Mobile Video Service Performance Study Whitepaper Series: 2 Video Coverage A Video-Centric Mobile Network Planning Methodology Building a better connected world October 2015 Contents With the emergence of experience economy, operators are performing a revaluation of priority criteria changing from network centric to experience centric operation. Providing diversified services based on the basic network capability has become one of the primary requirements of operators. More and more people are readily willing to pay for a better online video experience. User experience has become an important way for operators to improve efficiency and enhance user satisfaction and loyalty. This document describes how mobile video services support core functionality and explains the video experience evaluation method used to provide guidelines on how to plan and construct the mobile network based on video experience standard. This document is aimed at providing quality assistance to operators for the creation of differentiated competitiveness through mobile video services. • 1. Introduction.........................................................................01 • 2. Video Industry Insights........................................................02 • 3. User Experience Is the Standard of MBB Network Construction Quality...............................................................05 • 4. Video Coverage Methodology........................................07 • 5. Benefits.................................................................................11 • 6. Appendix: Global vMOS Insight........................................12 »» 1. Introduction • Changed Operation Perspective Experience economy gradually becomes the core strategy and primary source of profit and revenue in the service industry. As basic telecom service providers, operators are altering their perspective from network centric to experience centric operation. Providing diversified services based on the basic network capability has become one of the primary demands of operators, and therefore, they can obtain extra operation value. As a basic service provided by operators, video service is hot. Providing good video service experience is the key to differentiated competitiveness for mobile broadband (MBB) operators. Family Health UGC Video = New Voice TV/ Movie Family Time SNS Sharing Enterprise UC Online Education Vertical Application Public Safety Entertainment Communication • Users' Willingness to Pay User experience is important for improving operators' ROI, which is worth considering even if consumers are price-sensitive. However, consumers are capable of making informed decisions in the selection of products with better service and quality and are willing to pay for better user experience when costs are acceptable. How much extra profit operators can be obtained by providing good user experience? A study about online video services in China shows that: Approximately 36% of participants are willing to pay for better video service experience. Depending on the economic status of the sales region, statistics reveal that users in eastern areas have the freedom to support funding for such products. 01 Users' Willingness to pay for better online video services in China Willing to pay for better video experience (average) 36% Willing to pay for better video experience (eastern areas) Willing to pay for better video experience (middle 38% 32% and western areas) It is fortunate that online video service users in China are willing to pay for better user experience. However, profit is not the only criterion when improving user experience. The predominant purpose of experience operation improvement is to enhance user loyalty by cost-efficient high-quality services as a mechanism to encourage added consumption. • User Experience Standard Building Users are willing to pay for HD, ultra HD, and 3D video experience, but are perhaps unwilling to pay for 4 Mbit/s, 8 Mbit/s, or 16 Mbit/s videos. User experience is an example of subjective perception and cannot be simply assessed by data rate. Mean opinion score (MOS) is used to evaluate voice quality. In the data era, a standard similar to MOS is required to evaluate video quality from various networks and devices in different scenarios. • User Experience-oriented Planning Methodology Complex mobile network planning and construction are operator's fundamental capability and production force. In voice service based network era, with voice quality as the core constituent, a comprehensive planning system including various KPIs is constructed through vigorous preparation, planning, and research. The following questions arise after mobile network entered the data era. How to evaluate network experience in the data era? In addition to bit error rate, access availability, and consistent rate experience in different geographical areas, what other KPIs should be considered to evaluate network experience? How to plan and construct a network based on the preceding KPIs? How do we construct the experience-oriented network planning methodology and apply it to actual networks? »» 2. Video Industry Insights • Longest Video Watching Time Spent on Mobile Devices Users used to watch scheduled TV programs on TV or online videos on PC or laptop. It is apparent from research and conducted studies that among all video watching terminals, users spend most time watching videos on mobile devices. 02 Time spent on mobie devices and tv US Daily Average(Min) Source: Flurry Analytics 180 170 160 150 140 130 120 110 177 168 168 162 158 168 168 109 Q1 2012 Q1 2013 Q1 2014 Q3 2014 Mobile devices have become the primary choice for information acquisition. Commercial enterprises adopt the mobile-preferred strategy, with major news pushed on mobile devices. This strategy causes higher mobile traffic requirement, accelerating the MBB network development. • SNSs Calling for Fast Development of Video Services Mobile Internet leads to not only the vigorous development of social media, but also the emergence of user-generated content (UGC). With UGC, every user can be a content broadcast source. For example, the Periscope enables a user to publish live videos to users all around the world. The author of a content is the broadcast of social media. New video production ways emerge. In mobile Internet era, 44% users use cameras on mobile phones per day. About 90% users use the cameras at least once a week. Simplified video clip APPs, such as the Snapchat, Instagram, and Meipai diversify the UGC propagation. GoPro, the first-person perspective action camera, brings about the boom of extreme-action videography. About 4000 videos are uploaded on the GoPro channel of YouTube. Social live video APPs increase the consumption of mobile data. About 300-hour videos are uploaded to YoTube every minute. Half times of YoTube video watching are over mobile devices. The traffic consumption on mobile devices accounts for 40%. About 75% video watching on Facebook are performed on mobile devices. Social videos develop in a specialized and high-quality direction, allowing social APPs to consume the majority of mobile video traffic. Facebook user`s video sharing and uploading increases dramatically. Times of video watching per day: billion 4.5 4 4 Other video related statistics: 3.5 3 3 50%+ active users watching 1 or more times videos(American) 2.5 53% watched via sharing 2 1.5 1 75% watched via mobile devices 1 0 Q3 14 Q4 14 Q1 15 03 • Mobile HD Video Is the Beginning As technology develops and the cost of HD screens decreases, the resolution of smartphones becomes higher. According to the survey performed by Huawei mLAB on newly delivered smartphones during the first half of 2015, the resolution of 77% of smartphones is higher than 720p and that of the flagship models for each device manufacturer is higher than 1080p. A 720p screen has become the basic configuration of smartphones. The hardware capability of mobile devices has been fully equipped with the ability to support the display and popularity of HD videos of 720p or higher. A large proportion of videos uploaded are higher than 720p, which increase the consumption on videos. For example, the proportion of higher than 720p videos uploaded on YouTube reaches up to 51%. From the perspective of network, the bit rate of a 720p video differs depending on encoding technologies. However, the average bit rate of 720p videos is about 1.5 Mbit/s. Networks with 3 to 5 Mbit/s bandwidth can provide favorable 720 video service experience, with considering the initial waiting delay, handovers in mobile scenarios, and fast fading. The deployment of LTE network in large scale can provide a 5 Mbit/s data rate any time anywhere. Terminal Ready Content Ready Network Ready Terminal with higher than 720P resolution Content with higher than 720P resolution Global MBB throughputs Source: Huawei mLab, 2015 Source: Youtube, 2015 16 77% 51% 55% 38% 23% Source: Open Signal, 2015 10% 18% 31% 15 >5Mbps Countries 14 13 10 9 8 7 5 3 opt10 opt9 opt8 opt7 opt6 opt5 opt4 opt3 opt2 opt1 In high-end field, 2K resolution becomes the competitiveness of leading operators. There are various types of 2K resolution. In the movie projection industry, 2K resolution means 2048 x 1080 pixels. The resolution of iPhone6 Plus is 1920 x 1080, which is very close to 2K resolution. The standard 2K resolution of mobile phones refers to four times of 720p, equal to 2560X1440. Currently, there are 40 types of 2K mobile phones. It is expected that more than 10% newly delivered mobile phones are 2K mobile phones until next year. The 2K video experience will be the target of mobile networks and become the mainstream. • Video Is the Strategy Services, such as videos are required to help operators become the industrial chain leader from pipe services. Voice, instant communication, and a small number of Internet of Things (IoT) are not enough. A large number of future services will be carried on basic videos. After several years of exploration, mature video business models are formed. For example, the South Korean LG U+ binds the main package with self-managing video content, which implements the fast go-to-market and profit of video services. Vodafone cooperates with OTT providers and implements channel service values by binding the large traffic package with OTT content. Verizon acquired AOL to build global digital media platform and ecology. AT&T becomes the leader of PayTV by the acquisition of DirectTV. The global leading operators are moving to the upstream of video industry chain with its ubiquitous high-quality network and videobased service, embracing the Internet and ecology. 04 »» 3. User Experience Is the Standard of MBB network Construction Quality What makes a good MBB network? What qualifies as a positive MBB experience? What is the relationship between user experience and networks? Can user experience be the evaluation standard for network construction? The answers to these questions have continually been researched and discussed. However, these issues still pose challenges to network planning and construction. Investing limited capital to network construction is the ultimate choice for operators' strategy makers. However, there are many disputes over what should be considered the standard. From the KPI system to the KQI system, a comprehensive quality evaluation is available. However, the standard is rarely adopted because it requires complex calculation and KPIs, as well as the actuality that a good KPI result does not bring good user experience. The MBB development, especially the rapid development of video services, shows that user experience of video services can be measured as the network construction standard. The performance of mobile video services directly affects user behavior, which inevitably affects operators' service revenue. Operators can meet the increasing user requirements of video service performance by the evaluation and planning of video services on networks. Mobile network planning is targeted from coverage- and throughputoriented to user-experience-oriented. • Throughput Is Not Equal to User Experience for Video Services Mobile users expect the ability to view video without stalling at any time and place. However, the traditional throughput-based single KPI evaluation system cannot reflect user experience. According to a survey by Huawei mLAB, the top 5 factors affecting mobile video service performance are initial buffering delay, stalling times, stalling duration, video resolution, and the screen size of mobile devices. Top 5 factors affecting mobile video performance initial buffering delay screen size video resolution stalling times stalling duration 2014Q4,mLab consumers survey • Experience-based Planning Target Is the Result of Market Strategy Network planning based on video service experience should consider geographical differences. User video consumption behavior (such as video playback duration, resolution requirements, and tolerable waiting delay) differ from region to region. Therefore, consumption behavior should be analyzed prior to network planning to identify various types of video consumption areas, such as heavy video consumption area or light video consumption area. This is to set appropriate video performance targets for specific areas, mapping the planning objectives and marketing strategies. 05 Urban area: heavy-medium video service consumption area@good experience Stadium:heavy video service consumption area@excellent experience Suburban area: medium video service consumption area@good experience CBD:heavy video service consumption area@excellent experience • Change from Coverage- and Throughput-oriented to UserExperience-oriented Network Planning and Construction Connection experience is the core of the differentiated competitiveness for operators' networks. Ensuring connection experience is the main objective for network planning. Video service performance involves multiple dimensions, such as larger access bandwidth and lower end-to-end delay. However, traditional network planning cannot generate all KPIs and predict experience improvement. Therefore, planning efficiency is low. The video experience-based planning method uses normalized KPIs and displays the planning results visually. As a result, improvements in video experience are evident during the planning. Currently, HD video services have the highest requirements on mobile networks. Therefore, video-based mobile network planning and construction can meet the experience requirements of video services as well as web pages, SNSs, and music services. 06 »» 4. Video Coverage Methodology Faced with MBB network planning challenges, Huawei has introduced methodology based on video coverage experience for mobile network planning and construction. • 4.1 Procedures Video Coverage mainly covers the following three key steps: 1. Network evaluation and target setting: Quantify the mobile video performance based on objective KPIs. Analyze video consumption habits and identify the target planning areas. 2. Gap and root cause analysis: Locate the problems and causes of areas where the video performance does not meet the requirements. 3. Solution implementation and iteration-based optimization: Perform network coverage and capacity planning based on the root causes. The solution includes carrier expansion, sector splitting, site addition, and the deployment of new features. Iteration planning continuously improves the compliance ratio. vMOS baseline Network evaluation and target setting Evaluation and target setting 50*50m grid-level displaying Problem location Gap and root cause analysis Gap analysis Root cause analysis vMOS simulation and forecast Network planning and solution implementation Iteration-based optimal planning Solution implementation • 4.2Network Evaluation and Target Setting The vMOS Is Used for Mobile Video experience Evaluation Back in 2012, the ITU-T released the vMOS evaluation standards for video services. The vMOS references the definition of the speech MOS, monitors video quality loss after the network disconnects, detects video QoE, and locates problems. The streaming media transmission mechanism changes. The adaptive streaming over HTTP replaces the progressive download over HTTP and becomes the main streaming media transmission mechanism. The vMOS standard does not consider the streaming mechanism change and the impacts of different resolution videos on user experience. Therefore, requirements for network construction based on operator experience cannot be met. Huawei believes video evaluation standards centered on user experience should be built to evaluate user experience of various videos from different networks, screens, and scenarios. Together with ergonomic experiments, sample research, and in-depth technical research, Huawei mLAB discovers that the top3 factors applicable to all scenarios are sQualtiy, sLoading, and sStalling. Huawei has put forward a new evaluation standard, the U-vMOS. Therefore, video experience evaluation can be defined, measured, and managed. 07 = f(sQuality, sLoading, sStalling) vMOS Latency<=100ms Stalling Ratio=0% 4.9 4.9 4.8 4.8 vMOS value 5 Excellent 4 Good 3 Fair Slightly annoying 2 Poor Annoying 1 Bad Very annoying 4.5 4.5 4 4 2.94 2.7 2 2.88 Latency=3s Stalling Ratio=10% 2.4 720 1080 2K User perception Imperceptible Perceptible but not annoying 4K Resolution Mapping Video Experience-based Planning with Business Goal Operators also focus on investment income when user experience is improved. Accurate investments in different areas to achieve higher return on investment (ROI) becomes a main business objective for operators. Performance requirements of mobile video services vary in different areas. Based on insights into video traffic distribution, video resolution, and mainstream video APPs, high and medium video consumption areas are discovered. Along with user behavior and operator strategy, the vMOS target is determined to plan the networks in high and medium video consumption areas. Therefore, video service performance is improved, traffic is increased, and more benefits are brought for operators. The video coverage planning strategy is determined based on the insight results. Networks can be divided into the following areas. Characteristics Heavy video consumption area Planning strategy 1080p/2K ultra-HD video quality is planned to build competitiveness. Medium video consumption area 720p HD video quality is planned everywhere to stimulate video consumption. Light video consumption area 720p HD video quality is planned everywhere to cultivate video consumption habits. • 4.3 Gap and Root Cause Analysis Grid-level Visualized Analysis Based on the video performance evaluation on the live network, the vMOS value is geographically displayed as a 50 x 50 m grid. If the value is lower than the target vMOS value, a grid not complying with the requirements is formed. In this case, root cause analysis and specified network planning are performed. 08 Target vMOS=3.8 2.7 and below 2.8~3.2 3.3~3.7 3.8 and above Conversion and Mapping Between the Video Experience Target and Traditional Network KPIs The traditional network KPIs, such as coverage, interference, and number of active users cannot be directly used for measuring video performance. By analyzing mass data on the live network, Huawei converts these traditional KPIs to vMOS KPIs of video performance using data mining. As a result, a precise mapping model of traditional KPIs to vMOS KPIs is formed. vMOS mapping model Coverage Interference Active users number PRB/TCP resource This process covers several important steps: identifying factors, data clustering and classifying, key feature selection, and fitting to a model. Firstly, find out network KPIs related to the vMOS. Then, select the most relevant data to fit the relationship curve between the vMOS and traditional network KPIs. Finally, the best vMOS mapping model is formed. 09 Identifying factors Data clustering and classifying Key feature selection Fitting to a model For example, for 720p videos, interference, transmitted carrier power, and the number of active users are used for fitting to the initial buffer delay model. This model maps traditional network KPIs and the vMOS. Then, the vMOS is used for network planning and construction. er time Initial buff o ati er ois ln na Sig r we po r e i r car ted t i nsm Tra • 4.4 Solution Implementation and Iteration Planning Solution Implementation The Video Coverage method is used to perform coverage, capacity, and network structure optimization based on the root cause of each grid value lower than the vMOS value. Optimization includes capacity expansion, sector splitting, site addition, deployment of new features, and CDN sinking to ensure that each area reaches the preset video performance goal. Edge performance improvement introduces inter-site carrier aggregation and adaptive inter-cell interference coordination to improve cell edge coverage. Indoor performance improvement deploys micro base stations in office buildings and shopping malls to improve the indoor capacity. The end-to-end delay optimization deploys TCP optimization, video scheduling, adaptive bit rate, and CDN sinking to reduce initial buffering delay. Iteration Planning Achieves Continuous Improvement The vMOS scores are the final target. Network planning stops until the vMOS scores meet requirements. Then, the planning simulation is determined. 10 Before planning »» Iteration planning simulation1: Grids Iteration planning simulation2: Grids compliance rate>75% compliance rate>90% 5. Benefits Iteration optimization is used to reduce network planning times and costs. Therefore, network planning efficiency is improved and network construction costs are reduced. Precise planning implements the mapping between the market objectives and network construction. In the previous network construction, channels were built. However, user experience has become the target objective now. With precise planning, the Video Coverage method establishes mapping between video experience and network KPIs. Therefore, a certain standard of video quality and performance can be expected. As a result, optimal mapping between business objectives and network construction is achieved and profits are increased. Network Construction Case A leading operator quickly learns the importance of video services in the MBB era. However, this operator's video score ranked lower than its competitors during a third-party test due to long delay and stalling time, leading to a poor user experience. This operator adopts Huawei Video Coverage methodology and sets the target vMOS value to 3.5 in key areas (over 90% grids meet the requirement). After the solution is deployed, the operator reaches its target. Areas with vMOS scores lower than 3.0 are eliminated and video performance is greatly enhanced. As a result, data traffic is increased, with the average consumption per user increasing from 1.3 GB to 1.5 GB and operator`s profit is increased. Before: After: vMOS: 2.9 20% 38% vMOS: 2.2 1.3GB vMOS<3 grid rate video traffic rate average traffic vMOS: 3.5 5% 50% 1.5GB vMOS: 3.1 11 »» 6. Appendix: Global vMOS Insight User experience is much less satisfactory. The aim of LTE networks is to provide videos with 4.0 vMOS scores. To study the present state of video service experience, Huawei mLAB uses the Speed video client to select samples of 4G mobile networks in seven cities around the world. The study shows that the average video experience vMOS score on most networks is lower than 3.5 with unsatisfactory user experience. The evaluation score of stalling is higher than 4.0. The evaluation score of initial buffer delay is lower than 3.5. Therefore, improving initial buffering delay is the key to improving user experience. Test APP:SpeedVideo (Downloaded at http://mbblab.tk, supported by Huawei mLAB) Test Time:2015.7-2015.9 Test Results: Global vMOS Test Result Source: Huawei mLAB Test samples: sampling video experience data from seven cities in the world. Sampled video are YouTube 720 HPD, YouTube 1080P Dash and YouTube 2K Dash. The initial buffer time for HPD video is 8 seconds, and for Dash video is 2 seconds. City Video Streaming Samples Seoul YouTube 2K* -DASH 4228 Dubai YouTube 1080p -DASH 2459 Mexico City YouTube 1080p- DASH 1883 Sydney YouTube 1080p- DASH 107 Riyadh YouTube 1080p- DASH 100 Chon Buri YouTube 720p- HPD 190 Rio de Janeiro YouTube 720p- HPD 207 * YouTube 2K resolution is 2560*1440. 12 End-to-end (E2E) round trip time (RTT) becomes the constraints of video experience. The E2E RTT for a good video experience on LTE network is less than 40 ms and that on UMTS network is less than 80 ms. Huawei mLAB finds that for the same video initial buffer delay, different E2E RTTs correspond to different bandwidth requirements. The shorter the initial buffer delay is, the greater the E2E RTT impact is and the greater bandwidth gains the RTT brings. To achieve better video experience, LTE network should support the smooth playing of videos higher than 1080p and the E2E RTT should be less than 40 ms. UMTS network should support the smooth playing of 720p videos and the E2E RTT should be less than 80 ms. The network capability requirements are as follows when the target initial buffer delay is two seconds. Target Initial Buffering Time = 2S Mbps 30 27.2 25 20 15 10 5 0 10.6 8.4 6.7 5 4.1 4.4 3.1 2.4 2.1 720p E2E Latency 14.6 10.6 1080p 40ms 60ms 2k 80ms 100ms Reference Mobile Video Service Performance Research, http://www.huawei.com/minisite/hwugm15/mobile-vmos.html 13 Copyright © 2015 Huawei Technologies Co., Ltd. All rights reserved. No part of this document may be reproduced or transmitted in any form or by any means without prior written consent of Huawei Technologies Co., Ltd. Trademarks and Permissions and other Huawei trademarks are the property of Huawei Technologies Co., Ltd. All other trademarks and trade names mentioned in this document are the property of their respective holders. Notice The product, service, or feature that you purchase should be restricted by the Huawei commercial contract and the clauses in the contract. All or a part of products, services, or features described in this document may not be purchased or used. Every effort has been made in the preparation of this document to ensure the accuracy of the contents, but the statements, information, and recommendations in this document do not constitute a warranty of any kind, expressed or implied. The information in this document is subject to change without notice. Every effort has been made in the preparation of this document to ensure the accuracy of the contents, but the statements, information, and recommendations in this document do not constitute a warranty of any kind, expressed or implied. Huawei Technologies Co., Ltd. Address: Huawei Industrial Base Bantian, Longgang Shenzhen 518219 People's Republic of China Website: http://www.huawei.com Email: support@huawei.com