CN105897736A - Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service - Google Patents
Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service Download PDFInfo
- Publication number
- CN105897736A CN105897736A CN201610326745.6A CN201610326745A CN105897736A CN 105897736 A CN105897736 A CN 105897736A CN 201610326745 A CN201610326745 A CN 201610326745A CN 105897736 A CN105897736 A CN 105897736A
- Authority
- CN
- China
- Prior art keywords
- user
- interbehavior
- video
- assessed
- performance
- 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.)
- Pending
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
- H04L65/60—Network streaming of media packets
- H04L65/65—Network streaming protocols, e.g. real-time transport protocol [RTP] or real-time control protocol [RTCP]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The invention discloses a method and a device for assessing quality of experience (QoE) of a TCP (Transmission Control Protocol) video stream service, and aims to monitor and record user interaction behaviors including pause, fast forward, backward and a selected resolution during watching of a video to be assessed on a client. The method comprises the following steps: acquiring a plurality of performance quantitative indexes representing the playing smoothness, content attraction degree and picture definition of the video to be assessed according to network layer performance, application layer performance and user interaction behaviors during watching of the video to be assessed on the client; inputting the performance quantitative indexes into a BP (Back Propagation) neural network model to obtain a subjective experience MOS (Mean Opinion Score) value; and determining the quality of experience (QoE) of the video to be assessed according to the MOS value. Through application of the scheme provided by the invention, an obtained assessment result can be closer to real experience of a user; the QoE assessment accuracy of the TCP video stream service is increased; and a good assessment effect is achieved.
Description
Technical field
The present invention relates to video stream traffic user experience quality assessment technology field, particularly relate to a kind of TCP
Video stream traffic user experience quality appraisal procedure and device.
Background technology
Along with the fast development of the Internet and emerging in multitude of media information, video stream traffic is increasingly becoming net
One of topmost business of network service provider.Meanwhile, along with the fast development of intelligent terminal, pass through
The user of network viewing video also gets more and more.User when watching Internet video, be not intended merely to video clear,
Smooth, it is also desirable to the content of video can be enriched interesting, and the satisfaction of Video service is clear, smooth, interior
That holds considers, i.e. user is more and more higher to the requirement of Video service.Therefore, in order to strive for more using
Family, obtains user's satisfaction to video stream traffic, improves video stream traffic quality, beyond doubt video flowing industry
The key that business provider is successful in industry competition.
At present, user is characterized to regarding usually through user experience quality (Quality of Experience, QoE)
Frequency flows the satisfaction of business.Compared to traditional service quality (Quality of Service, QoS), QoE
Consider the subjective factors of user, closer to the sense of reality of user, it as user at certain objective ring
To the service used or the overall degree of recognition of business in border, seem for service or service provider
Particularly important.Therefore, the appraisal procedure of research video stream traffic QoE has the biggest realistic meaning.
Such as, the application for a patent for invention of Application No. 201410325896.0, disclose a kind of TCP video
The QoE training of stream business and the scheme of assessment, the detailed process of the program is: according to video stream traffic
The transfer rate of network performance, such as video, draws video stream traffic network QoS, and sets up network performance
Index system;Initial buffering time according to the application layer performance of video stream traffic, such as video, heavily buffer
Frequency and averagely buffer duration, draws video stream application layer QoS, and sets up the application layer of video stream traffic
Can index system;Set up the mapping function of video stream traffic network QoS and video stream application layer QoS;Root
Test factually and determine the average suggestion of Consumer's Experience (Mean Opinion Score, MOS) value, draw QoE,
Wherein, the subjective scoring of video stream traffic is determined by MOS value according to the experimenter invited in experiment;Build
The mapping relations of vertical video stream application layer QoS Yu QoE;According to video stream traffic network QoS and video flowing
The mapping function of application layer QoS, and according to the mapping relations of video stream application layer QoS Yu QoE, set up
Video stream traffic network QoS and the mapping model of QoE;Finally, according to framework for network performance metrics, answer
With layer train diagram adjusting and video stream traffic network QoS and the mapping model of QoE, carry out QoE assessment.
But, the QoE evaluation scheme disclosed in above-mentioned patent application, only considered the net of video stream traffic
Network layers performance and application layer performance, less consideration user watches impression during video, and user watches video
Time interbehavior, this does not meets real application scenarios.User's interbehavior and user are experienced as video industry
A unavoidable important factor in order in business stream QoE assessment, in video traffic QoE assesses,
If do not considered user's interbehavior and user's impression, the true body obtaining QoE assessment result with user can be made
Testing and produce bigger deviation, Evaluated effect is poor.
Summary of the invention
The purpose of the embodiment of the present invention is to provide a kind of TCP video stream traffic user experience quality assessment side
Method and device, to improve the accuracy of QoE assessment so that QoE assessment result and the actual experience of user
More closely, raising Evaluated effect.
To achieve these goals, the embodiment of the invention discloses a kind of TCP video stream traffic user's body to check the quality
Amount appraisal procedure, described method includes:
S101, monitor and record client watch video to be assessed time user's interbehavior, including suspend,
F.F., retrogressing and selected resolution;
S102, according to described client watch video to be assessed time network layer performance, application layer performance and institute
Stating user's interbehavior, the broadcasting fluency, content draws and the picture that obtain sign video to be assessed are clear
Multiple performance quantizating index of degree;
S103, by each described performance quantizating index input BP neural network model, it is thus achieved that user's subjective experience
MOS value;
S104, according to described MOS value, determine the user experience quality QoE of video to be assessed.
Preferably, after performing step S101, and before performing step S102, described method also includes:
Screen described user's interbehavior.
Preferably, described screening described user interbehavior includes:
Screen and delete the random user interbehavior in described user's interbehavior;
And/or, screen and delete the small probability user's interbehavior in described user's interbehavior.
Preferably, the determination method of described random user interbehavior includes:
The effective time scope of user's interbehavior is determined according to the first function;Wherein, described first function is:δ is described effective time scope, tstartAnd tendRespectively
Represent the moment commencing play out and playing end of video to be assessed, tiThe moment occurred for user's interbehavior,
T is default time range;
Determine the output valve of the second function according to described effective time scope, the output of described second function will be made
It is worth null user's interbehavior and is defined as random user interbehavior;Wherein, described second function is:DiIt is ti-δ arrives tiIn+δ time range, niThe moment t that individual user's interbehavior occursa+jWith
tiRelative distance difference d(a+j)iSum;
The determination method of described small probability user's interbehavior includes:
When described user's interbehavior is the small probability user's interbehavior adding up acquisition in advance, it is defined as little
Probability user's interbehavior.
Preferably, described BP neural network model is made up of input layer, hidden layer and output layer, input layer
The number of node is 8, and the number of hidden layer node is 13, and the number of output layer node is 1.
Preferably, the performance quantizating index playing fluency of described sign video to be assessed includes: initially delay
Rush the time, averagely buffer duration, heavily buffer frequency, average suspend duration and suspend frequency;
The performance quantizating index of the content draws of described sign video to be assessed includes: F.F. ratio, retrogressing
Ratio;
The performance quantizating index of the image sharpness of described sign video to be assessed includes: video average resolution
Rate.
The embodiment of the invention also discloses a kind of TCP video stream traffic user experience quality apparatus for evaluating, described
Device includes: user's interbehavior monitoring modular, performance quantizating index acquisition module, MOS value computing module
And evaluation module,
Described user's interbehavior monitoring modular, for monitoring and record when client watching video to be assessed
User's interbehavior, including: time-out, F.F., retrogressing and selected resolution;
Described performance quantizating index acquisition module, for watching net during video to be assessed according to described client
Network layers performance, application layer performance and described user's interbehavior, obtain the broadcasting smoothness characterizing video to be assessed
Degree, content draws and multiple performance quantizating index of image sharpness;
Described MOS value computing module, for inputting BP neutral net mould by each described performance quantizating index
Type, it is thus achieved that user's subjective experience MOS value;
Described evaluation module, for according to described MOS value, determines the user experience quality of video to be assessed
QoE。
Preferably, described device also includes: user's interbehavior screening module, for mutual to described user
Behavior monitoring module monitors to user's interbehavior screen, and be transferred to described performance quantizating index and obtain
Delivery block.
Preferably, described user's interbehavior screening module, including:
Random user interbehavior screening unit, random for screen and delete in described user's interbehavior
User's interbehavior;
Small probability user's interbehavior screening unit is little for screen and delete in described user's interbehavior
Probability user's interbehavior.
Preferably, described random user interbehavior screening unit, including:
Effective time range determination submodule, during for determining user's interbehavior effective according to the first function
Between scope;Wherein, described first function is:δ is
Described effective time scope, tstartAnd tendRepresent respectively video to be assessed commence play out and play end time
Carve, tiIn the moment occurred for user's interbehavior, T is default time range;
Random user interbehavior determines submodule, for determining the second function according to described effective time scope
Output valve, the output valve null user interbehavior making described second function is defined as random user
Interbehavior, and delete this random user interbehavior;Wherein, described second function is:
DiIt is ti-δ arrives tiIn+δ time range, niThe moment t that individual user's interbehavior occursa+jWith tiRelative distance
Difference d(a+j)iSum.
A kind of TCP video stream traffic user experience quality appraisal procedure of embodiment of the present invention offer and device,
Can monitor and record client watch video to be assessed time user's interbehavior, including time-out, F.F.,
Retreat and selected resolution;Watch network layer performance during video to be assessed according to described client, answer
With layer performance and described user's interbehavior, obtain and characterize the broadcasting fluency of video to be assessed, content attraction
Degree and multiple performance quantizating index of image sharpness;Each described performance quantizating index is inputted BP nerve net
Network model, it is thus achieved that user's subjective experience MOS value;According to described MOS value, determine the use of video to be assessed
Family Quality of experience QoE.The scheme provided due to the present invention, not only allows for Internet and application layer performance refers to
The mark impact on TCP video stream traffic QoE, it is also contemplated that the impact of user's interbehavior so that acquisition
The actual experience of assessment result and user closer to, improve the accuracy of TCP video stream traffic QoE assessment,
Evaluated effect is good.Certainly, arbitrary product or the method for implementing the present invention must be not necessarily required to reach above simultaneously
Described all advantages.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to enforcement
In example or description of the prior art, the required accompanying drawing used is briefly described, it should be apparent that, describe below
In accompanying drawing be only some embodiments of the present invention, for those of ordinary skill in the art, do not paying
On the premise of going out creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
A kind of TCP video stream traffic user experience quality assessment side that Fig. 1 provides for the embodiment of the present invention 1
The flow chart of method;
A kind of TCP video stream traffic user experience quality assessment side that Fig. 2 provides for the embodiment of the present invention 2
The flow chart of method;
Fig. 3 is that application does not consider the assessment result that the BP neural network model of user's interbehavior is estimated
Schematic diagram;
A kind of TCP video stream traffic user experience quality that Fig. 4 provides for the application embodiment of the present invention 2 is commented
The assessment result schematic diagram that the method for estimating is estimated;
Fig. 5 is that the HTTP Video transmission system experiment carrying out subjective assessment experiment in the embodiment of the present invention 3 is put down
Platform schematic diagram;
The schematic diagram of the BP network neural model that Fig. 6 is used by the embodiment of the present invention 1;
A kind of TCP video stream traffic user experience quality assessment dress that Fig. 7 provides for the embodiment of the present invention 5
The structural representation put;
A kind of TCP video stream traffic user experience quality assessment dress that Fig. 8 provides for the embodiment of the present invention 6
The structural representation put.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clearly
Chu, be fully described by, it is clear that described embodiment be only a part of embodiment of the present invention rather than
Whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not making creation
The every other embodiment obtained under property work premise, broadly falls into the scope of protection of the invention.
Embodiments provide a kind of TCP video stream traffic user experience quality appraisal procedure and device,
Illustrate separately below.
Embodiment 1
As it is shown in figure 1, the embodiment of the present invention 1 provides a kind of TCP video stream traffic user experience quality assessment
Method, comprises the steps:
S101, monitor and record client watch video to be assessed time user's interbehavior, including suspend,
F.F., retrogressing and selected resolution;
When user is by client viewing video, employ the script of written in JavaScript in client, logical
Cross the API of HTML Video Events, make user can control the player on browser, and can monitor also
Record user user's interbehavior during viewing video.
S102, according to described client watch video to be assessed time network layer performance, application layer performance and institute
Stating user's interbehavior, the broadcasting fluency, content draws and the picture that obtain sign video to be assessed are clear
Multiple performance quantizating index of degree;
(1) about the fluency of video playback
Whether smoothness is to affect the topmost factor of Consumer's Experience to video playback, and the present invention chooses to have studied and carries
The Internet gone out and application layer performance index, and characterize suspend this user's interbehavior performance refer to
Mark, carrys out the fluency that quantitation video is play.Wherein, the performance indications of this user's interbehavior of time-out are characterized
Calculation can have multiple, the number of times that such as suspends, the duration etc. of time-out, this is not done by the present embodiment
Concrete restriction.
(2) about the Attraction Degree of video content
Under normal circumstances, user also can pay close attention to very much the interest of video content.On the one hand, the content of video is such as
Fruit is the most interesting, and user watches the focus/climax parts of video the most repeatedly.On the other hand, the content of video
If the most barren, user can be fed up with, and then may select F.F. or directly redirect.Therefore, this
Bright F.F., two user's interbehaviors of retrogressing of having quantified are to characterize the Attraction Degree of video content.Wherein, characterize
F.F., the calculation of the performance indications retreating the two user's interbehavior can have multiple, such as F.F.
Time span, the frequency of F.F., the frequency etc. of retrogressing, this is not specifically limited by the present embodiment.
(3) about the definition of video pictures
It is appreciated that user's desirably video pictures high definition as far as possible.But, at video display process
In, due to limited bandwidth or bad network condition, user is sometimes at the definition of video pictures with regard
Balance is made between the fluency that frequency is play.Such as, in order to watch video glibly, user may
Video is switched to low resolution by high-resolution, a kind of concession that i.e. user makes for smoothness viewing video
Behavior.But, also it is not excluded for sacrificing fluency and adhering to seeing the user of HD video.
Therefore, the present invention has quantified this user's interbehavior of the video resolution selected by user, to characterize
The definition of video pictures, concrete, characterize this user's interbehavior of video resolution selected by user
The calculation of performance indications can have multiple, such as average video resolution, minimum video resolution are held
Continuous duration accounts for the ratio etc. of whole video duration, and this is not specifically limited by the present embodiment.
S103, by each described performance quantizating index input BP neural network model, it is thus achieved that user's subjective experience
MOS value;
Owing to when the input and output parameter of BP neural network model has dependency, could apply
BP neural network model is estimated.Therefore, in order to make BP neural network model be output as Consumer's Experience
Mass M OS value, it is necessary to the input parameter making input BP neural network model is relevant to MOS value.
It is understood that in video display process, the experience state (being properly termed as hidden state) of user
Cannot learn, but in reality, determine that existence, please oneself such as user or be sick of.And user
Interbehavior during viewing video is observable, differentiates including time-out, F.F., Switch Video
Rates etc. operate.Therefore, it can the interbehavior of user is considered as that user watches the experience state of video
Plant external reflection.It is to say, user's interbehavior can reflect that user's body is checked the quality under many circumstances
Amount, user's interbehavior and user experience quality have dependency.
But, for the sake of rigorous, the present invention also utilizes hidden Markov model (Hidden Markov
Model, HMM) to user's interbehavior (Observable state) and Consumer's Experience phase (hidden state) whether
Relevant verifying, the result shows, user's interbehavior and Consumer's Experience have dependency, can be by
Embody the performance quantizating index input parameter as BP neutral net of user's interbehavior, by BP nerve net
The output parameter of network model is as MOS value.
It addition, prior art it has been proved that characterize client watch video to be assessed time network layer performance and
The performance quantizating index of application layer performance is relevant to Consumer's Experience, accordingly it is also possible to quantify to refer to by these performances
It is denoted as the input parameter for BP neutral net.
Concrete, utilizing HMM model, association user's interbehavior is as follows with the process of user experience quality:
Step one, analytic statistics Consumer's Experience status switch and interbehavior;
At moment t, the experience s of the viewing video of usert∈ S={S1,S2,…,SN, wherein S1,S2,…,SNIt is N
Individual possible hidden state, i.e. Consumer's Experience.And thus produce M observed result, i.e. user is at viewing video
During interbehavior vt∈ V={V1,V2,…,VM}.The sequence of the experience state composition that user watches video claims
For Markov chain.
Step 2, employing Baum-Welch method estimate systematic parameter λ of HHM model;
HHM model can be described, i.e. HMM by five elements (2 state sets and 3 probability matrixs)
Systematic parameter λ={ N, M, A, B, π }, be abbreviated as λ={ A, B, π }.N represents the hidden status number of HMM
Mesh, S={S1,S2,…,SNIt it is hidden state set;M represents consequent Observable result number,
V={V1,V2,…,VMIt it is observation set.A represents hidden state transition probability matrix, describe HMM each
Transition probability between hidden state;B represents observation probability matrix, describes the general of observation under hidden state
Rate is distributed;π represents initial state probabilities matrix, describes the initial probability distribution of hidden state.
A={aij| 1≤i, j≤N}, wherein, aij=P (Sk=j | Sk-1=i), ij ∈ S.
B={bj(k) | 1≤j≤N, 1≤k≤M}, wherein, bj(k)=P (Vk|Sj)。
π={ πi| 1≤i≤N}, wherein, πi=P (S1=i).
Experience state and measuring behavior state during user watches video are it has been determined that for a length of T
Observation sequence O={O1,O2…,OT, selection can be by this maximized parameter of observation sequence probability of occurrenceMake
For the best estimate of HMM, i.e. need to find parameterMake the likelihood function Pr (v | λ) representing probability maximum
Change.Use Baum-Welch method, initial estimation amount calculate log-likelihood function and select new estimator.
Specifically, if the initial estimation amount of λ is λi-1, use following this estimator of log-likelihood function correction:
Wherein, U represents λi-1To λ makeover process seemingly
So function, Q represents specific hidden status switch.
Take new estimator λi=argmaxλQ(λ,λi-1), to specific hidden status switch Q, can obtain,
Wherein, the hidden state during q refers to specific hidden status switch Q
Value,Refer to hidden state qt-1To hidden state qtTransition probability value,It it is hidden state qtCorresponding observation
Probit, πq0It it is hidden state q0Corresponding initial state probabilities value.
Therefore, likelihood function becomes:
Independent estimations model parameter can be gone out, by calculating 0 moment shape by optimizing three, the right of above equation
State SjThe ratio occurred can determine that πj, by calculating in state SiTime SiIt is converted to SjRatio determine aij;
Calculating state Sj, and state SjUnder the conditions of observe VkRatio can get bj(k).Final parameter value λ can
Obtained by iterative computation.Concrete iterative calculation method belongs to prior art, and here is omitted.
The parameter lambda obtained estimated by step 3, basis, it was predicted that Consumer's Experience state during time t;
After HMM model determines, it was predicted that Consumer's Experience state equivalent during time t is in given observation sequence and mould
Determine most probable hidden state under the conditions of shape parameter, for following the tracks of User Status continuously, wall scroll need to be found and most preferably use
Family status switch, realizes by maximizing posterior probability Pr (Q | O, λ) or Pr (Q, O | λ), by MAP rule,
Estimate user viewed status q during viewing videot。
S104, according to described MOS value, determine the user experience quality QoE of video to be assessed.
MOS value is the quantitative description to QoE, and the value of MOS value is generally the integer of 1 to 5, from 1 to 5
Correspondence is the most dissatisfied, dissatisfied, general, satisfied respectively and feels quite pleased five kinds of QoE.
The TCP video stream traffic user experience quality appraisal procedure provided due to the embodiment of the present invention 1, not only
Consider Internet and the impact on TCP video stream traffic QoE of the application layer performance index, it is also contemplated that user
The impact of interbehavior so that the actual experience of assessment result and user that this appraisal procedure obtains closer to,
Improve the accuracy of TCP video stream traffic QoE assessment, Evaluated effect is good.
Embodiment 2
As in figure 2 it is shown, the another kind of TCP video stream traffic user experience quality that the embodiment of the present invention provides is commented
Estimating method, as a kind of preferred version of embodiment 1, embodiment 2 is with the difference of embodiment 1,
After execution of step S101, and before performing step S102, described method also includes:
S105, screen described user's interbehavior.
It is understood that user may make various interbehavior during viewing video,
But some of which behavior is user infrequently to be done, and the probability of generation is little, and weighing factor is low, and researching value is not
Greatly;Some user behavior is typically random appearance, with uncertainty.Therefore, to user's interbehavior
Suitably screen, the accuracy of the QoE assessment of TCP video traffic can be improved further.
Same, if to the user in the sample data by subjective assessment experiment acquisition above addressed
After interbehavior screens, the sample data training BP neural network model after recycling screening, training obtains
The BP Model of Neural Network obtained is more excellent.
Specifically, can be screened by user's interbehavior that monitoring is obtained by following manner:
Screen and delete the random user interbehavior in described user's interbehavior;
And/or, screen and delete the small probability user's interbehavior in described user's interbehavior.
Further, the determination method of random user interbehavior includes:
Step one, determine the effective time scope of user's interbehavior according to the first function;
Wherein, described first function is:δ is described
Effective time scope, tstartAnd tendRepresent the moment commencing play out and playing end of video to be assessed, t respectivelyi
In the moment occurred for user's interbehavior, T is default time range;
Step 2, determine the output valve of the second function according to described effective time scope, described second letter will be made
The output valve null user interbehavior of number is defined as random user interbehavior;
Wherein, described second function is:DiIt is ti-δ arrives tiIn+δ time range, niIndividual
The moment t that user's interbehavior occursa+jWith tiRelative distance difference d(a+j)iSum.
Further, the determination method of small probability user interbehavior includes:
When described user's interbehavior is the small probability user's interbehavior adding up acquisition in advance, it is defined as little
Probability user's interbehavior.
The TCP video stream traffic user experience quality appraisal procedure that the application embodiment of the present invention 2 provides, compares
In prior art, not only considering user's interbehavior basis on the impact of TCP video stream traffic QoE
On so that the actual experience of assessment result and user that this appraisal procedure obtains closer to, improve TCP and regard
The accuracy of frequency stream business QoE assessment.And, after user's interbehavior is screened, can be further
Improving the accuracy of TCP video stream traffic QoE assessment, Evaluated effect is more preferably.
In order to prove the TCP video stream traffic user experience quality assessment that the embodiment of the present invention 2 provides further
The helpfulness of method, the QoE appraisal procedure that Application Example 2 is provided by the present invention does not consider user with application
The BP neural network model of interbehavior carries out the assessment performance of QoE assessment and compares, comparative result such as figure
Shown in 3 and Fig. 4, wherein Fig. 3 is that application does not consider that the BP neural network model of user's interbehavior is estimated
Assessment result schematic diagram, Fig. 4 be application the embodiment of the present invention 2 provide a kind of TCP video stream traffic user
The assessment result schematic diagram that Quality of experience appraisal procedure is estimated.
In figs. 3 and 4, discrete point represents the assessment result of single sample data;Solid line is canonical reference
Line, represents and utilizes BP network neural model evaluation to obtain the MOS value that MOS value obtains with subjective assessment experiment
Essentially equal;Dotted line is the result that discrete point carries out linear fit, represents and utilizes BP network neural model to comment
Estimating the mutual relation obtaining the MOS value that MOS value obtains with subjective assessment experiment, R represents the phase of linear fit
Guan Xing.R is the biggest, meanwhile, Linear Quasi zygonema and canonical reference line closer to, show to utilize BP network neural
It is the most close with the MOS value that subjective assessment experiment obtains that model evaluation obtains MOS value.
Comparison diagram 3 and Fig. 4 understands, the TCP video stream traffic user experience quality that the embodiment of the present invention 2 provides
Appraisal procedure, owing to considering user's interbehavior, therefore, assessment result more connects with the real experiences of user
Closely, assessment result is more accurate.
Embodiment 3
As embodiment 1 or a kind of preferred version of embodiment 2, embodiment 3 and embodiment 1 or embodiment 2
Difference is, the BP neural network model employed in the present embodiment is by utilizing sample data to train
Obtaining, the detailed process of training BP neural network model is as follows:
Step one, carries out user and watches the user experience quality subjective assessment experiment of video, to obtain sample number
According to.
In experiment, experimenter's M-F is coordinated, from different background environments.Build as shown in Figure 5
HTTP Video transmission system, in Figure 51 representative server, 2 represent router, and 3 represent user's viewing regards
The client of frequency, controls, by TC order, the network environment that netem simulation is different on router 2, and network is joined
Arranging as shown in table 1 of number.
The network parameter arranged in the experiment of table 1 subjective assessment
In the subjective assessment of video is tested, the system monitoring record user interbehavior when watching video,
Obtain user's interbehavior sequence.Experimenter, can be real-time according to the needs of oneself in video display process
Provide scoring, produce Consumer's Experience status switch.At the end of video playback, experimenter provides one again
Last TOP SCORES.After experiment terminates, appraisal result is done statistical analysis, reject invalid data, enter
And draw effective user's subjective experience MOS value.
The user's interbehavior sequence obtained during subjective assessment is tested and final user subjective experience MOS
Value is defined as sample data.Wherein it is possible to utilize the same way disclosed in embodiment 2, mutual to user
Small probability user's interbehavior and random user interbehavior in behavior sequence screen, after screening
User's subjective experience MOS value of user's interbehavior sequence and correspondence is as sample data, so that training obtains
BP neural network model more excellent.
Step 2, the sample data that subjectively-based evaluation experiment obtains, train BP neutral net.
The present embodiment uses three layers of BP neutral net, and its structure is made up of input layer, hidden layer, output layer.
Input layer comprises 8 input nodes, the sample data that subjective assessment experiment obtains is converted into 8 individual characteies energetic
Index is as the input parameter of BP neural network model, and this energetic index of 8 individual character is: initial buffering time,
Averagely buffer duration, heavily buffer frequency, averagely suspend duration, suspend frequency, redirect ratio, backward forward
Redirect ratio, average video resolution.Output layer comprises an output node, i.e. assessment MOS value.For
Choosing of node in hidden layer, is trained by the neutral net constructing different Hidden nodes, and according to
The error convergence speed of each neutral net and the comparison of the mean square error size of sign fitting degree, optional
The network structure of N number of hidden node, tests prove that, N can take 13.
In order to reach convergence rate and higher precision faster, the present invention select based on
Levenberg-Marquardt (LM) learning algorithm training BP neutral net.
The learning process of BP neural network algorithm is divided into forward input and two processes of back propagation, forward-propagating
Middle input information processes through hidden layer from input layer, and is transmitted to output layer.During the forward-propagating of information,
By input information i.e. pth learning sample Xp={ xp1, xp2..., xpMInput BP neutral net after, from
Input layer processes through hidden layer, and is transmitted to output layer, calculates the desired output y of hidden layer jth nodepjWith
The actual output z of output layer nodep1:
Input layer to the transmission function of hidden layer node is: ypj=f1(x)=tanh (x), ypjFor the output valve of jth hidden layer node, wijFor i-th input layer and jth
Network weight between individual hidden layer node, xpiFor the input value of i-th input layer, θjImply for jth
The threshold vector of node layer;
Hidden layer node to the transmission function of output layer node is: zp1=f2(x')=ax'+b,
zp1For the MOS value of output layer node output, vj1For the internodal network of jth hidden layer node and output layer
Weights, θ1For the threshold vector of output layer node, f2X ()=ax+b, a and b is constant.
If the expected value of output node is tp1, it is desirable to value tp1For the user subjective experience MOS in sample data
Value, then the error criterion function of output node is:
If desired output can not be obtained at output layer, then proceed to back propagation by error signal e (w) along former
The connection path come returns, and is optimized the weights between each node layer, until can be expected at output layer
Output valve.The method of adjustment of the weights between each node layer of Levenberg-Marquardt optimized algorithm is:
Δ w=(JTJ=+ μ I)-1JTE, wherein, J is the error Jacobian matrix to weights differential, and e is for by mistake
Difference vector, μ is learning rate.If wkRepresent internodal weights and the threshold value institute group of node of kth time iteration
The vector become, the vector that new internodal weights and the threshold value of node are formed is: wk+1=wk+Δw。
When the internodal weights in BP neutral net, the threshold vector of node and internodal transmission function are equal
After determining, BP neural network model has just been trained.
Fig. 6 shows the schematic diagram of the BP network model employed in embodiment 3,
The TCP video stream traffic user experience quality appraisal procedure that the embodiment of the present invention 3 provides, compared to existing
There is technology, not only on the basis of considering the impact on TCP video stream traffic QoE of user's interbehavior,
Make the actual experience of assessment result that this appraisal procedure obtains and user closer to, improve TCP video flowing
The accuracy of business QoE assessment.It is additionally, since and employs structure preferably BP neural network model, therefore,
Make the actual experience of assessment result and user closer to, assessment result is more accurate, and Evaluated effect is more carefully.
Embodiment 4
As a kind of preferred version of any embodiment in embodiment 1 to 3, in embodiment 4 and embodiment 1 to 3
The difference of any embodiment is, wherein characterizes video playback fluency, content draws and picture clear
The performance quantizating index of clear degree is specific as follows:
(1) the performance quantizating index of the fluency characterizing video playback includes:
Three the application layer performance indexs researched and proposed: initial buffering time, averagely buffer duration,
Heavily buffer frequency, and two indexs relevant to user behavior: averagely suspend duration, suspend frequency, come
The fluency that quantitation video is play.
It should be noted that no matter be packet loss or time delay, network layer performance is the most all reacted in application layer,
Therefore, above three application layer performance index can be with concentrated expression Internet and the performance of application layer.
Initial buffering time Tinit, tolerance starts to be loaded into the time interval commenced play out first from video.
Average buffering duration Trebuf, in video display process, measure again the meansigma methods at buffer time interval.
Heavily buffering frequency frebuf, in video display process, the frequency that tolerance video weight buffered event occurs.
Average time-out duration Tpause, in video display process, the meansigma methods of tolerance time-out period
Suspend frequency fpause, in video display process, the frequency that tolerance suspending event occurs.
(2) the performance quantizating index of the Attraction Degree characterizing video content includes: F.F. ratio and the ratio of retrogressing.
Wherein, F.F. ratio:TJF,kRefer to the time span that kth time redirects forward, nJF
It is the number of times redirected forward, lJFThe total time length redirected before sensing, lvideoTotal duration for evaluated video.
Wherein, ratio is retreated:TJBRefer to the time span redirected backward the m time,
nJBIt is the number of times redirected backward, lJBThe total time length redirected after sensing.
(3) the performance quantizating index of the definition characterizing video pictures includes: average video resolution Dswitch。
Wherein, R is the video resolution selected for user, drFor regarding that user selects
Frequency division resolution, lrFor user in resolution drThe duration of lower viewing video to be assessed.Wherein, the selectable value of R
Including: 720P, 480P and 360P.
The TCP video stream traffic user experience quality appraisal procedure that the embodiment of the present invention 4 provides, compared to existing
There is technology, not only on the basis of considering the impact on TCP video stream traffic QoE of user's interbehavior,
Make the actual experience of assessment result that this appraisal procedure obtains and user closer to, improve TCP video flowing
The accuracy of business QoE assessment.It is additionally, since and employs 8 performances that can reflect user experience quality
Quantizating index is as the input parameter of BP neural network model, therefore so that assessment result is true with user's
Experience closer to, further increase TCP video stream traffic QoE assessment accuracy, Evaluated effect is more preferable.
Embodiment 5
Corresponding to said method embodiment, as it is shown in fig. 7, present invention also offers a kind of TCP video flowing industry
Business user experience quality apparatus for evaluating, described device includes: user's interbehavior monitoring modular 701, performance
Quantizating index acquisition module 702, MOS value computing module 703 and evaluation module 704,
User's interbehavior monitoring modular 701, for monitoring and record when client watching video to be assessed
User's interbehavior, including: time-out, F.F., retrogressing and selected resolution;
When user is by client viewing video, use JavaScript by HTML5 video in client
The API of event controls the player on browser, monitors and records user user during seeing video and hand over
Behavior mutually.
Performance quantizating index acquisition module 702, for watching net during video to be assessed according to described client
Network layers performance, application layer performance and described user's interbehavior, obtain the broadcasting smoothness characterizing video to be assessed
Degree, content draws and multiple performance quantizating index of image sharpness;
Wherein, characterize video to be assessed plays fluency, content draws and many individual characteies of image sharpness
Energetic index is consistent with embodiment 1, does not the most do repeated description.
MOS value computing module 703, for inputting BP neutral net mould by each described performance quantizating index
Type, it is thus achieved that user's subjective experience MOS value;
Also due to when the input and output parameter of BP neural network model has dependency, Cai Nengying
It is estimated with BP neural network model.Therefore, in order to make BP neural network model be output as user's body
The amount of checking the quality MOS value, it is necessary to the input parameter making input BP neural network model is relevant to MOS value.
It is understood that in video display process, the experience state (being properly termed as hidden state) of user
Cannot learn, but in reality, determine that existence, please oneself such as user or be sick of.And user exists
Interbehavior during viewing video is observable, including time-out, F.F., Switch Video resolution etc.
Operation.The one that therefore, it can be considered as the interbehavior of user the experience state that user watches video is external
Reflection.It is to say, user's interbehavior can reflect user experience quality, user under many circumstances
Interbehavior and user experience quality have dependency.
But for the sake of rigorous, as in Example 1, the present embodiment is also adopted by HHM model to
Family interbehavior (Observable state) is the most relevant to Consumer's Experience phase (hidden state) to be verified, checking
Result shows, user's interbehavior and Consumer's Experience have dependency, can will embody user's interbehavior
The output parameter of BP neural network model, as the input parameter of BP neutral net, is made by performance quantizating index
For MOS value.
Concrete, use HHM modelling verification user's interbehavior and the process of user experience quality and enforcement
In example 1 unanimously, repeated description is not the most done.
It addition, prior art it has been proved that characterize client watch video to be assessed time network layer performance and
The performance quantizating index of application layer performance is relevant to Consumer's Experience, accordingly it is also possible to quantify to refer to by these performances
It is denoted as the input parameter for BP neutral net.
Evaluation module 704, for according to described MOS value, determines the user experience quality of video to be assessed
QoE。
MOS value is the quantitative description to QoE, and the value of MOS value is generally the integer of 1 to 5, from 1 to 5
Correspondence is the most dissatisfied, dissatisfied, general, satisfied respectively and feels quite pleased five kinds of QoE.
The TCP video stream traffic user experience quality apparatus for evaluating provided due to the embodiment of the present invention 5, not only
Consider Internet and the impact on TCP video stream traffic QoE of the application layer performance index, it is also contemplated that user
The impact of interbehavior so that the actual experience of assessment result and user that this apparatus for evaluating obtains closer to,
Improve the accuracy of TCP video stream traffic QoE assessment, Evaluated effect is good.
Embodiment 6
As shown in Figure 8, on the basis of embodiment 5, the embodiment of the present invention 6 additionally provides another kind of TCP and regards
Frequency stream service-user Quality of experience apparatus for evaluating, is with the difference of embodiment 5, described device also wraps
Include: user's interbehavior screening module 705, for what described user's interbehavior monitoring module monitors was arrived
User's interbehavior screens, and is transferred to described performance quantizating index acquisition module.
It is understood that user may make various interbehavior during viewing video,
But some of which behavior is user infrequently to be done, and the probability of generation is little, and weighing factor is low, and researching value is not
Greatly;Some user behavior is typically random appearance, with uncertainty.Therefore, to user's interbehavior
Suitably screen, the accuracy of the QoE assessment of TCP video traffic can be improved further.
Concrete, user's interbehavior screening module 705, including:
Random user interbehavior screening unit, hands over for deleting the random user in described user's interbehavior
Behavior mutually;
Small probability user's interbehavior screening unit, uses for deleting the small probability in described user's interbehavior
Family interbehavior.
Wherein, random user interbehavior screening unit, including:
Effective time range determination submodule, during for determining user's interbehavior effective according to the first function
Between scope;Wherein, described first function is:δ is
Described effective time scope, tstartAnd tendRepresent respectively video to be assessed commence play out and play end time
Carve, tiIn the moment occurred for user's interbehavior, T is default time range;
Random user interbehavior determines submodule, for determining the second function according to described effective time scope
Output valve, the output valve null user interbehavior making described second function is defined as random user
Interbehavior, and delete this random user interbehavior;Wherein, described second function is:
DiIt is ti-δ arrives tiIn+δ time range, niThe moment t that individual user's interbehavior occursa+jWith tiRelative distance
Difference d(a+j)iSum.
It addition, the determination method of small probability user's interbehavior includes:
When described user's interbehavior is the small probability user's interbehavior adding up acquisition in advance, it is defined as little
Probability user's interbehavior.
The TCP video stream traffic user experience quality apparatus for evaluating that the application embodiment of the present invention 6 provides, compares
In prior art, not only considering user's interbehavior basis on the impact of TCP video stream traffic QoE
On so that the actual experience of assessment result and user that this apparatus for evaluating obtains closer to, improve TCP and regard
The accuracy of frequency stream business QoE assessment.And, after user's interbehavior is screened, can be further
Improving the accuracy of TCP video stream traffic QoE assessment, Evaluated effect is more preferably.
Embodiment 7
As embodiment 5 or a kind of preferred version of embodiment 6, embodiment 7 and embodiment 5 or embodiment 6
Difference is, the BP neural network model employed in the present embodiment is by utilizing sample data to train
Obtain, consistent described in concrete training process and embodiment 3, the most do not do repeated description.
But it should be recognized that preferably, the BP neural network model that training obtains by input layer, hidden layer,
Output layer forms.Input layer comprises 8 input nodes, the sample data that subjective assessment experiment obtains is converted
Being the 8 energetic indexs of the individual character input parameters as BP neural network model, this energetic index of 8 individual character is:
Initial buffering time, averagely buffer duration, heavily buffer frequency, averagely suspend duration, suspend frequency, forward
Redirect ratio, redirect ratio, average video resolution backward.Output layer comprises an output node, i.e. comments
Estimate MOS value.Node in hidden layer is chosen, is entered by the neutral net constructing different Hidden nodes
Row training, and according to the error convergence speed of each neutral net and the mean square error size that characterizes fitting degree
Comparison, the network structure of optional N number of hidden node, tests prove that, the assessment knot that N obtains when taking 13
Fruit is optimum.
The TCP video stream traffic user experience quality apparatus for evaluating that the embodiment of the present invention 7 provides, compared to existing
There is technology, not only on the basis of considering the impact on TCP video stream traffic QoE of user's interbehavior,
Make the actual experience of assessment result that this apparatus for evaluating obtains and user closer to, improve TCP video flowing
The accuracy of business QoE assessment.It is additionally, since and employs structure preferably BP neural network model, therefore,
Make the actual experience of assessment result and user closer to, assessment result is more accurate, and Evaluated effect is more preferable.
Embodiment 8
As a kind of preferred version of any embodiment in embodiment 5 to 7, in embodiment 8 and embodiment 5 to 7
The difference of any embodiment is, wherein characterizes video playback fluency, content draws and picture clear
The performance quantizating index of clear degree is specific as follows:
(1) the performance quantizating index of the fluency characterizing video playback includes:
Three the application layer performance indexs researched and proposed: initial buffering time, averagely buffer duration,
Heavily buffer frequency, and two indexs relevant to user behavior: averagely suspend duration, suspend frequency, come
The fluency that quantitation video is play.
It should be noted that no matter be packet loss or time delay, network layer performance is the most all reacted in application layer,
Therefore, above three application layer performance index can be with concentrated expression Internet and the performance of application layer.
Initial buffering time Tinit, tolerance starts to be loaded into the time interval commenced play out first from video.
Average buffering duration Trebuf, in video display process, measure again the meansigma methods at buffer time interval.
Heavily buffering frequency frebuf, in video display process, the frequency that tolerance video weight buffered event occurs.
Average time-out duration Tpause, in video display process, the meansigma methods of tolerance time-out period
Suspend frequency fpause, in video display process, the frequency that tolerance suspending event occurs.
(2) the performance quantizating index of the Attraction Degree characterizing video content includes: F.F. ratio and the ratio of retrogressing.
Wherein, F.F. ratio:TJF,kRefer to the time span that kth time redirects forward, nJF
It is the number of times redirected forward, lJFThe total time length redirected before sensing, lvideoTotal duration for evaluated video.
Wherein, ratio is retreated:TJBRefer to the time span redirected backward the m time,
nJBIt is the number of times redirected backward, lJBThe total time length redirected after sensing.
(3) the performance quantizating index of the definition characterizing video pictures includes: average video resolution Dswitch。
Wherein,R is the video resolution selected for user, drFor regarding that user selects
Frequency division resolution, lrFor user in resolution drThe duration of lower viewing video to be assessed.Wherein, the selectable value of R
Including: 720P, 480P and 360P.
The TCP video stream traffic user experience quality apparatus for evaluating that the application embodiment of the present invention 8 provides, compares
In prior art, not only considering user's interbehavior basis on the impact of TCP video stream traffic QoE
On so that the actual experience of assessment result and user that this apparatus for evaluating obtains closer to, improve TCP and regard
The accuracy of frequency stream business QoE assessment.It is additionally, since and employs 8 and can reflect user experience quality
Performance quantizating index is as the input parameter of BP neural network model, therefore so that assessment result is with user's
Actual experience closer to, further increase TCP video stream traffic QoE assessment accuracy, Evaluated effect
More preferably.
For device embodiment, owing to it is substantially similar to embodiment of the method, so the comparison described is simple
Single, relevant part sees the part of embodiment of the method and illustrates.
It should be noted that in this article, the relational terms of such as first and second or the like be used merely to by
One entity or operation separate with another entity or operating space, and not necessarily require or imply these
Relation or the order of any this reality is there is between entity or operation.And, term " includes ", " bag
Contain " or its any other variant be intended to comprising of nonexcludability, so that include a series of key element
Process, method, article or equipment not only include those key elements, but also include being not expressly set out
Other key elements, or also include the key element intrinsic for this process, method, article or equipment.?
In the case of there is no more restriction, statement " including ... " key element limited, it is not excluded that including
The process of described key element, method, article or equipment there is also other identical element.
Each embodiment in this specification all uses relevant mode to describe, phase homophase between each embodiment
As part see mutually, what each embodiment stressed is different from other embodiments it
Place.For device embodiment, owing to it is substantially similar to embodiment of the method, so describe
Fairly simple, relevant part sees the part of embodiment of the method and illustrates.
The foregoing is only presently preferred embodiments of the present invention, be not intended to limit the protection model of the present invention
Enclose.All any modification, equivalent substitution and improvement etc. made within the spirit and principles in the present invention, all wrap
Containing within the scope of the present invention.
Claims (10)
1. a TCP video stream traffic user experience quality appraisal procedure, it is characterised in that described method
Including:
S101, monitor and record client watch video to be assessed time user's interbehavior, including suspend,
F.F., retrogressing and selected resolution;
S102, according to described client watch video to be assessed time network layer performance, application layer performance and institute
Stating user's interbehavior, the broadcasting fluency, content draws and the picture that obtain sign video to be assessed are clear
Multiple performance quantizating index of degree;
S103, by each described performance quantizating index input BP neural network model, it is thus achieved that user's subjective experience
MOS value;
S104, according to described MOS value, determine the user experience quality QoE of video to be assessed.
Method the most according to claim 1, it is characterised in that after performing step S101, and holding
Before row step S102, described method also includes:
Screen described user's interbehavior.
Method the most according to claim 2, it is characterised in that described screening described user interbehavior
Including:
Screen and delete the random user interbehavior in described user's interbehavior;
And/or, screen and delete the small probability user's interbehavior in described user's interbehavior.
Method the most according to claim 3, it is characterised in that described random user interbehavior is really
The method of determining includes:
The effective time scope of user's interbehavior is determined according to the first function;Wherein, described first function is:δ is described effective time scope, tstartAnd tendRespectively
Represent the moment commencing play out and playing end of video to be assessed, tiThe moment occurred for user's interbehavior,
T is default time range;
Determine the output valve of the second function according to described effective time scope, the output of described second function will be made
It is worth null user's interbehavior and is defined as random user interbehavior;Wherein, described second function is:DiIt is ti-δ arrives tiIn+δ time range, niThe moment t that individual user's interbehavior occursa+jWith
tiRelative distance difference d(a+j)iSum;
The determination method of described small probability user's interbehavior includes:
When described user's interbehavior is the small probability user's interbehavior adding up acquisition in advance, it is defined as little
Probability user's interbehavior.
5. according to the method according to any one of claim 1-4, it is characterised in that described BP neutral net
Model is made up of input layer, hidden layer and output layer, and the number of input layer is 8, hidden layer node
Number is 13, and the number of output layer node is 1.
6. according to the method according to any one of claim 1-4, it is characterised in that described sign is to be assessed
The performance quantizating index playing fluency of video includes: initial buffering time, averagely buffer duration, the most slow
Rush frequency, average time-out duration and suspend frequency;
The performance quantizating index of the content draws of described sign video to be assessed includes: F.F. ratio, retrogressing
Ratio;
The performance quantizating index of the image sharpness of described sign video to be assessed includes: video average resolution
Rate.
7. a TCP video stream traffic user experience quality apparatus for evaluating, it is characterised in that described device
Including: user's interbehavior monitoring modular, performance quantizating index acquisition module, MOS value computing module and comment
Estimate module,
Described user's interbehavior monitoring modular, for monitoring and record when client watching video to be assessed
User's interbehavior, including: time-out, F.F., retrogressing and selected resolution;
Described performance quantizating index acquisition module, for watching net during video to be assessed according to described client
Network layers performance, application layer performance and described user's interbehavior, obtain the broadcasting smoothness characterizing video to be assessed
Degree, content draws and multiple performance quantizating index of image sharpness;
Described MOS value computing module, for inputting BP neutral net mould by each described performance quantizating index
Type, it is thus achieved that user's subjective experience MOS value;
Described evaluation module, for according to described MOS value, determines the user experience quality of video to be assessed
QoE。
Device the most according to claim 7, it is characterised in that described device also includes: user is mutual
Behavior screening module, for described user's interbehavior monitoring module monitors to user's interbehavior carry out
Screening, and it is transferred to described performance quantizating index acquisition module.
Device the most according to claim 8, it is characterised in that described user's interbehavior screening module,
Including:
Random user interbehavior screening unit, random for screen and delete in described user's interbehavior
User's interbehavior;
Small probability user's interbehavior screening unit is little for screen and delete in described user's interbehavior
Probability user's interbehavior.
Device the most according to claim 9, it is characterised in that described random user interbehavior sieves
Menu unit, including:
Effective time range determination submodule, during for determining user's interbehavior effective according to the first function
Between scope;Wherein, described first function is:δ is
Described effective time scope, tstartAnd tendRepresent respectively video to be assessed commence play out and play end time
Carve, tiIn the moment occurred for user's interbehavior, T is default time range;
Random user interbehavior determines submodule, for determining the second function according to described effective time scope
Output valve, the output valve null user interbehavior making described second function is defined as random user
Interbehavior, and delete this random user interbehavior;Wherein, described second function is:
DiIt is ti-δ arrives tiIn+δ time range, niThe moment t that individual user's interbehavior occursa+jWith tiRelative distance
Difference d(a+j)iSum.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610326745.6A CN105897736A (en) | 2016-05-17 | 2016-05-17 | Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610326745.6A CN105897736A (en) | 2016-05-17 | 2016-05-17 | Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105897736A true CN105897736A (en) | 2016-08-24 |
Family
ID=56717439
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610326745.6A Pending CN105897736A (en) | 2016-05-17 | 2016-05-17 | Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105897736A (en) |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106454512A (en) * | 2016-09-06 | 2017-02-22 | 北京小度互娱科技有限公司 | Player event statistical method and device player and terminal equipment |
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 |
CN107087160A (en) * | 2017-04-28 | 2017-08-22 | 南京邮电大学 | A kind of Forecasting Methodology of the user experience quality based on BP Adaboost neutral nets |
CN107820125A (en) * | 2017-11-28 | 2018-03-20 | 北京小度互娱科技有限公司 | Method and device based on user behavior optimization Video Applications experience |
CN108668123A (en) * | 2017-03-31 | 2018-10-16 | 华为技术有限公司 | A kind of acquisition methods and network element device of video tastes evaluation result |
CN108833976A (en) * | 2018-06-27 | 2018-11-16 | 深圳看到科技有限公司 | A kind of panoramic video dynamic cuts the picture quality evaluations method and device after stream |
CN109451303A (en) * | 2018-12-24 | 2019-03-08 | 合肥工业大学 | A kind of modeling method for user experience quality QoE in VR video |
CN109598704A (en) * | 2018-11-19 | 2019-04-09 | 电子科技大学 | A kind of night soil microscopic image clarity evaluation method based on BP neural network |
CN109710552A (en) * | 2018-12-29 | 2019-05-03 | 深圳市联影医疗数据服务有限公司 | Bus transfer method for evaluating quality, system and computer storage medium |
CN109729433A (en) * | 2019-01-22 | 2019-05-07 | 北京奇艺世纪科技有限公司 | A kind of video playing appraisal procedure and device |
CN110855669A (en) * | 2019-11-14 | 2020-02-28 | 北京理工大学 | Video QoE index prediction method suitable for encrypted flow based on neural network |
WO2020077682A1 (en) * | 2018-10-17 | 2020-04-23 | 网宿科技股份有限公司 | Service quality evaluation model training method and device |
US10694221B2 (en) | 2018-03-06 | 2020-06-23 | At&T Intellectual Property I, L.P. | Method for intelligent buffering for over the top (OTT) video delivery |
CN112862250A (en) * | 2021-01-12 | 2021-05-28 | 浙江知行教育科技有限公司 | College learning evaluation system and method based on big data |
CN113542880A (en) * | 2020-04-20 | 2021-10-22 | 中国移动通信集团河北有限公司 | Method and device for detecting quality of short video service |
US11429891B2 (en) | 2018-03-07 | 2022-08-30 | At&T Intellectual Property I, L.P. | Method to identify video applications from encrypted over-the-top (OTT) data |
CN115225936A (en) * | 2021-04-19 | 2022-10-21 | 中国移动通信集团河北有限公司 | Definition index determination method, device, equipment and medium for video resources |
CN115278354A (en) * | 2022-06-14 | 2022-11-01 | 北京大学 | Method and system for evaluating and optimizing video transmission quality based on user behavior index |
WO2022267720A1 (en) * | 2021-06-23 | 2022-12-29 | 华为技术有限公司 | Data transmission method and communication apparatus |
CN116017092A (en) * | 2022-12-12 | 2023-04-25 | 中国科学院计算技术研究所 | QoE quantization method and transmission system parameter adjustment method based on QoE quantization method |
CN116074586A (en) * | 2022-12-13 | 2023-05-05 | 北京中电飞华通信有限公司 | Video experience quality prediction method and device, electronic equipment and storage medium |
-
2016
- 2016-05-17 CN CN201610326745.6A patent/CN105897736A/en active Pending
Non-Patent Citations (2)
Title |
---|
于静伟: "IPTV多媒体业务平台设计与质量保障研究", 《中国优秀硕士学位论文全文数据库》 * |
刘帆: "面向视频流业务的用户体验质量评估方法", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106454512A (en) * | 2016-09-06 | 2017-02-22 | 北京小度互娱科技有限公司 | Player event statistical method and device player and terminal equipment |
CN108668123A (en) * | 2017-03-31 | 2018-10-16 | 华为技术有限公司 | A kind of acquisition methods and network element device of video tastes evaluation result |
CN107071399B (en) * | 2017-04-26 | 2019-02-19 | 华为技术有限公司 | A kind of method for evaluating quality and device of encrypted video stream |
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 |
CN107087160A (en) * | 2017-04-28 | 2017-08-22 | 南京邮电大学 | A kind of Forecasting Methodology of the user experience quality based on BP Adaboost neutral nets |
CN107820125A (en) * | 2017-11-28 | 2018-03-20 | 北京小度互娱科技有限公司 | Method and device based on user behavior optimization Video Applications experience |
CN107820125B (en) * | 2017-11-28 | 2020-03-17 | 北京小度互娱科技有限公司 | Method and device for optimizing video application experience based on user behavior |
US11166053B2 (en) | 2018-03-06 | 2021-11-02 | At&T Intellectual Property I, L.P. | Method for intelligent buffering for over the top (OTT) video delivery |
US11606584B2 (en) | 2018-03-06 | 2023-03-14 | At&T Intellectual Property I, L.P. | Method for intelligent buffering for over the top (OTT) video delivery |
US10694221B2 (en) | 2018-03-06 | 2020-06-23 | At&T Intellectual Property I, L.P. | Method for intelligent buffering for over the top (OTT) video delivery |
US11699103B2 (en) | 2018-03-07 | 2023-07-11 | At&T Intellectual Property I, L.P. | Method to identify video applications from encrypted over-the-top (OTT) data |
US11429891B2 (en) | 2018-03-07 | 2022-08-30 | At&T Intellectual Property I, L.P. | Method to identify video applications from encrypted over-the-top (OTT) data |
CN108833976A (en) * | 2018-06-27 | 2018-11-16 | 深圳看到科技有限公司 | A kind of panoramic video dynamic cuts the picture quality evaluations method and device after stream |
WO2020077682A1 (en) * | 2018-10-17 | 2020-04-23 | 网宿科技股份有限公司 | Service quality evaluation model training method and device |
CN109598704A (en) * | 2018-11-19 | 2019-04-09 | 电子科技大学 | A kind of night soil microscopic image clarity evaluation method based on BP neural network |
CN109451303A (en) * | 2018-12-24 | 2019-03-08 | 合肥工业大学 | A kind of modeling method for user experience quality QoE in VR video |
CN109710552A (en) * | 2018-12-29 | 2019-05-03 | 深圳市联影医疗数据服务有限公司 | Bus transfer method for evaluating quality, system and computer storage medium |
CN109729433B (en) * | 2019-01-22 | 2021-05-07 | 北京奇艺世纪科技有限公司 | Video playing evaluation method and device |
CN109729433A (en) * | 2019-01-22 | 2019-05-07 | 北京奇艺世纪科技有限公司 | A kind of video playing appraisal procedure and device |
CN110855669B (en) * | 2019-11-14 | 2020-10-09 | 北京理工大学 | Video QoE index prediction method suitable for encrypted flow based on neural network |
CN110855669A (en) * | 2019-11-14 | 2020-02-28 | 北京理工大学 | Video QoE index prediction method suitable for encrypted flow based on neural network |
CN113542880A (en) * | 2020-04-20 | 2021-10-22 | 中国移动通信集团河北有限公司 | Method and device for detecting quality of short video service |
CN112862250A (en) * | 2021-01-12 | 2021-05-28 | 浙江知行教育科技有限公司 | College learning evaluation system and method based on big data |
CN112862250B (en) * | 2021-01-12 | 2023-12-26 | 北京漂洋过海科技有限责任公司 | College learning evaluation system and method based on big data |
CN115225936A (en) * | 2021-04-19 | 2022-10-21 | 中国移动通信集团河北有限公司 | Definition index determination method, device, equipment and medium for video resources |
CN115225936B (en) * | 2021-04-19 | 2023-07-14 | 中国移动通信集团河北有限公司 | Method, device, equipment and medium for determining definition index of video resource |
WO2022267720A1 (en) * | 2021-06-23 | 2022-12-29 | 华为技术有限公司 | Data transmission method and communication apparatus |
CN115278354A (en) * | 2022-06-14 | 2022-11-01 | 北京大学 | Method and system for evaluating and optimizing video transmission quality based on user behavior index |
CN115278354B (en) * | 2022-06-14 | 2024-06-28 | 北京大学 | Method and system for evaluating and optimizing video transmission quality based on user behavior index |
CN116017092A (en) * | 2022-12-12 | 2023-04-25 | 中国科学院计算技术研究所 | QoE quantization method and transmission system parameter adjustment method based on QoE quantization method |
CN116074586A (en) * | 2022-12-13 | 2023-05-05 | 北京中电飞华通信有限公司 | Video experience quality prediction method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105897736A (en) | Method and device for assessing quality of experience (QoE) of TCP (Transmission Control Protocol) video stream service | |
US11670185B2 (en) | Adaptive machine learning system | |
US11829874B2 (en) | Neural architecture search | |
CN106023015B (en) | Course learning path recommendation method and device | |
CN100588271C (en) | System and method for measuring video quality based on packet measurement and image measurement | |
US8832117B2 (en) | Apparatus, systems and methods for interactive dissemination of knowledge | |
Yang et al. | Survey on QoE assessment approach for network service | |
CN109242710B (en) | Social network node influence ordering method and system | |
CN108428227A (en) | Non-reference picture quality appraisement method based on full convolutional neural networks | |
US20210158714A1 (en) | Method and Apparatus for Inquiry Driven Learning | |
Chang et al. | Active learning for crowdsourced QoE modeling | |
CN110110899B (en) | Knowledge mastery degree prediction method, adaptive learning method and electronic equipment | |
Demirbilek et al. | Machine learning--based parametric audiovisual quality prediction models for real-time communications | |
JP5552717B2 (en) | Learning support device, learning support method, and program | |
CN105761154A (en) | Socialized recommendation method and device | |
Dorça et al. | A new approach to discover students learning styles in adaptive educational systems | |
Almond | Comparison of two MCMC algorithms for hierarchical mixture models | |
CN109190040A (en) | Personalized recommendation method and device based on coevolution | |
US20190111317A1 (en) | System and method of basketball testing | |
Ahmad et al. | Significance of Cross-Correlated QoS Configurations for Validating the Subjective and Objective QoE of Cloud Gaming Applications | |
Khalida et al. | Enhancing Usability of the Academic Information System at Bhayangkara University: A Design Thinking and System Usability Approach | |
Georgiadis et al. | Reinforcing stealth assessment in serious games | |
CN107133292A (en) | Object recommendation method and system | |
CN102724543B (en) | HMM-based method for implementing media quality analysis evaluation control in IP network | |
KR102683091B1 (en) | Method and device for classifying quality of online lecture based on machine learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160824 |
|
RJ01 | Rejection of invention patent application after publication |