Edge computing accelerates live video scenes: clearer, smoother, and more real-time

With 4G、5G The development of network technology and the popularity of smart phones , Webcast has gradually become the main mode of transmission of new media , In social entertainment 、 Product display 、 Open government affairs 、 Exhibition release and other fields are widely used .

Facing different cities across the country 、 Users from different operators , How to ensure clear live video 、 Fluent 、 Real time viewing and interactive experience ？ This article mainly shares the application of edge computing in live video scenes , The core content is as follows ：

1. What is live broadcast ？

2. System architecture of live broadcast

3. Edge computing accelerates live broadcast

4. summary

01  What is live broadcast ？

according to 《 Radio and television dictionary 》 Definition , Live broadcast refers to the post synthesis of radio and television programs 、 The broadcast mode of simultaneous broadcasting . In this way, the post synthesis process of the program is the broadcast process of the program , It is not recorded or recorded in advance , But in the live or studio at the same time 、 The studio completes the production and broadcasting of the program .

Live broadcasting can fully reflect the advantages of radio and television media ：

1. You can broadcast the latest news at any time through live broadcast , Ensure the timeliness of news reports .
2. It can synchronously report the occurrence and development of news events .
3. Strong sense of presence , Achieve good broadcasting effect .

With 4G、5G The development of network technology and the popularity of smart phones , Webcast based on Internet and streaming media technology gradually began to develop . Rely on the internet widely 、 Convenient network resources , Webcast encodes audio and video signals 、 Compress 、 After packaging , Real time transmission and playback on the Internet through streaming media transmission protocol . Compare radio and television live , Webcast vs. anchor in professional skills 、 form 、 Live broadcast content and other aspects are more diverse and loose , And it can provide more interactive two-way communication with the audience , Webcast has gradually become the main mode of transmission of new media , In social entertainment 、 Product display 、 Open government affairs 、 Exhibition release and other fields are widely used .

Webcast is based on the content scene , It is mainly divided into e-commerce live broadcast 、 Live broadcast of the game 、 Live reality TV 、 Live concert 、 Sports live broadcast and other types . according to CNNIC released 《 The first 48 The second statistical report on the development of Internet in China 》 data , By 2021 year 6 month , The number of live webcast users in China reaches 6.38 Billion , Year-on-year growth 7539 ten thousand , Accounting for the overall number of Internet users 63.1%. Benefit from a wide range of user groups , Webcast will continue to highlight its economic and social value .

02  Architecture of live broadcast system

Infrastructure

The live broadcast infrastructure consists of the host 、 Server side 、 The audience side consists of three parts ：

• The anchor side ： Video production source . Collect audio through hardware devices 、 After video data , Encoding 、 Compress 、 After encapsulation , Push the stream to the corresponding media service of the server through the network .

• Audience ： Pull the target video stream on the server media server through the network , Play after audio and video decoding through local player .

• Server side ： Contains a series of media servers , Realize the control surface of live video （ User authentication 、 Room management, etc ） And the data side （ Push and pull streaming of audio and video streams ） Unified access 、 Management and scheduling capabilities . meanwhile , Media services can also be based on the business needs of the live broadcast platform , Connect the video stream 、 transcoding 、 Content review 、 Recording and other functions .

Business structure

Generally speaking, the live broadcast service architecture is mainly composed of the host 、 Media services 、 Play end 、 Live broadcast business platform and other parts ：

• The anchor side ： Through the live broadcast APP Integrated streaming SDK, Realize live broadcast related collection 、 code 、 Streaming and value-added services .

• Audience ： Through the live broadcast APP Integrated playback in SDK, Realize streaming related to live broadcast 、 decode 、 Playback and playback experience optimization capabilities .

• Live broadcast business platform ： The anchor and audience are releasing / When subscribing to live content , All need to rely on live broadcast APP： For example, the anchor only has the corresponding live broadcast APP After the platform is registered and approved , Before you can create a room for live broadcast , Then the audience also broadcast live APP Enter a certain anchor's room to watch live broadcast and interaction in real time . Corresponding to live broadcast APP The back-end platform is the live broadcast business platform , Realize user registration and management 、 Room management 、 Authentication and authorization management capabilities .

• Media services ： Realize the access to live streaming 、 Release 、 Forwarding capability , Merge live streaming and content based on business needs 、 review 、 transcoding 、 Recording and other capabilities .

In the commercial live broadcast APP in , There will also be barrages 、 Reward 、 Business modules such as product recommendation , Provide more fun 、 Novel live interaction and experience .

03  Edge computing accelerates live broadcast

As the business grows , The live broadcast platform needs to be available to different cities across the country 、 Users from different operators provide coverage and live broadcasting services , The following problems will follow ：

1. High concurrency bottleneck ： The live broadcast center is difficult to carry millions 、 Tens of millions of users have concurrent requests .
2. Bandwidth bottleneck ： The network resources of the live broadcast center are difficult to meet the video access and distribution of a large number of users .
3. Inconsistent experience ： The difference in physical distance between users in different places brings additional network delay , This leads to inconsistent user experience .

Edge computing aims to be closer to the user's network edge , Provide standard computing power and IT service . Through time-lapse sensitive business 、 The traffic access service is deployed locally , On the one hand, it can access users nearby and reduce the service response delay , On the other hand, the business can be processed locally at the edge, which can effectively share the central bottleneck and improve the business capacity .

 

The push-pull flow accelerates

If the live streams of all anchors are pushed to the live broadcast center , Then all the audience also streamed from the live broadcast center , It is bound to bring huge business pressure to the live broadcast center . Simultaneous anchor / The network link between the audience and the live broadcast center has long-distance transmission , Its network performance is limited by bandwidth 、 Instability and other factors will also affect the final user experience .

The edge instance provided by edge computing builds a live broadcast network , Deploy push-pull flow capabilities closer to the edge of end users , Access nearby 、 Forward user live stream data , Reduce user delay . At the same time, the node where the edge instance is located has the operator's high-quality Special line exit , It can effectively guarantee users - Edge node - Network transmission quality of live broadcast center , Improve business stability .

• Upstream streaming accelerates ： Based on user geographical location affinity , Intelligent scheduling users （ The host ） Provide live video streaming services to the nearest edge computing node , Reduce user access response delay and improve experience .

• The downward pull accelerates ： Based on user address location affinity , Intelligent scheduling users （ The audience ） Provide live video streaming services to the nearest edge computing node ： If the live stream has been cached locally , Direct distribution ; If there is no local cache , Then go back to the source live broadcast center to pull the stream , Effectively reduce user streaming delay , And greatly save the cost of bandwidth expenditure in the live broadcast center .

Media processing accelerates

General computing power provided by edge computing , It can support the processing of user media data of live video broadcast locally , For example, transcoding video live media services 、 section 、 Confluence and other services are directly deployed locally at the edge , On the one hand, local media data processing can Improve the response efficiency of user interaction related businesses , On the other hand, compress the data at the data source 、 Optimization can improve the edge - The data transmission efficiency of the center optimizes the cost . meanwhile , From the perspective of the overall system architecture , The distributed deployment of live broadcast services can be realized by adopting the edge computing architecture with scattered physical locations , To achieve higher concurrency 、 More stable business capabilities .

 

Consider implementing the following media service processing at the edge node ：

• Narrowband HD / transcoding ： On the premise of ensuring the quality of video users watch , Compress the live streaming video , Reduce the demand for backhaul bandwidth .

• Confluence ： Merge multiple video streams of users at the edge , Reduce the demand for backhaul bandwidth .

• Turn to encapsulate ： For different terminal types of users （IOS、Android、HTML5 etc. ） Dynamic conversion streaming media encapsulation protocol .

• transcoding ： Realize the conversion of different coding formats of video streams .

• Super resolution ： Improve the image quality of the original film through algorithm , Meet the user needs of high-quality pictures .

• Dynamic resolution ： Dynamically switch videos with different bit rates based on users' real-time network environment , Keep the video flowing .

It is worth mentioning that , More and more interesting gameplay and special effects have emerged in the live broadcast scene , such as Tiktok is quite hot “ Ants, hey ”、“ Cartoon style ” Wait for video special effects , It can greatly enrich user interest and enhance user stickiness . High quality video special effects , It also puts forward higher resource requirements for user terminals , But many mid-range 、 Low end terminals are limited by hardware performance bottlenecks, which makes it difficult for users to obtain the expected effect of the product, thus affecting the experience . Edge computing nodes provide general GPU Computing resources , It can effectively assist the terminal to achieve high-quality video special effect rendering , Ensure that users get the expected live experience .

AI Application acceleration

With AI The development and maturity of technology , More and more AI Technology is applied to live broadcast scenes , Such as through AI Algorithm in content understanding 、 Content innovation 、 Achieve better quality in terms of content transmission 、 Innovative live broadcast experience .

 

From the source of computing power , Central cloud 、 Edge node 、 User terminals can provide AI Computing resources needed by business ：

1. Central cloud , With a large scale 、 High performance AI Computing power advantage , Can satisfy all kinds of AI Business scenario resource requirements ; Limited by user distance , It has the disadvantage of low timeliness when serving long-distance users .
2. user terminal , Some medium and high-end mobile terminals can provide certain AI Calculation force is used for AI Business . Of the user terminal AI Computing power has the lowest delay for users 、 Experience the best , For the business side, there is no need to invest resources 、 Cost optimization ; However, due to the variety of user terminal models 、 Performance levels are uneven , It requires the business party to invest considerable resources in compatibility and adaptation , And the final effect is limited, and the hardware performance is difficult to achieve the expected effect .
3. Edge of computing , It can provide general computing power closer to the user edge , Such as consistent with the data center GPU The graphics card , You can meet AI Computing resources needed for business streaming or rendering , At the same time, it is closer to users Ensure the timeliness of the business , So it can be very good Coordination Center AI Calculate force and end side AI Computing power is in time delay 、 Compatibility 、 Computational power 、 The difference in cost .

04  summary

Edge computing provides full coverage of provinces, cities and operators across the country , In the live video scene, users will interact with push-pull streams related to traffic access 、 Transcoding and merging services are deployed closer to the edge of users , Push and pull the live video of the anchor and the audience nearby , Ensure low delay service response , Improve the smooth business experience of HD live broadcast . meanwhile , With the interest of live video content 、 innovative 、 The increasing demand for immersion , Edge computing provides low latency 、 Standardization 、 Isomerized computing resources will continue to help the live broadcast scene provide a more extreme business experience .

 

 

Volcano engine edge computing node is dedicated to providing Stable 、 High performance 、 A new generation of edge computing cloud platform services with rich functions , By covering the edge nodes of provinces, cities and operators all over the country , Help businesses quickly deploy to each edge layer between users and cloud centers .

Current volcanic engine edge calculation It has been broadcast live in Tiktok 、 Today's headline 、 Watermelon Video 、 Flying book and other applications have landed . Welcome to pay attention to the edge computing node or open the service ～