As 5G goes mainstream across theworld, 5G-enabled networks will transmit high-quality videos on an unprecedented scale and at a super-low cost, which in turn will intensify competition among operators in the video field. Mobile data currently accounts for 14% of global IP traffic, and the figure will rise to 20% by 2022. Of all mobile data traffic, video traffic makes up nearly 2/3, with the proportion forecasted to increase to 82% at the end of 2022.
In the coming years, video is expected to become the main driver of 5G traffic, with short video and livestreaming among the earliest applications to benefit from 5G. China mobile users spent an average of 28 hours online per week in June 2020, a year-on-year net increase of 2 hours. The usage of short video peaked at 11:00 to 13:00 and at 17:00 to 22:00, respectively, with the combined duration of the peaks accounting for 53.7% of the total viewing time. Users tend to watch livestreams in an even more concentrated timeframe, with the usage duration in the 18:00 to 23:00 period making up more than 40% of the overall viewing time. 5G has the advantages of high bandwidth, low latency and massive connections, and so is well placed to eliminate the bottlenecks currently experienced by short video in terms of traffic, rate, and cost.
A dramatic increase in 5G connections and traffic, however, presents the following challenges to operators:
—As the adoption of 5G increases, VR/AR and 8K+ ultra high definition (UHD) services begin to explode and generate increasing volumes of traffic. But at the same time, mobile users demand a better service experience, which can only be delivered with high bandwidth and low latency.
—Scenarios including emergencies, online celebrities, and hit shows, as well as places like stadiums, subway stations, high-speed railway stations, airports and densely populated campuses, produce highly concentrated and provisional traffic that ebbs and flows quickly. 

Disadvantages of Traditional CDN 
The architecture of traditional content distribution network (CDN) can no longer meet the traffic and user requirements of 5G networks because: 
—The traditional CDN uses a hierarchical architecture to provide video streaming service for users. The origin server that is farthest from users has a larger storage capacity to guarantee the integrity and reliability of the stored content while providing users with a small amount of service. The edge nodes nearest to users have a higher throughput and cache the most popular content locally so that the bulk of service requests can be directly terminated on them. However, when content requested by users does not exist at the edge nodes, the request has to go to the regional nodes or further to the origin server after first passing through the regional nodes. Requesting and pushing content in such a cascading way poses a number of problems, including slow speed, heavy source traffic, and excessive backhaul and backbone traffic.
—To cope with the rapid traffic fluctuations of 5G networks, new edge nodes often need to be added to the CDN and they must begin to process services shortly after installation. With the traditional architecture, content takes dozens of hours to be pushed down to the new nodes, which means that traffic fluctuations cannot be alleviated in a short period of time. However, after traffic peaks disappear, the new nodes are lightly loaded, which causes a waste of resources. 
—In the traditional CDN, content is distributed among different devices in the same node. If some devices store popular content, load may be unbalanced between different devices of a node. A few devices may produce poor service because they are overloaded.

5G CDN: New Architecture for 5G Networks
5G CDN removes the barriers of the traditional CDN by using a smart distribution system to distribute content more flexibly through multiple layers and hence markedly boost content transfer efficiency (Fig. 1).

Key Technologies of 5G CDN
Automatic Replication of Popular Content 
In the traditional CDN, content is evenly distributed among devices. When certain content becomes popular, a large number of users concurrently visit the device that stores the content, straining the processing capability of the device and degrading the service it delivers.
The core scheduling layer of ZTE 5G CDN can automatically analyze the popularity of media content. When the popularity of certain content reaches the preset threshold, the CDN can quickly replicate the content from the device containing it, whichever network layer the device is located at, to other devices so that all the devices can balance the load. Users' service requests can also be flexibly scheduled according to the load carried by the devices.  
The automatic replication technology allows content to be effectively shared in the CDN, thus greatly reducing source traffic and lightening the load on the backhaul network. In some circumstances, it can cut up to 50% source traffic for operators.

MEC and Fast Resource Scaling
ZTE 5G CDN supports containerized deployment of virtual content distribution networks (vCDNs). It integrates a multi-access edge computing (MEC) platform to allocate and manage resources for vCDNs in a unified manner (Fig. 2).

Smart Scheduling
The smart scheduling technology can distribute popular traffic evenly among different nodes or different devices within a node.
The core scheduling layer of ZTE 5G CDN uses a smart scheduling decision support system to collect its operational data. The AI technology is introduced to make decisions on handling hot traffic and schedule traffic loads so that load is balanced among intra-nodes, inter-nodes, areas, and users. This is to prevent a single node or device from carrying an excessive volume of traffic that would stall and degrade service. Meanwhile, high-priority users are allocated to optimal nodes to differentiate service levels.

Low-Latency Livestreaming 
ZTE 5G CDN leverages the industry-leading low-latency livestreaming technology to dynamically create live channels and achieve end-to-end latency of less than 2.5 s, which are especially crucial to guaranteeing a superior experience for latency-sensitive livestreaming services such as TV sports and games (Fig. 3).

Evolution to Edge Cloud 
Operators have rich network infrastructure resources, unique edge connectivity, cloud-network convergence, and abundant facilities like equipment rooms and hardware. Edge clouds and CDNs are similar in terms of service and deployment location. As the 5G age presents increasing demands for edge clouds, operators are also accelerating the deployment of MEC capabilities on CDNs to take full advantage of their existing CDN resources. 
ZTE and China Mobile have started to deploy edge cloud and CDN resources in phases. They will gradually cloudify CDNs and put them on existing edge clouds. On this basis, they aim to enrich the platform as a service (PaaS) components and achieve the coordinated sharing of platform capabilities.
By coordinating equipment room locations, equipment resources and edge cloud platforms, ZTE deeply integrates 5G networks with MEC CDN service platforms and uses industry-leading technologies to continuously optimize user viewing experience.

Summary
ZTE 5G CDN solution is specially designed for mobile application scenarios and employs technologies that involve containerized CDNs, dynamic scaling, hot content statistics, prediction and replication, smart scheduling, and low latency. The solution can offload around 70% of traffic storms and tackle problems caused by rapid rises in video traffic. By drastically reducing traffic pressure, eliminating traffic overloads, and solving poor quality for 5G backhaul and backbone networks, the solution provides a reliable, cost-efficient alternative to expensive backhaul upgrade. Operators can use the 5G CDN to easily deliver a better 5G video viewing experience to a surging user base.
The architecture design of ZTE 5G CDN fully considers 5G development and industry standards and supports 5G multimedia services like VR, AR and XR that require high bandwidth, low latency and strong interaction. The containerized architecture can also rapidly evolve into edge clouds, helping operators implement a new operation model that features cloud-network convergence.