CN107888579A - A kind of mobile video user experience quality index modeling method of non-interfering type - Google Patents
A kind of mobile video user experience quality index modeling method of non-interfering type Download PDFInfo
- Publication number
- CN107888579A CN107888579A CN201711077314.1A CN201711077314A CN107888579A CN 107888579 A CN107888579 A CN 107888579A CN 201711077314 A CN201711077314 A CN 201711077314A CN 107888579 A CN107888579 A CN 107888579A
- Authority
- CN
- China
- Prior art keywords
- video
- classification
- feature
- model
- actual value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/80—Responding to QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
- H04L65/60—Network streaming of media packets
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
A kind of mobile video user experience quality index modeling method of non-interfering type, step are：The video for playing HPD patterns and HLS patterns is clicked on mobile phone, catches the flow information of these videos, while the interim card situation of each video is recorded using interim card analysis tool, and the actual value of marking video transmission mode classification；By analyzing video cardton situation, the actual value of video cardton degree is obtained；By analyzing video flow, the actual value of video definition and the network characterization of flow are obtained；To the video of two kinds of transmission modes, machine learning is carried out using the true classification of the network characterization and transmission mode of video flow, trains the model that can be used for transmission mode classification；To the video of each transmission mode, machine learning is carried out using the network characterization and user experience quality index actual value of video flow respectively, trains the model that can be used for evaluating experience quality of user index.
Description
Technical field
The present invention relates to the HPD analogue videos for the mobile terminal that can be used in the case of encryption and regarding for HLS analogue videos
Keep pouring in defeated method for classifying modes, and the modeling method per class video user Quality of experience index.
Background technology
In recent years, as people watch the growth of the demand of video on mobile phone, major video website is all proposed movement
The video player at end, and take different transmission of video pattern.The most commonly used transmission of video pattern includes HPD
And HAS.
HPD is HTTP progressive download, that is, a kind of simply from HTTP WEB servers progress file download
Common mode." progressive " this term comes from when player client allows media file also to download and begins to broadcast
Put, after all completing to write on disk when whole file download, it is all first to be placed directly in play content under normal circumstances
In the caching of browser.Support HTTP1.1 standards client can by WEB server carry out bytes range request come
It is addressed to the relevant position for the media file for not downloading completion.Client also takes the pattern that shunting is transmitted, in a video
The transmission of another stream is begun to when streaming is also unclosed.
HAS is HTTP code rate self adaptation stream, and a key of HAS technologies is exactly the cutting whole into sections of media data, Mei Gefen
The time span of block is identical, general 2~10 seconds.In video coding layer, it means that each piecemeal is completely regarded by several
Frequency GOP forms (each piecemeal has a crucial I frame), with this ensure each piecemeal and the media piecemeal in past and future without
Association.Media piecemeal is stored in HTTP Web servers, and client is in a linear fashion to Web server request media point
Block, and the download of media piecemeal is carried out in a manner of traditional HTTP, when media piecemeal is downloaded to client, client is according to suitable
Sequence plays this series of media piecemeal.HLS is the HAS total solution of Apple Inc., and it is made up of three parts：Server group
Part, distributed components and client.First, encoder receives audio frequency and video input, and using H.264 coding techniques, exports MPEG-
2TS flows, and is then spaced according to set time using Slice Software and TS files one by one are cut and saved as to TS code streams.This
A little TS files are deployed on Web server, and Slice Software further creates the index text comprising these TS file-related informations simultaneously
Part, index file are saved as .M3U8 files.The URL of index file is issued on Web server, and client reads index text
Part, then ask media file to server in order and without a pause show them.
To use network traffics to distinguish HPD patterns and HLS patterns, have a relatively simple method, i.e., in network traffics
The middle GET request for finding index file (.M3U8 files).If this request can be found, then video is passed with HLS patterns
It is defeated, otherwise transmitted with HPD patterns.But this sorting technique needs to use URL information, once it is encrypted to video
Afterwards, for the information in URL with regard to invisible, this method is naturally also no longer feasible.
The user experience quality of video is mainly made up of two indices --- definition and interim card situation.To video definition
Obtain the information that can also be used in GET request URL, but this method is not suitable for encryption situation equally.And to video
The acquisition of interim card situation, can not be for HPD patterns and the general method in network-side of HLS patterns before.
The content of the invention
The present invention will overcome the disadvantages mentioned above of prior art, there is provided a kind of mobile terminal that can be used in the case of encryption regards
The sorting technique of the HPD and HLS transmission of video patterns of frequency, and the modeling method of the user experience quality index per class video.
To realize object above, the technical solution used in the present invention is：A kind of mobile video user's body of non-interfering type
Check the quality figureofmerit modeling method, comprise the following steps：
(1) clicked on mobile terminal and play the video of HPD patterns and the video of HLS patterns, in video flow logging modle
The middle flow information for recording these videos, at the same in video cardton situation logging modle record video interim card situation；
(2) model for video mode classification is trained, including：
21) to each video, in transmission of video mode flag module transmitting classification to it is marked；
22) suitable feature is extracted in the flow information of each video；
23) machine learning is carried out in the classification in 1) and the feature in step 22) using each video, training can be with
Model for classification；
(3) to the video of each transmission mode, train for evaluating video user Quality of experience index --- interim card feelings
The model of condition, including：
31) classification that each video cardton situation is obtained in video cardton condition category actual value analysis module is true
Value；
32) extracted in user experience quality index classification characteristic extracting module useful in the flow information of each video
Network characterization, and it is extended using statistical nature；
33) feature selecting is carried out using classification of each video in step 31) and the feature in step 32)；
34) machine learning, training are carried out using classification of each video in step 31) and the feature in step 33)
Go out the model that can be used for evaluation；
(4) to the video of each transmission mode, train for evaluating video user Quality of experience index --- definition
The model of situation, including：
1) the classification actual value of each video definition is obtained in video definition classification actual value extraction module；
2) extracted in user experience quality index classification characteristic extracting module useful in the flow information of each video
Network characterization, and it is extended using statistical nature；
3) feature selecting is carried out using classification of each video in step 41) and the feature in step 42)；
4) machine learning is carried out using classification of each video in step 41) and the feature in step 43), trained
It can be used for the model of evaluation；
Compared with prior art, the beneficial effects of the invention are as follows：In the model come using training, all measurements refer to
Mark is obtained from network-side, it is not necessary to any change and deployment are made in client, user will not be interfered；All surveys
Figureofmerit obtains below application layer, even if video flow is encrypted, the method stands good.
Brief description of the drawings
Fig. 1 is the video cardton situation of the present invention, definition evaluation rubric figure.
Fig. 2 is the training process of video cardton situation of the invention, definition evaluation model and transmission mode disaggregated model
Schematic diagram.
Embodiment
The present invention is using the network characterization below application layer to the HPD analogue videos of mobile terminal and the biography of HLS analogue videos
Defeated pattern is classified, and provides a kind of video user Quality of experience index modeling side for being applied to two kinds of transmission modes
Method, this method is equally only using the network characterization below application layer, therefore both approaches are applied to encryption situation.Specifically
Step is as follows：
(1) video network stream information and client screenshotss information are captured：
(1.1) the continuous one week video that HPD patterns are played in Android mobile phone, video include SD, high definition, super clear three kinds
Definition, and network bandwidth is controlled simultaneously so that the playing process of different video has different interim card situations.In the broadcasting of video
During using wireshark crawl video flow, and with appetizer-toolkit instruments every 0.5s interception Mobile phone screen
Curtain.It is finally HPD patterns by these video markers.
(1.2) play within continuous one week HLS analogue videos in Android mobile phone, video include SD, high definition, super clear three kinds it is clear
Clear degree, and network bandwidth is controlled simultaneously so that the playing process of different video has different interim card situations.In the broadcasting of video
Using wireshark crawl video flows in journey, and Mobile phone screen is intercepted every 0.5s with appetizer-toolkit instruments
Curtain.It is finally HLS patterns by these video markers.
(2) model for video mode classification is trained；
(2.1) to each video, category label is carried out to its transmission mode；
(2.2) all stream information and video block messages of each video are found from the flow information being truncated to, are recorded
The average time started intervals of preceding 3 stream, the average time started intervals of all streams, the ratio of video block number and fluxion.
(2.3) machine learning is carried out using classification of each video in (2.1) and the feature in (2.2), trained
It can be used for the model of classification, the most suitable algorithm is DecisionTable；
(3) train for evaluating video user Quality of experience index --- the model of interim card situation, the video of each pattern
Train a single model；
(3.1) to each video, video duration and interim card duration are obtained according to its sectional drawing, if interim card duration is with video
Long ratio is 0, is without interim card, if being less than or equal to 0.1 more than 0, labeled as slight interim card, if being more than by this video marker
0.1, then labeled as serious interim card；
(3.2) suitable 19 features (as shown in table 1) are extracted in the flow information of each video, it is special to each
Sign is extended using 7 statistical natures (as shown in table 2) to it, finally gives 133 network flow characteristics；
(3.3) classification of each video in (3.1) and the feature in (3.2) are used, is commented using CfsSubsetEval
Valency strategy and BestFirst search strategies carry out feature selecting, to remove influence of the uncorrelated features to classification results；
(3.4) machine learning is carried out using classification of each video in (3.1) and the feature in (3.3), trained
It can be used for the model of evaluation, for HPD analogue videos and HLS analogue videos, the most suitable algorithm is all
RandomSubSpace；
(4) train for evaluating video user Quality of experience index --- the model of definition situation, each pattern regard
Frequency trains a single model；
(4.1) GET request to video content in each video network flow is extracted, by analyzing the letter in its URL
Breath obtains the classification actual value of this video definition；
(4.2) suitable 19 features (as shown in table 1) are extracted in the flow information of each video, it is special to each
Sign is extended using 7 statistical natures (as shown in table 2) to it, finally gives 133 network flow characteristics；
The video flow feature of table 1
The statistical nature of table 2
| Maximum |
| Minimum value |
| Average value |
| Standard deviation |
| 25th hundredths |
| 50th hundredths |
| 75th hundredths |
(4.3) classification of each video in (4.1) and the feature in (4.2) are used, is commented using CfsSubsetEval
Valency strategy and BestFirst search strategies carry out feature selecting, to remove influence of the uncorrelated features to classification results；
(4.4) machine learning is carried out using classification of each video in (4.1) and the feature in (4.3), trained
It can be used for the model of evaluation, for HPD analogue videos, the most suitable algorithm RandomForest, regarded for HLS patterns
Frequently, algorithm the most suitable is MultiplayerPerception.
Content described in this specification embodiment is only enumerating to the way of realization of inventive concept, protection of the invention
Scope is not construed as being only limitted to the concrete form that embodiment is stated, protection scope of the present invention is also and in art technology
Personnel according to present inventive concept it is conceivable that equivalent technologies mean.
Claims (1)
1. a kind of mobile video user experience quality index modeling method of non-interfering type, comprises the following steps：
(1) clicked in mobile terminal and play the video of HPD patterns and the video of HLS patterns, recorded in video flow logging modle
The flow information of these videos, at the same in video cardton situation logging modle record video interim card situation；
(2) model for video mode classification is trained, including：
21) to each video, in transmission of video mode flag module transmitting classification to it is marked；
22) suitable feature is extracted in the flow information of each video；
23) machine learning is carried out in the classification in 21) and the feature in 22) using each video, trains and can be used for point
The model of class；
(3) to the video of each transmission mode, train for evaluating video user Quality of experience index --- interim card situation
Model, including：
31) the classification actual value of each video cardton situation is obtained in video cardton condition category actual value analysis module；
32) network useful in the flow information of each video is extracted in user experience quality index classification characteristic extracting module
Feature, and it is extended using statistical nature；
33) feature selecting is carried out in the classification in 31) and the feature in 32) using each video；
34) machine learning is carried out in the classification in 31) and the feature in 33) using each video, trains and can be used for commenting
The model of valency；
(4) to the video of each transmission mode, train for evaluating video user Quality of experience index --- definition situation
Model, including：
41) the classification actual value of each video definition is obtained in video definition classification actual value extraction module；
42) network useful in the flow information of each video is extracted in user experience quality index classification characteristic extracting module
Feature, and it is extended using statistical nature；
43) feature selecting is carried out in the classification in 41) and the feature in 42) using each video；
44) machine learning is carried out in the classification in 41) and the feature in 43) using each video, trains and can be used for commenting
The model of valency.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711077314.1A CN107888579B (en) | 2017-11-06 | 2017-11-06 | Non-interference mobile video user experience quality index modeling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711077314.1A CN107888579B (en) | 2017-11-06 | 2017-11-06 | Non-interference mobile video user experience quality index modeling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107888579A true CN107888579A (en) | 2018-04-06 |
| CN107888579B CN107888579B (en) | 2020-06-09 |
Family
ID=61778760
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711077314.1A Active CN107888579B (en) | 2017-11-06 | 2017-11-06 | Non-interference mobile video user experience quality index modeling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107888579B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109905696A (en) * | 2019-01-09 | 2019-06-18 | 浙江大学 | A kind of recognition methods of the Video service Quality of experience based on encryption data on flows |
| CN110225417A (en) * | 2019-05-09 | 2019-09-10 | 网宿科技股份有限公司 | Data processing method and server, the method and server that detect Caton |
| CN110958491A (en) * | 2018-09-27 | 2020-04-03 | 中兴通讯股份有限公司 | Video Kanton model training method, video Kanton model prediction method, server and storage medium |
| CN111107423A (en) * | 2018-10-29 | 2020-05-05 | ***通信集团浙江有限公司 | Video service playing card pause identification method and device |
| CN113395602A (en) * | 2021-06-25 | 2021-09-14 | 合肥工业大学 | Modeling method for user experience quality QoE in adaptive point cloud video streaming media |
| WO2022116811A1 (en) * | 2020-12-04 | 2022-06-09 | 武汉绿色网络信息服务有限责任公司 | Method and device for predicting definition of video having encrypted traffic |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103888846A (en) * | 2014-03-04 | 2014-06-25 | 浙江大学 | Wireless video streaming service self-adaption rate control method based on QoE |
| US20160365120A1 (en) * | 2015-06-11 | 2016-12-15 | Eran Steinberg | Video editing system for generating multiple final cut clips |
| US20170213111A1 (en) * | 2016-01-21 | 2017-07-27 | Wizr Llc | Video processing |
| CN106998322A (en) * | 2017-02-20 | 2017-08-01 | 南京邮电大学 | A kind of stream sorting technique of the Mean Opinion Score characteristics of mean of use video traffic |
| CN107071399A (en) * | 2017-04-26 | 2017-08-18 | 华为技术有限公司 | The method for evaluating quality and device of a kind of encrypted video stream |
-
2017
- 2017-11-06 CN CN201711077314.1A patent/CN107888579B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103888846A (en) * | 2014-03-04 | 2014-06-25 | 浙江大学 | Wireless video streaming service self-adaption rate control method based on QoE |
| US20160365120A1 (en) * | 2015-06-11 | 2016-12-15 | Eran Steinberg | Video editing system for generating multiple final cut clips |
| US20170213111A1 (en) * | 2016-01-21 | 2017-07-27 | Wizr Llc | Video processing |
| CN106998322A (en) * | 2017-02-20 | 2017-08-01 | 南京邮电大学 | A kind of stream sorting technique of the Mean Opinion Score characteristics of mean of use video traffic |
| CN107071399A (en) * | 2017-04-26 | 2017-08-18 | 华为技术有限公司 | The method for evaluating quality and device of a kind of encrypted video stream |
Non-Patent Citations (2)
| Title |
|---|
| Y.T.LIN, E.M.R.OLIVEIRA,ET AL: "Machine Learning for Predicting QoE of Video", 《IEEE ICC 2017 COMMUNICATIONS QOS, RELIABILITY, AND MODELING SYMPOSIUM》 * |
| 钱力言: "无线视频流业务的OoE评估模型及其在传输控制中的应用", 《信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110958491A (en) * | 2018-09-27 | 2020-04-03 | 中兴通讯股份有限公司 | Video Kanton model training method, video Kanton model prediction method, server and storage medium |
| CN111107423A (en) * | 2018-10-29 | 2020-05-05 | ***通信集团浙江有限公司 | Video service playing card pause identification method and device |
| CN109905696A (en) * | 2019-01-09 | 2019-06-18 | 浙江大学 | A kind of recognition methods of the Video service Quality of experience based on encryption data on flows |
| CN110225417A (en) * | 2019-05-09 | 2019-09-10 | 网宿科技股份有限公司 | Data processing method and server, the method and server that detect Caton |
| CN110225417B (en) * | 2019-05-09 | 2022-06-10 | 网宿科技股份有限公司 | Data processing method and server, and method and server for detecting stuck state |
| WO2022116811A1 (en) * | 2020-12-04 | 2022-06-09 | 武汉绿色网络信息服务有限责任公司 | Method and device for predicting definition of video having encrypted traffic |
| CN113395602A (en) * | 2021-06-25 | 2021-09-14 | 合肥工业大学 | Modeling method for user experience quality QoE in adaptive point cloud video streaming media |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107888579B (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107888579A (en) | A kind of mobile video user experience quality index modeling method of non-interfering type | |
| CN108696403A (en) | A kind of encrypted video QoE evaluating methods based on the study of network flow latent structure | |
| CN103797483B (en) | Methods and systems for identifying content in data stream | |
| CN107241645B (en) | Method for automatically extracting goal wonderful moment through caption recognition of video | |
| CN105809190B (en) | A kind of SVM cascade classifier methods based on Feature Selection | |
| CN106454748B (en) | A kind of Children in Kindergartens video and data collection and transmission | |
| WO2017001607A1 (en) | Method and device for creating a database | |
| CN103797482A (en) | Methods and systems for performing comparisons of received data and providing follow-on service based on the comparisons | |
| DE602005011611D1 (en) | METHOD AND DEVICE FOR AWARDING INFORMAT | |
| CN111212317A (en) | Skip navigation method for video playing | |
| CN104869430B (en) | A kind of video speed player method and device | |
| US20170178692A1 (en) | Emotional timed media playback | |
| CN103327045A (en) | User recommendation method and system in social network | |
| DE112008001040T5 (en) | Wireless distribution of audio files and related information | |
| CN1719909A (en) | Method for measuring audio-video frequency content change | |
| DE102008044635A1 (en) | Apparatus and method for providing a television sequence | |
| RU2014151717A (en) | TRANSMISSION / RECEIVING DEVICE, METHOD AND DEVICE FOR CODING / DECODING | |
| CN107231581A (en) | Method, system and streaming media playing for video playback control server | |
| CN109511011A (en) | A kind of finger print data base construction method towards YouTube DASH encrypted video | |
| CN107566851A (en) | A kind of video segment storage method and system accessed applied to media data | |
| CN106998322A (en) | A kind of stream sorting technique of the Mean Opinion Score characteristics of mean of use video traffic | |
| CN105530523B (en) | A kind of service implementation method and equipment | |
| US20200097852A1 (en) | Systems and methods for detecting and grouping anomalies in data | |
| CN107943811A (en) | The dissemination method and device of content | |
| CN106372653A (en) | Stack type automatic coder-based advertisement identification method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |