Win in the era of big video: Video bearer network based on U-vmos

Wen/Tao Liufei

The video experience is now the Web business experience that users are most concerned about. Conviva User Video Experience Research report data shows that 35% of users use the video viewing experience as the primary condition for selecting a video service, or even exceeding the video content itself, and 84% of users stop watching within a minute of experiencing the difference.

The video experience is so important, but in the past there has been little concern in the industry, which is determined by the industrial chain structure. The quality of the content source and the network will affect the video experience, but the actual operation of the two are basically fragmented, even if the operator's own video business experience is not ideal, from the user terminal to the video server can cause the video experience deterioration, the operator is difficult to do fast problem delimitation and fault handling.

U-vmos Video Experience Standard

From the perspective of video experience and network optimization, Huawei combines a large number of user surveys and human engineering tests to objectively quantify the user's subjective perception of the video experience and ultimately optimize it on the basis of the VMOs (video Mean Opinion score) standard in ITU-D. The U-vmos video experience standard is presented.

U-vmos Station in the user perspective, from the "video quality, viewing experience and interactive operation" 3 aspects (including video source resolution, the number of slices, playback screen size, operation experience and play smoothness and other indicators) to give a score, using 5 points to indicate the quality of experience (5 of which are excellent, 4 points good, 3 points general, 2 points poor, 1 points difference). The final U-vmos score is calculated as follows:

According to Huawei Ilab's analysis of the 4K video viewing experience, the viewing experience is significantly degraded when bandwidth, latency, and packet loss exceed a range.

U-vmos Standard and corresponding network index (experimental statistics) table

fixed network optimization based on U-vmos

U-vmos to the network construction, through real-time acquisition of network indicators and their corresponding U-vmos scores, and then through the big data analysis to understand the impact of the key network indicators of experience, and finally give the network optimization and operation and maintenance recommendations.

The current mainstream video business in the highest resolution is "4K", the network supporting 4K video will certainly be able to provide a better viewing experience for video below 4K resolution, and according to U-vmos Standard, 4 points above is a good experience, combined with the example table, The best 4K video experience (up to 4 standard) corresponding to the key network indicators are: bandwidth 100M, delay 30ms, packet loss rate of 0.002%, which means that the network to reach the standard of 4 points, must be based on 100M above the end-to-end bandwidth capacity to plan the network.

From the view of network architecture, traditional networks can reduce the cost of network equipment port expansion after the increase of users by multi-level aggregation, but in the era of video service, the network traffic model changes, multi-level aggregation network can not play the effect of convergence, but each class of network needs to be expanded, bringing huge expansion cost pressure. Therefore, reduce the network level, reduce the traffic convergence ratio (from 10:1 to 2:1), is the main direction of the current network planning, will also be the main direction of future network architecture evolution.

The development of the carrier video service is broadly divided into 3 stages, and the business characteristics and experience guarantee of each stage are different from the key challenges of the network, so we need to adopt the strategy/response plan according to the characteristics of different stages.

First stage: Video (4K) business start-up phase. The number of users in this phase is less, the main concern is the single-user e2e business throughput, and the line quality is sufficient to support the opening of video services. It is suggested that the incoming copper wire be converted into fiber, and the access rate can be increased to 100m~1g (copper change light also improves the time delay and packet loss problem). For fiber-to-household difficulties, it is also possible to consider introducing vectoring/g.fast technology to increase the copper wire rate to 100m~1g (along with the shortening of the copper wire distance to improve packet loss and delay). In addition, other technologies can be used to enhance the experience, such as the deployment of TCP acceleration technology in the network to increase TCP throughput and improve the TCP-mode VOD experience. The test results show that TCP acceleration is more effective for random drops (packet loss due to poor copper quality) and can guarantee 80% bandwidth TCP throughput under <100MS delay and <10-3 packet loss rate. For BTV and UDP VOD experience, can not solve through the network itself, need to deploy Fcc/ret IPTV platform to enhance the experience.

Second stage: Users grow rapidly and take shape. This stage focuses on the development of video users, resulting in a high local network load, so in addition to user access to network optimization, but also pay attention to user concurrency caused by network congestion problems. In the access segment, you need to lower the olt ratio to build a nonblocking pon network, support all users with 1-channel 4K video through Gpon/epon, and follow the gradual upgrade of 10G Pon to support 1 multi-channel 4K video. In the metropolitan area network, according to the video concurrency peak planning bandwidth, it is recommended to follow 20% concurrent planning bandwidth (that is, the convergence ratio is less than 5:1). With the increase of users, taking into account the central CDN load and the backbone traffic pressure, it is necessary to reduce the urban/backbone expansion pressure by gradually moving down the CDN.

The third stage: already has the large-scale active user. At this stage, the whole metropolitan area Network Video user volume reaches millions, each region/different period of user volume, instantaneous maximum concurrency rate difference is great. If the whole network is based on the unified average concurrency rate for capacity planning, local areas are likely to be congested at some time leading to u-vmos deterioration, while the CDN and network capacity in other regions are in light load, and if the whole network is programmed at the maximum concurrency rate, the network average utilization will be lower. Therefore, this phase should focus on the emergency response and refinement of operations. It is recommended to set up a collection system for network metrics, collect Network load/quality status regularly, and provide big data base for video service platform and network collaboration. When the user on demand, according to the CDN and Network Load situation, the CDN allocates the optimal server, the network controller assigns the optimal network path.

optimization of home network based on U-vmos

Home network is a complaint of the prevalence, according to a carrier video operation and maintenance department feedback, more than 60% of users complained about the final confirmation is the user's home network problems. The current home network WiFi bandwidth is more than 50M, the prevalence of indoor signal intensity is uneven and interference, weak signal or signal interference under the delay, jitter and packet loss rate is high and low bandwidth problems, it can be said that the need to upgrade optimization, specifically from the "coverage, bandwidth and network quality" 3 aspects:

Home Network coverage: in order to be able to reach more than 100M connection speed anywhere in the home, the home WiFi network should be distributed to the centralized deployment, deploying multiple WiFi access points (i.e. APs) to meet the needs of the home multi-terminal in any location of the gigabit network.

Home Bandwidth:The 4K experience requires a home network to deploy the network in accordance with the capabilities of more than 100M, while the current popular home WiFi uses the 802.11n standard (bandwidth actual capacity is less than 50M) to meet business needs. 802.11AC standard WiFi throughput capacity of up to 1000M, recommended home deployment support 802.11AC standard WiFi devices, while supporting the 4K video business, but also for the future support more home business ready.

WiFi signal interference: WiFi needs to consider interference problems, the proposed inter-room interconnection using G.hn PLC scheme, through the wired load reduction of interference between AP; only 1 APs are arranged in each room, each AP set different frequency points to reduce interference between the same band WiFi signal.

video operations centered on the U-vmos experience

Video Services bring new challenges to operations: In addition to connectivity and traffic, focus on end-to-end video experiences and fast fault handling. The traditional operation tools lack the path information of the business flow, can not perceive the experience change in real time, lack the problem of data locating, and can not handle all kinds of faults well. Based on U-vmos's video experience-centric operations, the operation is extended to the user's home terminal, to the details of the business, and the service object extends from the operator to the user, which is a good solution to these problems.

To create a U-vmos experience-centric operation, it is necessary to achieve the whole network information can be perceived and manageable, to achieve real-time measurement of u-vmos indicators. Anticipate the end-to-end experience state in advance with big data analytics, allowing you to dynamically adjust to the optimal network. Change the traditional "problem re-processing" mode, do "Early perception of problems, eliminate problems" and the problem of rapid processing. Operation and maintenance process real-time specific can be divided into the following several steps:

Real-time U-vmos video experience Detection: based on the quality perception of the network device or video client U-vmos-aware module, to the video traffic real-time analysis, when the video indicator deterioration, the initiative to emit u-vmos deterioration alarm. This step can lay the groundwork for active operations.

Real-time video business path restoration: When the user watches the video, the video service probe tracks the business flow information in real time, and realizes the real-time restoration of the single business path through the comprehensive analysis of business flow information and network information.

Real-time video fault location: through the business path real-time restore analysis of each link state, real-time positioning problems.

problem-Solving closed-loop: solve the problem of non-physical connection in real-time with the fault point found in real-time delimitation location, and predict the change of U-vmos index of whole network by the analysis of each single point fault by operation and maintenance system.

Video is the future of various types of Internet services carrier and network traffic King, has gradually become a carrier's basic business, and the best video experience is the key to the success of operators business. U-vmos Video experience standards for the video industry to establish the experience of measurement norms and optimization direction, the industry chain can be combined to optimize the video experience around the experience standard, operators can also be based on U-vmos to create a video experience-centric bearer network, to provide the best video experience for end users. Huawei Ilab has been committed to the improvement and application of U-VMOS standards, and hopes to work with the industry to promote the development of video industry.

(For more information about Huawei, please follow Huawei's developer community, Huawei's own open door: http://developer.huawei.com/cn/ict/, don't ask me what I call, others call me Lei Feng )

Win in the era of big video: Video bearer network based on U-vmos