efficiency and effective information crawling are the main problems faced by users, with 44% and 36% of the respondents respectively thinking that the amount of data is too large and the difficulty in crawling effective information is the main problems.
The traditional view believes that solving data diversity and heterogeneity should start with basic software. Different micro devices may require different operating systems, different perceived information requires different data structures and databases, and different systems need to use different middleware. The correct choice of these three systems can block the heterogeneity of the data. But taking this approach, the cost expenditure will be huge. Mobile edge computing can first filter the data and upload the filtered data to the cloud, thereby achieving the smooth transmission, filtering and fusion of data, which is of great significance to timely and correctly perception of data.
For the massive and high growth issues of IoT data, if you directly build more and larger data centers, it will greatly increase management costs and reduce system reliability. Mobile edge computing is a small information center very close to the terminal information source, pushing applications, processing and storage to the mobile boundary, so that massive amounts of data can be processed normally without having to completely build more data centers.
5G technology takes "large capacity, large bandwidth, large connection, low latency, and low power consumption". According to the UN International Telecommunication Union (ITU) standard requirements for 5G, the 5G standard includes three application scenarios: enhanced mobile broadband (eMBB), ultra-high reliability low-latency communication (URLLC), and massive machine communication (mMTC), and defines the following key indicators: peak throughput rate of 10Gbps, delay of 1ms, connection number of 1 million, and high-speed mobility of 500km/h.
In the current network architecture, due to the high-location deployment of the core network, the transmission delay is relatively large, which cannot meet the ultra-low latency business needs; in addition, the complete termination of services in the cloud is not completely effective, especially some regional services do not end locally, which not only wastes bandwidth but also increases latency. Therefore, the delay indicator and the number of connection indicators determine that the end points of 5G services cannot be all on the cloud platform on the backend of the core network.
Mobile edge computing just meets this requirement. On the one hand, mobile edge computing is deployed at the edge location, and edge services run on terminal devices, and feedback is faster, solving the delay problem; on the other hand, mobile edge computing reduces content and computing capabilities, provides intelligent traffic scheduling, localizes services, and caches content locally, so that some regional businesses do not have to end in the cloud.
In addition, mobile edge computing is closely related to network slicing technology, C/U separation, etc. in 5G technology.
Network slicing is considered by many well-known operators and equipment manufacturers to be an ideal network architecture in the 5G era.
Since mobile networks need to serve various types and needs, if a proprietary network is built for each service, the cost will be difficult to estimate. Network slicing technology allows operators to split multiple virtual end-to-end networks based on a hardware infrastructure. Each network slicing is logically isolated from the device to the access network to the transmission network and then to the core network, adapting to the different characteristic needs of various types of services, ensuring that from the core network to the access network, including terminals and other links, it can dynamically, real-time and effectively allocate network resources, thereby ensuring the quality of quality, delay, speed, bandwidth, etc.
The business perception function of mobile edge computing is similar to network slicing technology to a certain extent. One of the main technical features of mobile edge computing is low latency, which allows mobile edge computing to support the most demanding business types, which also means that mobile edge computing is a key technology in ultra-low latency slicing.With the application of mobile edge computing, the connotation of network slicing technology will be expanded from simply splitting multiple virtual end-to-end networks to splitting virtual end-to-end networks with different high-demand time delays, which will help the development of 5G network slicing technology.
In the 5G era, mobile networks are facing exponentially growing traffic demand, so it has become a way to expand network capacity by using higher frequencies with a wider spectrum. However, high frequency bands are prone to severe propagation losses compared to lower frequency bands, and to solve this problem, operators generally place cells operating in higher frequency bands within cell coverage of lower frequency bands. However, as deployment becomes increasingly dense, the coverage of a single cell in the ultra-intensive networking scenario is smaller, which will lead to frequent switching of end users with higher mobile speeds, resulting in a significant decline in user experience. At the same time, such frequent handover will cause huge redundant control signaling interactions and reduce the efficiency of heterogeneous networks. In order to solve this problem, C/U separation technology is proposed.
C/U separation (transformation and control separation) technology refers to centralizing control functions from the perspective of network reconstruction, centralizing control surfaces from the perspective of architectural design, and further simplifying the user surface or forwarding surfaces to reduce costs and improve efficiency.
In C/U separation technology, the user plane gateway can be independently sink to the mobile edge due to the service being moved down, and the mobile edge computing will always exist in terms of the traffic billing function and security guarantee requirements. C/U separation technology can solve this problem and help the development of mobile edge computing. It is worth mentioning that the vBRAS innovation plan based on C/U separation technology, which cooperated with China Mobile Research Institute and ZTE, won the "2017 GTB Infrastructure Innovation Award", which fully demonstrates the industry's recognition of the concept of C/U separation technology.
The "low power consumption and large connection" requirements in one of the three major application scenarios of 5G can provide support capabilities with network connections of over 100 billion, and meet the density index requirements of 1 million/km2. Under such massive data and high connection density indicators, it is very important to ensure low latency and low power consumption. 5G even proposed a business goal of 1ms end-to-end delay to support the needs of business such as industrial control. To achieve low latency and low power consumption, on the one hand, it is necessary to significantly reduce the air-interface transmission delay, and on the other hand, it is necessary to minimize the forwarding nodes and shorten the "distance" between the source and the destination node.
. The current mobile technology does not optimize the delay sufficiently. LTE technology can increase the air interface throughput by 10 times, but can only optimize the end-to-end delay by 3 times. The reason is that when the air interface efficiency is greatly improved, the network architecture is not fully optimized and becomes a bottleneck for business delay. Although the LTE network implements a flat structure of 2 hops, the base stations are often hundreds of kilometers away from the core network, and the multiple convergence and forwarding equipment are used, coupled with unpredictable congestion and jitter, which cannot achieve low latency guarantees at all.
Mobile edge computing is deployed at the mobile edge, which will effectively integrate the technologies of wireless network and the Internet, and add computing, storage, processing and other functions to the wireless network side to build a mobile edge cloud, providing information technology service environment and cloud computing capabilities. Since application services and content are deployed at the mobile edge, this can reduce forwarding and processing time in data transmission, reduce end-to-end delay, meet low-latency requirements, and reduce power consumption.
Currently, traditional operator networks are "dumb pipelines" and are not intelligent. In the context that communication networks are carrying more multimedia applications based on new smart terminals and IP-based, operator fees and business models are relatively single, and they lack control over services and users. For example, there are currently a large number of monthly packages, which is difficult to meet the differentiated needs of users. Under certain conditions, users who use less traffic are actually subsidizing users who use high traffic.In addition, since the service is not prioritized, many services that occupy a large amount of bandwidth cannot generate sufficient value, such as some video streaming media, P2P services, etc., while some businesses with high real-time requirements and high value, such as mobile office services, cannot obtain priority protection.
Faced with this challenge, operators have proposed the "smart pipeline" strategy. According to Ericsson's definition, the broad definition of intelligent pipeline is: a data pipeline that allocates reasonable network resources based on customer value and business value and provides corresponding billing means. The key to realizing "smart pipeline" lies in accurately distinguishing user categories and truly grasping user needs. To achieve this goal, some operators have begun to use the URL information obtained from deep packet analysis to match key fields, and how to perceive user needs and portray customers.
As analyzed above, one of the important features of an intelligent 5G network is content perception. Through the content analysis of network traffic, the network's business stickiness, user stickiness and data stickiness can be increased. One of the key technologies of mobile edge computing is business and user perception. By identifying services and users at the mobile edge, fully optimize and utilize local network resources, improve network service quality, and provide users with differentiated services, bringing a better user experience.
Among domestic operators, China Unicom and China Mobile are active promoters of mobile edge computing. China Mobile and China Unicom jointly conducted relevant tests with the company, and China Mobile even released relevant plans. China Mobile also used MEC equipment to deploy in the F1 event venue in Shanghai. According to actual measured data, the live broadcast time is only 0.5 seconds, and users can hardly feel it. If you use the traditional live broadcast method now, put the server on the Internet and then transmit the length stream to the scene through the network, the delay is about 50 seconds, so the experience given to the user is a very huge improvement. This should be seen that localized service provision can indeed improve the user's experience well.
SDN is a new network innovative architecture and an implementation method of network virtualization. It converts hardware-intensive traditional networks into software-driven new networks, which can be fully programmed, and can simplify operations and quickly realize new service delivery. Mobile edge computing platforms can provide application programming interfaces (APIs) and open basic capabilities to third parties, which is consistent with the SDN concept.
In fact, with the increase in the use of mobile terminals, it has put huge pressure on the cloud computing network, and this situation will only worsen with the increase in the use of mobile devices around the world. Overload resources and latency will lead to a decline in the end user experience, and creating a unified system for cloud and edge computing resources is an effective way to deal with overload resources and latency challenges. However, the unification of resource systems for cloud and edge computing is also faced with challenges. There must be a local coordinator to allocate resources for tasks in a dynamic and unpredictable environment. The system must be updated in real time to provide the best information about available resources and have an open programmable interface to complete tasks in the most efficient way.
Research has found that creating an architecture that supports software-defined networks (SDNs) can effectively address these challenges. SDN can provide flexible and reliable real-time information of available resources, and a centralized controller enables the best decision-making of each unit in the overall system; using the SDN architecture will enable the network to interchange the resources of cloud computing and edge computing to meet the needs of agile and dynamic systems and provide users with the best service.
3
Mobile edge computing has rich application scenarios
Because mobile edge computing has technical characteristics such as high bandwidth, low latency and position perception, the application scenarios are very rich. Such as video optimization acceleration, vehicle networking, AR and surveillance video analysis are typical application scenarios of mobile edge computing.
In recent years, with the increase in network speed, video traffic has grown very rapidly. According to Cisco, global video traffic has increased from 13,483PB per month in 2012 to 46,237PB in 2017, an increase of nearly 2.5 times. As 5G commercialization approaches, network speed will further increase, which will greatly stimulate video traffic. According to Cisco's forecast, mobile video will grow 8.7 times from 2016 to 2021, enjoying the highest growth rate in the mobile app category. By 2021, mobile video will account for 78% of total mobile traffic.
When mobile video traffic is increasing rapidly, network delays greatly reduce the view of mobile video audiences. Mobile video stagnation and buffering remain a big problem for operators and their customers. In the United States, 69% of viewers have experienced various levels of network delay in watching mobile videos.
In the case where network congestion seriously affects the visual perception of mobile videos, mobile edge computing is a good solution.
(1) Local cache. Since the mobile edge computing server is a memory close to the wireless side, the content can be cached on the mobile edge computing server in advance. When there is a need to watch mobile videos, that is, the user initiates a content request, and the mobile edge computing server immediately checks whether there is any content requested by the user in the local cache, and if there is, it will serve directly; if there is no, it will go to the network service provider to obtain it and cache it locally. The next time other users have this type of demand, they can provide services directly. This reduces the request time and solves the problem of network congestion.
(2) Cross-layer video optimization. The cross-layer here refers to the interactive feedback of "upper and lower layer" information. By sensing the throughput of the lower layer of wireless physical layer, the server (upper layer) decides to send videos of different quality, clarity, etc. to users, reducing network congestion while improving line utilization, thereby improving user experience.
(3) User perception. Due to the service and user perception characteristics of mobile edge computing, customers with different needs can be distinguished, different service levels can be determined, differentiated wireless resource allocation and packet delay guarantee for users, and reasonable allocation of network resources to improve the overall user experience.
According to the definition of the Internet of Vehicles Industry Technology Innovation Strategic Alliance, Internet of Vehicles is a large system network that conducts wireless communication and information exchange between vehicles-X (X: vehicles, roads, pedestrians and the Internet, etc.) based on the agreed communication protocol and data interaction standards. It is an integrated network that can realize intelligent traffic management, intelligent dynamic information services and intelligent vehicle control. It is a typical application of Internet of Things technology in the field of transportation systems.
The prerequisite for implementing the above functions is to intelligently process the massive data collected by the Internet of Vehicles. The Internet of Vehicles has special requirements for data processing: First, low latency. When the vehicle is moving at high speed, the collision warning function should be realized, and the communication delay should be within a few ms; second, high reliability. Due to safe driving requirements, compared with ordinary communication, the Internet of Vehicles requires higher reliability. At the same time, since the vehicle is moving at high speed, the signal needs to achieve high reliability on the basis of being able to support high speed movement.
As the number of connected vehicles increases, the amount of data in the Internet of Vehicles will become larger and higher, and the requirements for delay and reliability will also become higher and higher. After the mobile edge computing is applied in the Internet of Vehicles, due to the position characteristics of the mobile edge computing, the Internet of Vehicles data can be stored nearby at a location close to the vehicle, so the delay can be reduced, which is very suitable for business types with extremely high requirements for delay standards such as anti-collision and accident warning in the Internet of Vehicles.
At the same time, the Internet of Vehicles is ultimately driven. During high-speed movement, the vehicle's position information changes very rapidly. The mobile edge computing server can be placed on the vehicle body, which can accurately sense changes in vehicle position in real time and improve communication reliability.In addition, the mobile edge computing server processes real-time Internet of Vehicles data with huge value, conducts data analysis in real time, and transmits the analysis results to other networked vehicles in the neighboring area with extremely low latency (usually milliseconds) so that the vehicle (driver) can make decisions. This method is more agile, more autonomous and more reliable than other methods of processing.
Augmented reality (AR) refers to computer technology that applies virtual information to the real world, and the real environment and virtual objects are superimposed on the same picture or space in real time at the same time. AR can greatly enhance people's experience, and one of the key technologies to achieve is ultra-low latency. The transmission delay directly determines the user's viewing experience, and the increase in the delay will cause the viewer to feel dizzy. According to the first VR head-mounted display technical benchmark completed by Digi-Capital, the delay time required is less than 19.3ms, otherwise a vertigo will occur.
The typical technical feature of mobile edge computing is low latency, so in AR, mobile edge computing has a broad application scenario. Mobile edge computing can greatly reduce latency, improve data processing accuracy, and enhance user experience by processing information transmitted by AR devices in real time.
At present, there are two common methods for data processing of surveillance videos: one is to process on the camera, and the other is to process on the server. Camera processing requires every camera to have data analysis capabilities, which is very expensive. However, when the server processing requires uploading a large amount of data to the server, it will increase the burden on the core network and have a large delay and be too low in efficiency.
By deploying the mobile edge computing server unit, the mobile edge computing server is used to localize the monitoring video data, and there is no need to upload a large amount of video data to the server, which reduces the burden on the core network, improves efficiency, and does not require the camera to have data analysis capabilities, which reduces the cost.
4
4 technical analysis of mobile edge computing
According to the edge computing reference architecture released by the Edge Computing Industry Alliance, mobile edge computing should be a system with both "hardware + software". The edge computing types can be divided into three types: local equipment, localized data centers (1-10 racks) and regional data centers.
(1) Local device: Suitable for home or small office applications. The size of the local device depends on the application scenario and the specified purpose, but the scheduling is "instant". Security systems running in buildings (Intel SOC devices) and storing local video content on DVR are typical examples of this edge computing. Another example is a cloud storage gateway, which is a local device, usually a network device or server that acts as a cloud storage API such as SOAP or REST. Cloud Storage Gateway enables users to integrate cloud storage into applications without having to move applications to the cloud.
(2) Localized data centers (1-10 racks): These data centers provide important processing and storage capabilities and can be deployed quickly in existing environments. These data centers can often be pre-designed to the order system and then assembled on site. Another form of localized data centers are prefabricated micro data centers that are assembled in factories and placed on site. These single housing systems can be made of a solid housing type (can be rainproof, corrosion-proof, fire-proof, etc.) or use a normal IT chassis in an office environment.
single rack version can save capital expenditures by leveraging existing buildings, refrigeration and electricity without having to build a new dedicated website.
(3) Regional Data Center: A data center with more than a dozen racks and closer to users and data sources than a centralized cloud data center is called a regional data center. Due to their huge scale, they will have more processing and storage capabilities than localized data centers (1 - 10 racks).Even if they are prefabricated, this will encounter licensing and local compliance issues as construction may be required, they will take longer than localized data centers and require dedicated power and cooling sources. The delay will depend on the physical proximity of the user and data and the number of hops in between.
China Unicom Network Technology Research Institute experts believe that the deployment location of mobile edge computing server (MEC SERVER) is relatively diverse:
MEC server is deployed on the RAN side base station aggregation point. At the same time, MEC servers can also be deployed behind a single base station on the RAN side, which is mainly aimed at hot spots, such as campuses, large shopping centers, etc. The advantage of this architectural solution is that it is more convenient to obtain wireless-related information on the base station side through monitoring and parsing the signaling of the S1 interface, but security issues such as billing and legal monitoring need to be further solved.
This scheme MEC server is deployed together with the P-GW on the CN side. This approach does not change the existing EPC architecture, and the MEC server is deployed with P-GW. The data service initiated by the UE passes through eNodeB, Hub Node, S-GW, P-GW+MEC servers, and then goes to the public Internet. This deployment method does not have billing, security and other issues.
also has a solution to deploy the MEC server and the D-GW on the CN side. This method changes the existing EPC architecture. The MEC server is deployed with D-GW. The original P-GW is split into P1-GW and P2-WG (i.e. D-GW), where P1-GW resides in its original position and D-GW moves downward (can be to the RAN side or to the CN edge). D-GW has functions such as billing, monitoring, and authentication. The MEC server can be integrated with the D-GW or can be deployed as a separate network element after the D-GW. The P1-GW and D-GW are private interfaces, and the same manufacturer requires equipment.
There are also two deployment plans for MEC server based on 5G architecture, one is deployed at GW-UP, and the other is deployed after NodeB.
The MEC server is deployed after NodeB: As shown in the MEC server 1 location in the figure below, the MEC server is deployed after NodeB (one or more NodeBs), making the data service closer to the user side. The data service initiated by the UE passes through NodeB and MEC server 1 and then goes to the Internet (third-party content provider server). In this way, security issues such as billing and legal listening need to be further solved.
MEC server is deployed at GW-UP: as shown in the MEC server 2 position in the figure below. After the C/U functions of the 5G network core network are separated, the U-Plane (corresponding to GW-UP) function is moved down (can be moved down to the RAN side or to the edge of the CN), and the C-Plane (corresponding to GW-CP) resides on the CN side. The MEC server is deployed at GW-UP, and compared with traditional public network solutions, it can provide users with low latency and high bandwidth services.
5
5 related layout company
Nokia is one of the first companies to focus on the field of mobile edge computing. The concept of mobile edge computing first appeared on a computing platform jointly developed by Nokia and IBM. At the same time, Nokia is also an ETSI member and is actively promoting the formulation of standards for mobile edge computing.
As early as 2014, Nokia supported China Mobile to demonstrate the mobile edge computing platform - Nokia's Liquid Applications. At the same time, Nokia has also proposed a cloud platform MEC solution. This solution is based on the cloud platform virtualization architecture. Using MEC virtual network elements, it can support access to macro stations and small base stations at the same time. It uses Nokia's general AirFrame cloud platform to integrate MEC and other applications and use an open API interface, which is characterized by high compatibility, high scalability and flexibility. The key driving force of AirFrame is that MEC can meet the extremely low latency, large throughput, secure programmable operation, and network agility necessary for 5G and the Internet of Things.
In 2016, Nokia released three mobile edge computing applications tailored for enterprises: target tracking, video surveillance and video analytics.
Nokia MEC has a wide range of cases around the world, such as the local computing smart port in South Korea, live video director of football matches in the UK, German road MEC combined with the Internet of Vehicles, and Shanghai International Racing Circuit multi-angle video live MEC networking scheme. The live video of this MEC solution is only about 0.5 seconds longer than the live live, providing the audience with an excellent viewing experience. LeTV live video is delayed by about 47.95 seconds compared to the MEC live video.
As the world's largest manufacturer of personal computer parts and CPUs, Intel is developing well in the field of the Internet of Things. According to the company's 2016 annual report, the Internet of Things business has accounted for 4.4% of its revenue. Intel believes that mobile edge computing will be an indispensable and important link in this. In the 5G era, its applications will be extended to transportation systems, intelligent driving, real-time tactile control, augmented reality and other applications.
Intel is also an important participant in the mobile edge computing industry. In 2014, Intel and some other manufacturers in the industry, including Huawei, Nokia, AT&T, DoCoMo and other manufacturers, established mobile edge computing through the ETSI Standardization Association. In 2016, Intel and Huawei, ARM and other companies launched the establishment of an edge computing industry alliance in Beijing to actively promote the development of the mobile edge computing industry. In the same year, Intel released the white paper "Real-time Live Broadcast Solution for Drones Carrying LTE Small Base Stations". The white paper provides a comprehensive introduction to the end-to-end solution based on mobile edge computing, launched by Intel and Baicells.
and with its own technical characteristics, Intel has launched the NEV SDK (Network Edge Virtualization Kit), which can assist partners in the mobile edge computing field to accelerate the development of related applications for the telecommunications field. In addition to infrastructure platform capabilities, the NEV SDK can also provide mobile edge computing application developers with IP services based on basic software environments with rich API interfaces and high-performance forwarding capabilities. According to Intel's plan, Intel will promote the development of mobile edge computing in multiple directions and differentiation.
Linghua Technologies - a world-class leader in embedded computing technology, headquartered in Taiwan, and has branches in the United States, Singapore , Beijing, Japan, South Korea and Germany. The company has a world-leading position in the fields of x86 computing, reinforcement design, high reliability and industrial I/O integration.
Linghua Technology is a member of the domestic edge computing industry alliance. While actively promoting the establishment of industry standards, it also continuously launches mobile edge computing products. In 2015, Linghua Technology announced the launch of the world's first reinforced, high-performance mobile network terminal computing platform EXTREME OUTDOOR SERVER, which is specially designed for rigorous outdoor telecommunications and network applications. It can be built outdoors to meet the needs of mobile edge computing.
At present, the company has launched the world's first high-performance server-level mobile edge computing platform SETO-1000. Linghua Technology SETO-1000 is a server designed specifically for extreme and harsh outdoor environments. The SETO-1000 supports two Intel® Xeon® E5 processors, up to 96Gb of memory, as well as a rich I/O interface and two hot-swap SATA hard drive slots. SETO-1000 provides a powerful general platform architecture for virtual wireless access devices in 2G, 3G and LTE, while also integrating security, remote management, open applications and other functions.
The company launched OCCERA (open telecommunications-grade edge computing architecture) in 2016, and launched three mobile edge computing products under this architecture.
Huawei is an active promoter of the mobile edge computing industry.In 2014, a joint promotion by six operators and suppliers including Huawei and Vodafone was established at ETSI. In 2016, Huawei, together with Intel, ARM and others, launched an edge computing industry alliance in China to jointly release the "Edge Computing Industry White Paper", and proposed the "OICT" concept for the first time in the industry, aiming to build an edge computing industry cooperation platform and promote open collaboration between OT and ICT industries. Huawei actively promotes extensive cooperation and docking between the alliance and domestic and foreign standard and industrial organizations, and accelerates the development of the alliance and standard output.
At the same time, Huawei is also a mobile edge computing solution provider. Huawei released the industry's first MEC@CloudEdge solution for future network architectures at the MEC Congress conference in Munich, Germany. As a 5G-oriented MEC solution, Huawei's MEC@CloudEdge solution deploys some of the service processing and resource scheduling functions of the application, content, and MBB core network to the network edge close to the access side. By bringing services close to users for processing, and collaborating applications, content and networks, it provides a reliable and ultimate business experience.
In terms of the combination of the Internet of Things and mobile edge computing, Huawei released the Internet of Things EC-IoT (Edge Computing IoT) solution based on edge computing to the world at the 2017 Mobile World Congress (MWC 2017). Innovatively introduce edge computing and cloud management into the field of the Internet of Things. Agile controllers based on SDN and Internet of Things gateways with edge computing capabilities (AR 500 series products) provide intelligent services nearby, and network management is fully clouded, realizing full-process industrial services and business model innovation, enabling industry digital transformation, and unleashing the huge potential of industrial innovation.
As the world's leading wireless communication solution supplier and communication equipment provider, ZTE is an industry giant in China that is paying attention to the field of mobile edge computing. In recent years, through cooperation with operators, it has achieved certain results in the field of mobile edge computing.
ZTE has a complete mobile edge computing MEC solution, including core technologies and patents such as virtualization, containers, high-precision positioning, diversion, CDN sinking, and other core technologies and patents. The related solutions cover business localization, local cache, Internet of Vehicles, Internet of Things and other scenarios. MEC technology can effectively integrate wireless networks and the Internet, and add computing, storage, processing and other functions on the wireless network side. Through service localization and API interfaces, information interaction between wireless networks and service servers is opened, effectively reducing the pressure on the transmission network, allowing operators to process information faster and realize differentiated services, truly changing users' business experience.
Recently, at the "2017 MEC Technology and Industry Development Summit" held in Beijing, ZTE announced that it has begun to work with three domestic operators to conduct MEC pilot and technology verification, and plans to conduct commercial deployment in 2018. Since 2016, ZTE has successively cooperated with the three major domestic operators to actively carry out various MEC pilot work. Among them, ZTE and Ningbo Telecom launched a park network cooperation and carried out local traffic unloading projects for the park network; in Beijing and Zhuhai, ZTE carried out precise indoor positioning projects with China Mobile; in Ningbo, based on MEC and NB-IoT, ZTE developed smart parking and smart park projects. At the 2016 Shanghai MWC, ZTE and China Unicom demonstrated VR services based on 5G MEC.
In 2016, ZTE and China Unicom demonstrated a mobile edge computing solution based on 5G architecture. In April 2017, the company and Beijing Mobile successfully completed the pilot verification of the mobile edge computing indoor high-precision positioning scheme based on the QCell room division scheme. After the innovative mobile edge computing positioning solution achieves rapid and flexible deep coverage of indoor 4G signals through QCell, based on ZTE's open mobile edge computing platform, it provides localized, low latency and high bandwidth services close to users, and also provides an open API interface to allow rich third-party applications and content to enter the pipeline to meet the diversified business needs of indoor users and realize value-added operator network pipelines.
In addition to mobile edge computing, there are continuous highlights in corporate governance. In addition to the prosperity cycle of IoT and 5G construction, the company has achieved steady recovery of valuations this year. According to the company's employee equity incentives for gambling performance, the company's net profit from 2017 to 2019 will not be less than 4.208 billion, 4.590 billion and 4.973 billion respectively. If we consider the growth rate of nearly 30% of the net profit in the first half of the year, it is expected that the company's net profit from 2017 to 2019 will be 4.5 billion, 4.8 billion and 5.2 billion, corresponding to PEs of 20X, 19X and 17X respectively. Maintain the "buy" rating and continue to recommend it!
Netsu Technology mainly provides services such as Internet content distribution and acceleration (CDN), cloud computing, cloud security, global distributed data center (IDC), and is the leader in the domestic CDN industry.
NetEase Technology accurately analyzed the relationship between MEC and CDN, and believed that one of the evolution directions of future CDN is to form an edge computing system. Therefore, NetEase Technology has a clear plan for MEC. NetEase Technology will carry the future by laying out centralized data center + edge computing nodes. In addition, the company is upgrading its existing CDN nodes to edge computing nodes with storage, computing, transmission and security functions, and deploying a larger number of edge computing nodes to metropolitan networks closer to users.
Affected by the intensified domestic CDN competition, we predict that the company's net profit from 1 billion to 1.2 billion yuan from 2017 to 2018, with a PE of 25X and 21X.
On June 13, 2017, Rihai Communication issued an announcement stating that the company's wholly-owned subsidiary Haiyuntai increased its capital with its own funds of 30 million yuan in cash. After the capital increase, Haiyuntai will hold 2% of Baicaibang's equity. In addition, Haiyuntai will appoint a director to Baicaibang. The following matters involved in Baicaibang must be reviewed and approved by the board of directors. Other directors must maintain consensus with the directors appointed by Haiyuntai on the matter: (1) Baicaibang's decision on the market sales of China Telecom Operator Group and its branches in various cities and the research institute; (2) After the capital increase agreement is signed, Baicaibang's introduction of OEMs or agents of China Telecom Operator Group and its branches in various cities and other partners.
Baicaipang is a world-leading provider of complete solutions for small base stations. It focuses on the research and development of wireless broadband access solutions, business operation platforms and future wireless broadband technology innovations related to small base stations. It has provided 4G smart small base stations and other products to multiple customers in many countries around the world, such as mobile operators, broadband access operators, cable TV operators, industry dedicated networks and enterprise networks, and is also committed to the research and development of next-generation wireless technologies such as 5G.
Baicaipang is a typical example of combining VR with mobile edge computing, live video and mobile edge computing technology in China. Baicaipang and Intel released the white paper "360-degree video live broadcast solution for drones equipped with LTE small base stations", introducing the end-to-end solution based on mobile edge computing. At the same time, in September 2016, Baicaipang and China Unicom exhibited a new MEC drone VR solution for 5G. This solution is a live video broadcast solution for drone VR based on MEC. The solutions jointly developed by China Unicom and Baicaipang combine a number of today's most advanced technologies, including panoramic video synthesis algorithm, panoramic video transmission protocol, MEC architecture, LTE/5G data channel QoS guarantee and other key technologies. The video stitching algorithm and panoramic video transmission protocol ensure seamless panoramic VR video; the MEC architecture makes the service closer to users, and combines LTE/5G transmission to ensure smooth VR panoramic video images, high-speed propagation and no interference, establishing a high-speed road for data transmission; the drone is equipped with a 360-degree panoramic high-definition camera without any blind spots in sight, and users can enter the panoramic video to manipulate it, observe it, and achieve an unprecedented immersive VR live broadcast experience.
Baicaipang's main business is a small base station, and the deployment of MEC servers cannot be separated from the influence of the base station location. Baicaibang's creative use of drones to move small base stations is in line with the idea of mobile edge computing.As a shareholding company, Rihai Communications will benefit from it after the development of MEC.
We believe that after the new shareholders entered Rihai Communications, they introduced new chairman and management, and the market grasp ability was further improved. After completing employee shareholding, the company has not relaxed, optimized its equity, and participated in Baicaipang. It is expected that corporate governance will be more scientific and its future strategy will be clearer. We expect the company's net profit from 2017 to 2018 to be 110 million and 210 million respectively, with the corresponding PE of 57X and 30X. It is recommended to pay attention to it!
6
Investment advice
Since mobile edge computing was born in 2013, it is still in the process of technological research and development and industrialization. However, as one of the core technologies of 5G and one of the development directions of CDN, it has good development prospects, so giants have made arrangements, including Nokia, Intel, Huawei, ZTE, etc.
From an investment perspective, we recommend focusing on four directions. First, edge computing device manufacturers, such as servers and gateways, such as Unigroup (Xinhua III), Inspur Information, ZTE, Rihai Intelligence, Seits, etc.; second, the opportunity for CDN service providers to deploy edge computing, such as NetEase Technology; third, the design and construction providers of edge computer rooms, such as data ports; fourth, cloud security manufacturers, cloud computing has been sinking to the edge, and the importance of data security and network security will be self-evident. You can pay attention to Shenxinshui, China News Service, and Hengwei Technology.
We believe that mobile edge computing is expected to develop together with 5G. Superimposed edge computing emphasizes getting closer to users and paying more attention to the richness of edge nodes, so it will be mainly deployed in the operator network in the early stage. Therefore, companies with first-mover advantages and good cooperation with operators may be more likely to stand out, such as ZTE, Rihai Intelligence, NetEase Technology, etc.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-