CN103888846B - Wireless video streaming service self-adaption rate control method based on QoE - Google Patents
Wireless video streaming service self-adaption rate control method based on QoE Download PDFInfo
- Publication number
- CN103888846B CN103888846B CN201410076993.0A CN201410076993A CN103888846B CN 103888846 B CN103888846 B CN 103888846B CN 201410076993 A CN201410076993 A CN 201410076993A CN 103888846 B CN103888846 B CN 103888846B
- Authority
- CN
- China
- Prior art keywords
- bit rate
- coding bit
- packet loss
- trend
- qoe
- 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.)
- Active
Links
Abstract
The invention relates to a wireless video streaming service self-adaption rate control method based on the QoE. The method includes the steps of setting up a QoE evaluation model at a receiving end to be used for calculating the quality of user experience, periodically feeding the packet loss rate, the end-to-end one-way time delay and user experience quality information which are obtained at the receiving end back to a sending end through a real-time transmission control protocol, conducting subdivision on network states through the sending end according to the user experience quality and by combining the packet loss rate with the increase or decrease trend of the end-to-end one-way time delay, and judging the network congestion degree. When the user experience quality decreases to a threshold value, the sending end of the wireless video streaming service starts a coding bit rate adjustment unit and adjusts the coding bit rate in a self-adaption mode by taking corresponding strategies according to the monitored network congestion degree, and therefore the user experience quality is improved. According to the method, the user experience quality of the video steaming service can be accurately evaluated in real time, network congestion is reduced through the self-adaption rate control method, and the user experience quality is effectively improved.
Description
Technical field
The present invention relates to mobile communication technology field, more particularly, to a kind of wireless video streaming service adaptation based on QoE
Method of rate control.
Background technology
With developing rapidly of wireless video streaming business, wireless video conference, video monitoring etc. are applied in the world
Rise and gradually incorporate the life of people.But, due to wireless network bandwidth resource-constrained and its unstability, video stream traffic
Can be affected by network fluctuation during transmission big data quantity, even can cause the loss of packet when serious, lead to
The video flow quality of user's viewing can not be ensured well.Therefore, the transmission of research valid wireless video stream traffic controls
The video flowing Quality of experience that method is watched to lifting user is most important.
In order to realize the lifting to wireless video streaming quality of service, existing control technology is generally in transmitting terminal to video flowing
Transmission rate, i.e. coding bit rate, carry out self-adaptative adjustment, when bandwidth throughput abundance network be in idle condition when, depending on
Frequency streaming server improves coding bit rate to lift the definition of video playback；When bandwidth throughput deficiency network is in congestion shape
During state, video stream server reduces coding bit rate to reduce the packet loss causing because of congestion.
Current method of rate control be mostly concerned with how being lifted service quality (Quality of Service,
QoS), as minimized packet loss, maximize handling capacity etc..But because QoS is a technical specification, it describes network and exists
The ability of service is provided, it can not directly reflect the satisfaction to business for the user on the basis of ensureing professional skill.However,
The final purpose of wireless video streaming business service is available to customer satisfaction system experiences quality, standardization body of International Telecommunication Union
User experience quality (Quality of Experience, QoE) is defined as weighing the index of user's subjective feeling, it refers to
A kind of application being perceived by terminal use or the overall acceptable degree of business, it can intuitively reflect user to using business
Subjective sensation.QoE contributes to Virtual network operator and understands the affecting parameters closely related with user satisfaction, finally improves user
Loyalty to keep on top in the fierce market competition.Therefore, rate controlled is carried out to wireless video streaming business based on QoE
It is not only the emphasis of academia research, be also that Virtual network operator ensures good Quality of experience with scale of keeping here and extend one's service
Crucial.
Because wireless video streaming business transmission mechanism is complicated, its QoE is subject to code encoding/decoding mode, network condition, terminal parameter
Impact etc. numerous influence factors.What QoE assessment technology was studied is the relation between QoE and its influence factor.Current QoE comments
Estimate technology majority to carry out in video stream server side, but video server can not regarding of being watched of direct access user terminal
Frequency stream information, and the QoE influence factor that existing QoE assessment technology considers is not comprehensive, and this leads to the assessment accuracy of QoE not
High.In addition, there will be self-adaptive quadtree method not careful to network state division, lead to its method of rate control easily to exist
There is to cause the concussion adjustment to coding bit rate during acute variation in network state, thus affecting the Consumer's Experience of wireless video streaming
Quality.Therefore, the how QoE of accurate evaluation wireless video streaming business carry out efficient adaptive rate controlled, to lift use
The subjective feeling to business for the family, not yet good solution at present.
Content of the invention
The invention aims to overcoming the shortcomings of existing solution, provide a kind of wireless video streaming based on QoE
Service adaptation method of rate control.Method of the present invention is set up QoE assessment models in receiving terminal and is used for calculating user's body
The amount of checking the quality, by RTCP Real-time Transport Control Protocol (Real-time Transport Control Protocol, RTCP) periodically
Packet loss, end-to-end One Way Delay and user experience quality feedback of the information that receiving terminal is obtained by ground are to transmitting terminal, transmitting terminal root
The user experience quality calculating according to QoE assessment models, joint packet loss is with end-to-end One Way Delay growth trend to network state
It is finely divided and judge network congestion degree.When the user experience quality being calculated by QoE assessment models drops to certain threshold value,
The transmitting terminal of wireless video streaming business starts coding bit rate adjustment unit the network congestion degree according to monitoring, takes corresponding
Strategy be adaptively adjusted coding bit rate, to realize the lifting of user experience quality.
For achieving the above object, the self-adaptive quadtree method of the present invention comprises time delay trend analysiss unit, packet loss
Statistic unit, QoE assessment unit, coding bit rate adjustment unit；
Described time delay trend analysiss unit is used for judging that wireless video streaming business data packet is transferred to reception from transmitting terminal
The One Way Delay growth trend at end, specifically：By RTP (Real-time Transport Protocol,
RTP) header timestamp information obtain packet One Way Delay, then by comparision testing judge One Way Delay for increase or
One Way Delay growth trend is finally input to coded-bit as a kind of configured information of network congestion degree by the trend reducing
Rate adjustment unit；
Described packet loss statistic unit is used for calculating wireless video streaming business within each coding bit rate adjustment cycle
End-to-end packet loss, specifically：Packet loss is obtained by the sequence number information statistics of RTP header, as network
Layer affecting parameters are input to QoE assessment unit, and by weighted mean method, packet loss are smoothed reducing because of network
The packet loss concussion that load state is mutated and causes, the packet loss after smoothing processing is as a kind of instruction letter of network congestion degree
Breath, anti-from receiving terminal by RTCP Real-time Transport Control Protocol (Real-time Transport Control Protocol, RTCP)
It is fed to the coding bit rate adjustment unit of transmitting terminal；
Described QoE assessment unit obtains for the cross-layer affecting parameters of input are mapped as user experience quality by calculating
Point, i.e. Mean Opinion Score value (Mean Opinion Score, MOS), and Mean Opinion Score value is input to coding bit rate tune
Whole unit；
Described coding bit rate adjustment unit is entered to network state with One Way Delay growth trend for joint packet loss
Row segment and judge network congestion degree, in conjunction with QoE assessment unit output Mean Opinion Score value, take accordingly strategy from
Adaptively adjust coding bit rate.
The comprising the concrete steps that of the inventive method：
Step 1：In transmitting terminal, wireless video streaming is encoded using H.264 mode, become with Z kind different coding ratio
The video flowing sequence of special rate, with set L '={ l1,l2,...,lZ},(l1＜ l2＜ ... ＜ lZ) represent, L ' is coding bit rate
Set, arranges { l1,l2,...,lZValue cover video flow quality and be changed into best experienced encoding bitrate value from worst,
The coding bit rate randomly selecting user's initial request wireless video streaming business is lk(k=1,2 ..., Z)；Wherein, k is coding
Bit rate level；
Step 2：Adjust the cycle in each coding bit rate, user experience quality is calculated by QoE assessment unit；
The QoE influence factor of wireless video streaming business is numerous, and the present invention considers the end-to-end cross-layer impact ginseng of impact QoE
Number, including application layer parameter (coding bit rate, frame per second, resolution), network layer parameter (packet loss), video content features parameter
(temporal information, spatial information, monochrome information, colouring information) and terminal unit parameter (screen size).Wherein, coded-bit
Rate refers to the bit number of transmitted per unit time video, and frame per second refers to the frame number of video display per second, and resolution refers to terminal institute
The pixel quantity of display, packet loss refers to that lost data packet number accounts for the ratio of sent packet, and video content features are
The space of finger video, time, brightness, colouring information, terminal size refers to the actual size of terminal screen.Above-mentioned application layer parameter
Can be by analyzing decoder end bit stream information (RTP bag bit number, sampling time, RTP packet number with network layer parameter
Deng) obtain；Video content features pass through to calculate frame pixel difference, edge block message, monochrome information, color at Video Decoder end
Information retrieval；Terminal size information is passed through to inquire about the international mobile equipment identification number (International of subscriber terminal equipment
Mobile Equipment Identity, IMEI) obtain.QoE assessment models are arranged on receiving terminal by the present invention, therefore more
Press close to the actual impression of subjective user, can more accurately reflect the Quality of experience to video stream traffic for the user.
In step 2 pass through radial basis function neural network (Radial Basis Function Neural Networks,
RBFN) algorithm sets up QoE assessment models, and radial basis function neural network algorithm comprises input layer, hidden layer, output layer, sets
Input layer has N number of input, and hidden layer has L hidden unit, and output layer has an output, i.e. user experience quality.Pass through
Radial basis function neural network algorithm is set up QoE assessment models and is included two steps：Training process and test process, idiographic flow
As follows：
Step (2-1) training process uses cross-layer affecting parameters and the subjective user Quality of experience value of training set to QoE mould
Type is trained, and training set is expressed asWherein vector xsRepresent s-th of radial basis function neural network algorithm
Input, s=1,2 ..., S, S represent the number of input sample, xsIncluding N number of cross-layer parameter, expression formula is xs=[x1,
x2,...,xN]T,ysExpression input sample is xsCorresponding user experience quality score value, span 1 are drawn by method of subjective appraisal
≤ys≤ 5, the comprising the following steps that of training process：
(2-1-1) randomly select L initial cluster center, make iterationses t=1；
(2-1-2) calculate input sample xsWith cluster centre cmThe distance between (t), expression formula is | | xs-cm(t)||,m
=1,2 ..., L, wherein cmT () represents the cluster centre of m-th hidden unit of the t time iteration；
(2-1-3) according to minimal distance principle to input sample xsClassified, that is, as m (xs)=min | | xs-cm(t)||
When, xsIt is classified as m class, be expressed as xs∈Rm(t), wherein RmT () represents m-th Clustering Domain, | | | | represent Euclidean distance；
(2-1-4) recalculate the cluster centre of each hidden unitWherein NmPoly- for m-th
Class field RmInput sample number included in (t)；
If (2-1-5) cm(t+1)≠cmT (), proceeds to step (2-1-2)；Otherwise terminate iteration and determine RBF
M-th hidden unit cluster centre of neural network algorithm is cm, and proceed to step (2-1-6)；
(2-1-6) defining P is overlap coefficient, calculates extension constant by closest distance methodci(i=1,2 ..., P) represent and cmP closest cluster centre, σmRepresent m-th hidden unit
Extension constant；
(2-1-7) output weight w=H of radial basis function neural network algorithm is determined by method of least square+Y, wherein w
Represent by output weight w of m-th hidden unitmThe weight vector constituting, expression formula is w=[w1,w2,...,wL]T, m=1,
2 ..., L, y=[y1,y2,...,yS]T, H=[hsm]S×LRepresent input sample xsOutput h in m-th hidden unitsmConstitute
Hidden unit matrix, hsmExpression formula isHTThe transposition of representing matrix H, H+The pseudoinverse of representing matrix H,
Expression formula is H+=(HTH)-1HT.
Step (2-2) test process is after step (2-1) completes the training process of radial basis function neural network algorithm,
Test set sample is inputted QoE assessment unit, if the QoE assessed value being calculated by radial basis function neural network algorithm is true with QoE
Error between real-valued is less than default error convergence threshold value, then QoE assessment models are set up and finished.
Step 3：Within each coding bit rate adjustment cycle, will be checked the quality by the calculated user's body of QoE assessment unit
Amount MOS and QoE threshold mosthIt is compared, if MOS >=MOSth, then keep the coding bit rate of transmitting terminal constant, if MOS is ＜
MOSth, then transmitting terminal startup coding bit rate adjustment unit, is entered to network state with One Way Delay growth trend in conjunction with packet loss
Row segments and judges network congestion degree, takes corresponding strategy to be adaptively adjusted coding bit rate.
As MOS ＜ MOS described in step 3thWhen, the step of self-adaptive quadtree method is：
(3-1) (Paired Comparison Test, PCT) numerical value S is tested by paired comparisonPCTCompare test with difference
(Paired Difference Test, PDT) numerical value SPDTDetermine One Way Delay growth trend from transmitting terminal to receiving terminal for the packet
DT, if SPCT＞ 0.55 or SPDT＞ 0.44, then One Way Delay show a rising trend, be designated as DT=1, otherwise One Way Delay becomes in reduction
Gesture, is designated as DT=0, wherein SPCTWith SPDTExpression formula be respectively
One Way Delay { the D of the K packet that receiving terminal is received by the timestamp information record of RTP header1,D2,...,
DK, and be divided into according to the priority order of arrival of K packetGroup, the mediant of every group of packet One Way Delay
ForThe value rule of I (X) is
(3-2) the smooth packet loss in t-th coding bit rate adjustment cycle of calculating is
Wherein p1,p2,...,pqThe front q coding bit rate representing closest adjusts cycle, i.e. t, and t-1, t-2 ..., t-q+1 are individual
Coding bit rate adjusts cycle corresponding history packet loss, λiRepresent the weight coefficient of history packet loss, span 0 ＜ λi＜
1.
(3-3) pass through RTCP Real-time Transport Control Protocol packet loss, end-to-end One Way Delay and user experience quality information is all
Feed back to transmitting terminal from receiving terminal to phase property, adjust the initial time in cycle, the coding of transmitting terminal in the t+1 coding bit rate
Bit rate adjustment unit according to t-th coding bit rate being calculated by step (3-1) and (3-2) adjust the cycle unidirectional when
Prolong growth trend DT (t) and packet loss p (t), take corresponding strategy to be adaptively adjusted coding bit rate.Two packet loss of setting
Threshold value plAnd ph, wherein plRepresent that user experience quality reaches and be satisfied with MOShWhen corresponding packet loss, phRepresent user experience quality
Reach acceptable degree MOSthWhen corresponding packet loss, and have 0 ＜ pl＜ ph＜ 1,1 ＜ MOSth＜ MOSh＜ 5, coding bit rate
The concrete adjustable strategies of adjustment unit are：
If (3-3-1). p (t) ＜ pl, illustrate that network condition is good and packet loss is minimum, therefore can be encoded by increase
Bit rate is determined by One Way Delay growth trend lifting QoE, the amplitude of increase：
If (3-3-1-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, then
Coding bit rate is dramatically increased, expression formula is BR (t+1)=min { BR (t)+Δ1,Rmax, wherein BR (t) represents t-th
Coding bit rate in the coding bit rate adjustment cycle,Δ1It is inversely proportional to packet loss p (t)
Growth factor, to ensure the increasing in p ＜ p of coding bit ratelIn the range of dynamically adjust, realize in the case that network state is good
Significantly raise coding bit rate, it is to avoid because the low coding bit rate of continued for too much time causes the user experience quality of difference, Rmax
Effect be constraint coding bit rate adjustment, to prevent coding bit rate from infinitely increasing；
If (3-3-1-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, then
Increase coding bit rate by a small margin, expression formula is BR (t+1)=min { BR (t)+Δ2,Rmax, whereinΔ2It is the growth factor being inversely proportional to packet loss p (t), Δ2Ensure that adjustment
Coding bit rate BR (t+1) afterwards is less than the twice of initial value BR (t), and this adjustment mode avoids when network condition is deteriorated
Cause network congestion by excessively increasing coding bit rate；
If (3-3-2). pl＜ p (t) ＜ ph, illustrate that network there occurs severe congestion, need further according to One Way Delay
Growth trend takes different adjustable strategies：
If (3-3-2-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend and having of improving
Ability alleviates severe congestion, in order to avoid frequently being adjusted the video flow quality fluctuation bringing by coding bit rate, now keeps former
Some coding bit rates are constant, i.e. BR (t+1)=BR (t)；
If (3-3-2-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, because
This should lower coding bit rate to be avoided congestion increases, because now network is in severe congestion state, then reduces coded-bit
Rate is BR (t+1)=max { (BR (t)-Δ3×BR(t)),Rmin, wherein Δ3It is the descending factors with exponential form, expression
Formula isΔ3Coding bit rate BR (t+1) can not only be realized with packet loss p
The increase of (t) and reduce, but also enable the increase with packet loss p (t), BR (t) fall increases, can protect simultaneously
Coding bit rate BR (t+1) after card adjustment is not less than the half of initial value BR (t).This adjustment mode can not only be light in network
Degree congestion when reduce coding bit rate, and avoid coding bit rate lower too low video flow quality is caused damage, Rmin's
Effect is constraint coding bit rate adjustment, to prevent coding bit rate from infinitely reducing；
If (3-3-3). p ＞ ph, illustrate that network excessively there occurs heavy congestion due to packet loss, need further according to list
Take different adjustable strategies to time delay growth trend：
If (3-3-3-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, but
Because network congestion seriously creates excessive packet loss, therefore still needing to reduction coding bit rate to alleviate network congestion, expression formula is
BR (t+1)=max { Δ × BR (t), Rmin, wherein Δ=max (Δ4,Δ5),
If (3-3-3-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, this
When should to reduce coding bit rate by a larger margin, expression formula be BR (t+1)=max { Δ × BR (t), Rmin), Δ=min (Δ4,
Δ5)；
Above-mentioned occur heavy congestion, i.e. p ＞ phWhen, Δ is the descending factors with multiplier form, it is ensured that coding
Bit rate BR (t+1) at least have dropped the half of initial value BR (t), realizes reducing packet loss thus alleviating network congestion.
Step 4：Repeat the above steps 2～step 3, until wireless video streaming service request terminates.
Wireless video streaming business based on QoE proposed by the invention be can be seen that by above-mentioned provided technical scheme
The advantage of self-adaptive quadtree method is：
First, to lift QoS as target, this is important to have ignored user's subjective feeling for existing conventional rate control method
Feature, different from the method for rate control based on QoS, method proposed by the invention considers user's viewing wireless video streaming industry
The subjective feeling of business, with lifted QoE as target design adaptive method of rate control.
Second, QoE assessment models proposed by the invention are located at receiving terminal, and so setting can be seen with direct access user
The video stream traffic parameter information seen, can more press close to the actual impression of subjective user, therefore QoE assessment models can be real
When, exactly reflect the Quality of experience to video stream traffic for the user.
3rd, consider end-to-end cross-layer impact by the QoE assessment models that the RBFN algorithm in machine Learning Theory is set up
Comprehensively, including video streaming content feature, application layer, Internet and terminal unit parameter, above-mentioned cross-layer parameter is from decoder for parameter
Bit stream information is extracted with mobile terminal side, and does not need source video information to make reference, and is therefore embodied as simple.This
Outward, the QoE assessment models set up can learn to extract and approach the relation between input and output by training process, therefore
QoE assessment accuracy is high.
4th, the wireless video streaming service adaptation method of rate control based on QoE proposed by the invention is to lift QoE
For target, combine packet loss and network state is finely divided and judges network congestion degree with One Way Delay growth trend.By
When the user experience quality that QoE assessment models calculate drops to certain threshold value, video stream traffic transmitting terminal starts adjustment unit simultaneously
According to the network congestion degree of monitoring, corresponding strategy is taken to make correct response to the network state of subdivision, therefore, it is possible to effective
The user experience quality of lifting video stream traffic.
The method being proposed of the present invention solves inaccurate to user experience quality assessment in current method of rate control
Problem, network state can be finely divided and accurately judge network congestion degree, and then take corresponding strategy adaptively
Adjustment coding bit rate, is effectively realized the lifting of user experience quality.
Brief description
Fig. 1 is the wireless video streaming service adaptation speed control system block diagram based on QoE of the present invention.
Fig. 2 is adaptive coding bit rate adjustment flow chart of the present invention.
Fig. 3 is end-to-end wireless video streaming business transmission instance block diagram of the present invention.
Fig. 4 is the system block diagram of QoE assessment models of the present invention.
Fig. 5 is QoE assessment models Establishing process figure of the present invention.
Specific embodiment
For making technical scheme, purpose and advantage clearer, below in conjunction with the concrete reality to the present invention for the accompanying drawing
The mode of applying is described in further detail.
The present invention is directed to the deficiency in current method of rate control, proposes on the basis of accurate evaluation user experience quality
A kind of wireless video streaming service adaptation method of rate control based on QoE, can be finely divided to network state and accurately
Judge the Congestion Level SPCC of network, take corresponding strategy to be adaptively adjusted coding bit rate, thus realizing the Quality of experience of user
Lifting.The present invention is applied to wireless network scenario, such as WLAN, WCDMA, CDMA2000, TD-SCDMA, LTE network etc..Described
The system block diagram of the self-adaptive quadtree method based on QoE is as shown in figure 1, comprise following functions unit：
Described time delay trend analysiss unit is used for judging that wireless video streaming business data packet is transferred to reception from transmitting terminal
The One Way Delay growth trend at end, specifically：Obtain the One Way Delay of packet by the timestamp information of RTP header, then lead to
Cross comparision testing and judge that One Way Delay is the trend increasing or reducing, finally using One Way Delay growth trend as network congestion
A kind of configured information of degree is input to coding bit rate adjustment unit；
Described packet loss statistic unit is used for calculating wireless video streaming business within each coding bit rate adjustment cycle
End-to-end packet loss, specifically：Packet loss is obtained by the sequence number information statistics of RTP header, affects ginseng as Internet
Number is input to QoE assessment unit, and by weighted mean method, packet loss is smoothed reducing because of Network load status
The packet loss concussion being mutated and causing, the packet loss after smoothing processing, as a kind of configured information of network congestion degree, passes through
RTCP feeds back to the coding bit rate adjustment unit of transmitting terminal from receiving terminal；
Described QoE assessment unit obtains for the cross-layer affecting parameters of input are mapped as user experience quality by calculating
Divide MOS, and MOS is input to coding bit rate adjustment unit；
Described coding bit rate adjustment unit is entered to network state with One Way Delay growth trend for joint packet loss
Row segments and judges network congestion degree, in conjunction with the MOS of QoE assessment unit output, takes strategy accordingly adaptively to adjust
Whole coding bit rate.
Adaptive coding bit rate adjustment flow chart of the present invention is as shown in Fig. 2 comprise the following steps that：
Step 1：In transmitting terminal, wireless video streaming is encoded to the video flowing sequence with Z=6 kind coding bit rate, with compiling
Code bit rate set L '={ 85.1kbps, 129.9kbps, 256.3kbps, 392.8kbps, 510.4kbp, 1030.2kbps }
Represent, this six values cover video flow quality and are changed into best experienced encoding bitrate value from worst, randomly select use
The coding bit rate of family initial request wireless video streaming business is 256.3kbps.
Step 2：Setting coding bit rate adjustment cycle T=1s, periodically calculates Consumer's Experience by QoE assessment models
Quality.
Fig. 3 is end-to-end wireless video streaming business transmission instance block diagram of the present invention, the QoE assessment models of the present invention
Be arranged on receiving terminal, using end-to-end cross-layer impact affecting parameters as input, include applying layer parameter (coding bit rate, frame per second,
Resolution), network layer parameter (packet loss), video content features parameter (temporal information, spatial information, monochrome information, color letter
Breath) and terminal unit parameter (screen size).Wherein, coding bit rate refers to the bit number of transmitted per unit time video,
Frame per second refers to the frame number of video display per second, and resolution refers to the pixel quantity shown by terminal, and packet loss refers to lost number
Account for the ratio of sent packet according to bag quantity, video content features refer to the space of video, time, brightness, colouring information, eventually
End size refers to the actual size of terminal screen.Above-mentioned application layer parameter and network layer parameter can be by analyzing decoder end ratio
Special stream information (RTP bag bit number, sampling time, RTP packet number etc.) obtains；Video content features pass through in Video Decoder
End calculates frame pixel difference, edge block message, monochrome information, colouring information extraction；Terminal size information is passed through to inquire about user eventually
The international mobile equipment identification number IMEI of end equipment obtains.
QoE assessment models system block diagram described in step 2 is as shown in Figure 4.
QoE assessment models are set up by RBFN algorithm, RBFN algorithm comprises input layer, hidden layer, output layer, set input
Layer has N number of input, and hidden layer has L hidden unit, and output layer has an output, i.e. user experience quality.By RBFN
Algorithm is set up QoE assessment models and is included two steps：Training process and test process, QoE assessment models Establishing process such as Fig. 5 institute
Show, comprise the following steps that：
Step (2-1) produces data set：Different cross-layer parameters are set, and wherein coding bit rate span is
16kbps-640kbps, frame per second can use 5-30fps, and resolution can use QCIF, CIF, 4CIF, packet loss span 0-20%,
Terminal size can use 110x50mm-250 × 200mm, and video content types are desirable at a slow speed, middling speed, the video flowing of quick motion.?
Under the various combination of cross-layer parameter, MOS scoring is carried out by the video flow quality that method of subjective appraisal is watched to user, MOS takes
Value scope [1,5], thus produce training set and test set sample data.Using training set sample data, model is instructed repeatedly
Practice, and determine cluster centre and extension constant, the comprising the following steps that of training process：
(2-1-1) the cluster centre number of input RBFN algorithm, overlap coefficient, stopping criterion for iteration, default error convergence
Threshold value；By the use of in data set, as training sample, training sample set is expressed as 80% sampleWherein vector xs
Represent s-th input of RBFN algorithm, s=1,2 ..., S, S represent the number of input sample, xsIncluding N number of cross-layer parameter, table
Reaching formula is xs=[x1,x2,...,xN]T,ysExpression input sample is xsCorresponding user experience quality is drawn by method of subjective appraisal
Score value.
(2-1-2) randomly select L initial cluster center, make iterationses t=1；
(2-1-3) calculate input sample xsWith cluster centre cmThe distance between (t), expression formula is | | xs-cm(t)||,m
=1,2 ..., L, wherein cmT () represents the cluster centre of m-th hidden unit of the t time iteration；
(2-1-4) according to minimal distance principle to input sample xsClassified, that is, as m (xs)=min | | xs-cm(t)||
When, xsIt is classified as m class, be expressed as xs∈Rm(t), wherein RmT () represents m-th Clustering Domain, | | | | represent Euclidean distance；
(2-1-5) recalculate the cluster centre of each hidden unitWherein NmCluster for m-th
Domain RmInput sample number included in (t)；
If (2-1-6) cm(t+1)≠cmT (), proceeds to step (2-1-3)；Otherwise terminate iteration and determine RBFN algorithm
M-th hidden unit cluster centre is cm, and proceed to step (2-1-7)；
(2-1-7) defining P is overlap coefficient, calculates extension constant by closest distance method
ci(i=1,2 ..., P) represent and cmP closest cluster centre, σmRepresent the extension constant of m-th hidden unit；
(2-1-8) output weight w=H of RBFN algorithm is determined by method of least square+Y, wherein w represent hidden by m-th
Output weight w of unitmThe weight vector constituting, expression formula is w=[w1,w2,...,wL]T, m=1,2 ..., L, y=[y1,
y2,...,yS]T, H=[hsm]S×LRepresent input sample xsOutput h in m-th hidden unitsmThe hidden unit matrix constituting, hsmTable
Reaching formula isHTThe transposition of representing matrix H, H+The pseudoinverse of representing matrix H, expression formula is H+=
(HTH)-1HT.
Step (2-2) test process is after step (2-1) completes the training process of RBFN algorithm, will be remaining in data set
20% sample input QoE assessment models as test sample, if the QoE assessed value being calculated by RBFN algorithm and QoE actual value
Between error preset error convergence threshold value, then QoE assessment models set up finish, step-up error convergence threshold be 10-3.
Step 3：Within each coding bit rate adjustment cycle, will be checked the quality by the calculated user's body of QoE assessment models
Amount MOS and QoE threshold mosthIt is compared, MOS is setth=3.5, represent that user experience quality reaches acceptable threshold value, if
MOS≥MOSth, then keep the coding bit rate of transmitting terminal constant, if MOS is ＜ MOSth, then transmitting terminal startup coding bit rate tune
Whole unit, is finely divided to network state and judges network congestion degree in conjunction with packet loss and One Way Delay growth trend, take
Corresponding strategy is adaptively adjusted coding bit rate.
As MOS ＜ MOS described in step 3thWhen, the step of self-adaptive quadtree method is：
(3-1) PCT numerical value S is tested by paired comparisonPCTCompare test PDT numerical value S with differencePDTDetermine packet from send out
Sending end to the One Way Delay growth trend DT of receiving terminal, if SPCT＞ 0.55 or SPDT＞ 0.44, then One Way Delay show a rising trend,
It is designated as DT=1, otherwise One Way Delay is in reduction trend, is designated as DT=0, wherein SPCTWith SPDTExpression formula be respectivelyReceiving terminal is received by the timestamp information record of Real-time Transport Protocol header
One Way Delay { the D of the K packet arriving1,D2,...,DK, and be divided into according to the priority order of arrival of K packetGroup, the mediant of every group of packet One Way Delay is The value rule of I (X) is
(3-2) the smooth packet loss in t-th coding bit rate adjustment cycle of calculating is
Wherein p1,p2,...,pqThe front q coding bit rate representing closest adjusts cycle, i.e. t, and t-1, t-2 ..., t-q+1 are individual
Coding bit rate adjusts cycle corresponding history packet loss, λiRepresent the weight coefficient of history packet loss, span 0 ＜ λi＜
1, q=8, weight coefficient can be set herein
(3-3) pass through rtcp protocol by packet loss, end-to-end One Way Delay and user experience quality property information cycle from
Receiving terminal feeds back to transmitting terminal, adjusts the initial time in cycle in the t+1 coding bit rate, and the coding bit rate of transmitting terminal is adjusted
Whole unit becomes according to the One Way Delay increase and decrease that t-th coding bit rate being calculated with (3-2) by step (3-1) adjusts the cycle
Gesture DT (t) and packet loss p (t), take corresponding strategy to be adaptively adjusted coding bit rate.Two packet loss threshold values p of settinglWith
ph, wherein plRepresent that user experience quality reaches and be satisfied with MOShCorresponding packet loss, p when=4.0hRepresent that user experience quality reaches
To acceptable thresholds MOSthCorresponding packet loss when=3.5, value is respectively pl=1.04%, ph=2.11%.Coded-bit
The concrete adjustable strategies of rate adjustment unit are：
If (3-3-1). p (t) ＜ pl, illustrate that network condition is good and packet loss is minimum, therefore can be encoded by increase
Bit rate is determined by One Way Delay growth trend lifting QoE, the amplitude of increase：
If (3-3-1-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, then
Coding bit rate is dramatically increased, expression formula is BR (t+1)=min { BR (t)+Δ1,Rmax, wherein BR (t) represents t-th
Coding bit rate in the coding bit rate adjustment cycle,Δ1It is inversely proportional to packet loss p (t)
Growth factor, to ensure the increasing in p ＜ p of coding bit ratelIn the range of dynamically adjust, realize in the case that network state is good
Significantly raise coding bit rate, it is to avoid because the low coding bit rate of continued for too much time causes the user experience quality of difference, Rmax
Effect be constraint coding bit rate adjustment, to prevent coding bit rate from infinitely increasing, take Rmax=1030.2kbps；
If (3-3-1-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, then
Increase coding bit rate by a small margin, expression formula is BR (t+1)=min { BR (t)+Δ2,Rmax, whereinΔ2It is the growth factor being inversely proportional to packet loss p (t), Δ2Ensure that adjustment
Coding bit rate BR (t+1) afterwards is less than the twice of initial value BR (t), and this adjustment mode avoids when network condition is deteriorated
Cause network congestion by excessively increasing coding bit rate；
If (3-3-2). pl＜ p (t) ＜ ph, illustrate that network there occurs severe congestion, need further according to One Way Delay
Growth trend takes different adjustable strategies：
If (3-3-2-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend and having of improving
Ability alleviates severe congestion, in order to avoid frequently being adjusted the video flow quality fluctuation bringing by coding bit rate, now keeps former
Some coding bit rates are constant, i.e. BR (t+1)=BR (t)；
If (3-3-2-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, because
This should lower coding bit rate to be avoided congestion increases, because now network is in severe congestion state, then reduces coded-bit
Rate is BR (t+1)=max { (BR (t)-Δ3×BR(t)),Rmin, wherein Δ3It is the descending factors with exponential form, expression
Formula isΔ3Coding bit rate BR (t+1) can not only be realized with packet loss p
The increase of (t) and reduce, but also enable the increase with packet loss p (t), BR (t) fall increases, can protect simultaneously
Coding bit rate BR (t+1) after card adjustment is not less than the half of initial value BR (t).This adjustment mode can not only be light in network
Degree congestion when reduce coding bit rate, and avoid coding bit rate lower too low video flow quality is caused damage, Rmin's
Effect is constraint coding bit rate adjustment, to prevent coding bit rate from infinitely reducing, takes Rmin=85.1kbps；
If (3-3-3). p ＞ ph, illustrate that network excessively there occurs heavy congestion due to packet loss, need further according to list
Take different adjustable strategies to time delay growth trend：
If (3-3-3-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, but
Because network congestion seriously creates excessive packet loss, therefore still needing to reduction coding bit rate to alleviate network congestion, expression formula is
BR (t+1)=max { Δ × BR (t), Rmin, wherein Δ=max (Δ4,Δ5),
If (3-3-3-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, this
When should to reduce coding bit rate by a larger margin, expression formula be BR (t+1)=max { Δ × BR (t), Rmin), Δ=min (Δ4,
Δ5)；
Above-mentioned occur heavy congestion, that is, as p ＞ phWhen, Δ is the descending factors with multiplier form, it is ensured that compiling
Code bit rate BR (t+1) at least have dropped the half of initial value BR (t), realizes reducing packet loss thus alleviating network congestion.
Step 4：Repeat the above steps 2～step 3, until wireless video streaming service request terminates.
Claims (1)
1. based on the wireless video streaming service adaptation method of rate control of QoE it is characterised in that comprising the concrete steps that of the method：
Step 1：In transmitting terminal, wireless video streaming is encoded using H.264 mode, become with Z kind different coding bit rate
Video flowing sequence, with set L '={ l1,l2,...,lZ},(l1＜ l2＜ ... ＜ lZ) represent, L ' is coding bit rate set,
Setting { l1,l2,...,lZValue cover video flow quality and be changed into best experienced encoding bitrate value from worst, at random
The coding bit rate choosing user's initial request wireless video streaming business is lk(k=1,2 ..., Z)；Wherein, k is coded-bit
Rate grade；
Step 2：Adjusting the cycle in each coding bit rate, user experience quality being calculated by QoE assessment unit it is simply that passing through footpath
Set up QoE assessment models to basis function neural network algorithm, radial basis function neural network algorithm comprise input layer, hidden layer,
Output layer, sets input layer and has N number of input, hidden layer has L hidden unit, and output layer has an output, i.e. user's body
The amount of checking the quality；
QoE assessment models are set up by radial basis function neural network algorithm and includes training process and test process, idiographic flow
As follows：
Step (2-1) training process is entered to QoE model with subjective user Quality of experience value using the cross-layer affecting parameters of training set
Row training, training set is expressed asWherein vector xsRepresent s-th input of radial basis function neural network algorithm,
S=1,2 ..., S, S represent the number of input sample, xsIncluding N number of cross-layer parameter, expression formula is xs=[x1,x2,...,xN]T,
ysExpression input sample is xsCorresponding user experience quality score value, span 1≤y are drawn by method of subjective appraisals≤ 5, training
The comprising the following steps that of process：
(2-1-1) randomly select L initial cluster center, make iterationses t=1；
(2-1-2) calculate input sample xsWith cluster centre cmThe distance between (t), expression formula is | | xs-cm(t) | |, m=1,
2 ..., L, wherein cmT () represents the cluster centre of m-th hidden unit of the t time iteration；
(2-1-3) according to minimal distance principle to input sample xsClassified, that is, as m (xs)=min | | xs-cm(t) | | when, xs
It is classified as m class, be expressed as xs∈Rm(t), wherein RmT () represents m-th Clustering Domain, | | | | represent Euclidean distance；
(2-1-4) recalculate the cluster centre of each hidden unitWherein NmFor m-th Clustering Domain Rm
Input sample number included in (t)；
If (2-1-5) cm(t+1)≠cmT (), proceeds to step (2-1-2)；Otherwise terminate iteration and determine Radial Basis Function neural
M-th hidden unit cluster centre of network algorithm is cm, and proceed to step (2-1-6)；
(2-1-6) defining P is overlap coefficient, calculates extension constant by closest distance methodci
(i=1,2 ..., P) represent and cmP closest cluster centre, σmRepresent the extension constant of m-th hidden unit；
(2-1-7) output weight w=H of radial basis function neural network algorithm is determined by method of least square+Y, wherein w represent
Output weight w by m-th hidden unitmThe weight vector constituting, expression formula is w=[w1,w2,...,wL]T, m=1,2 ...,
L, y=[y1,y2,...,yS]T, H=[hsm]S×LRepresent input sample xsOutput h in m-th hidden unitsmThe hidden unit constituting
Matrix, hsmExpression formula isHTThe transposition of representing matrix H, H+The pseudoinverse of representing matrix H, expression formula
For H+=(HTH)-1HT；
Step (2-2) test process is after step (2-1) completes the training process of radial basis function neural network algorithm, will survey
Examination collection sample input QoE assessment unit, if the QoE assessed value being calculated by radial basis function neural network algorithm and QoE actual value
Between error be less than default error convergence threshold value, then QoE assessment models are set up and are finished；
Step 3：Within each coding bit rate adjustment cycle, will be by QoE assessment unit calculated user experience quality MOS
With QoE threshold mosthIt is compared；
If MOS >=MOSth, then keep the coding bit rate of transmitting terminal constant；
If MOS is ＜ MOSth, then transmitting terminal startup coding bit rate adjustment unit, is adaptively adjusted coding bit rate, concrete step
Suddenly it is：
(3-1) pass through paired comparison test number SPCTCompare test number S with differencePDTDetermine packet from transmitting terminal to reception
The One Way Delay growth trend DT at end, if SPCT＞ 0.55 or SPDT＞ 0.44, then One Way Delay show a rising trend, be designated as DT=1,
Otherwise One Way Delay is in reduction trend, is designated as DT=0, wherein SPCTWith SPDTExpression formula be respectivelyReceiving terminal passes through the timestamp information of RTP header
Record the One Way Delay { D of the K packet receiving1,D2,...,DK, and according to K packet priority order of arrival by its
It is divided intoGroup, the mediant of every group of packet One Way Delay isK=1, the value rule of 2 ..., H, I (X)
It is
(3-2) the smooth packet loss in t-th coding bit rate adjustment cycle of calculating isI=1,2 ..., q, wherein
p1,p2,...,pqThe front q coding bit rate representing closest adjusts cycle, i.e. t, t-1, t-2 ..., t-q+1 coding
Bit rate adjusts cycle corresponding history packet loss, λiRepresent the weight coefficient of history packet loss, span 0 ＜ λi＜ 1；
(3-3) pass through RTCP Real-time Transport Control Protocol by packet loss, end-to-end One Way Delay and user experience quality property information cycle
Ground feeds back to transmitting terminal from receiving terminal, adjusts the initial time in cycle, the coded-bit of transmitting terminal in the t+1 coding bit rate
Rate adjustment unit increases according to the One Way Delay that t-th coding bit rate being calculated with (3-2) by step (3-1) adjusts the cycle
Subtract trend DT (t) and packet loss p (t), take corresponding strategy to be adaptively adjusted coding bit rate；Two packet loss threshold values of setting
plAnd ph, wherein plRepresent that user experience quality reaches and be satisfied with MOShWhen corresponding packet loss, phRepresent that user experience quality reaches
Acceptable degree MOSthWhen corresponding packet loss, and have 0 ＜ pl＜ ph＜ 1,1 ＜ MOSth＜ MOSh＜ 5, coding bit rate adjusts
The concrete adjustable strategies of unit are：
If (3-3-1). p (t) ＜ pl, illustrate that network condition is good and packet loss is minimum, therefore can be by increasing coding bit rate
To lift QoE, the amplitude of increase is determined by One Way Delay growth trend：
If (3-3-1-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, then significantly
Degree ground increases coding bit rate, and expression formula is BR (t+1)=min { BR (t)+Δ1,Rmax, wherein BR (t) represents t-th coding
Coding bit rate in the bit rate adjustment cycle,Δ1It is the growth being inversely proportional to packet loss p (t)
The factor, to ensure the increasing in p ＜ p of coding bit ratelIn the range of dynamically adjust, realize in the case that network state is good significantly
Raise coding bit rate, it is to avoid because the low coding bit rate of continued for too much time causes the user experience quality of difference, RmaxWork
With being constraint coding bit rate adjustment, to prevent coding bit rate from infinitely increasing；
If (3-3-1-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, then slightly
Degree ground increases coding bit rate, and expression formula is BR (t+1)=min { BR (t)+Δ2,Rmax, whereinΔ2It is the growth factor being inversely proportional to packet loss p (t), Δ2Ensure that adjustment
Coding bit rate BR (t+1) afterwards is less than the twice of initial value BR (t), and this adjustment mode avoids when network condition is deteriorated
Cause network congestion by excessively increasing coding bit rate；
If (3-3-2). pl＜ p (t) ＜ ph, illustrate that network there occurs severe congestion, need to be increased and decreased according to One Way Delay further
Trend takes different adjustable strategies：
If (3-3-2-1) DT (t)=0, illustrate One Way Delay be reduction trend, that is, network condition be in improve trend and have the ability
Alleviate severe congestion, in order to avoid frequently being adjusted the video flow quality fluctuation bringing by coding bit rate, now keep original
Coding bit rate is constant, i.e. BR (t+1)=BR (t)；
If (3-3-2-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, therefore should
Lower coding bit rate to avoid congestion increases, because now network is in severe congestion state, then reducing coding bit rate is
BR (t+1)=max { (BR (t)-Δ3×BR(t)),Rmin, wherein Δ3It is the descending factors with exponential form, expression formula isΔ3Coding bit rate BR (t+1) can not only be realized with packet loss p (t)
Increase and reduce, but also enable the increase with packet loss p (t), BR (t) fall increases, and ensure that tune simultaneously
Coding bit rate BR (t+1) after whole is optionally greater than the half of initial value BR (t)；RminEffect be constraint coding bit rate adjustment,
To prevent coding bit rate from infinitely reducing；
If (3-3-3). p ＞ ph, illustrate that network excessively there occurs heavy congestion due to packet loss, need further according to unidirectional when
Prolong growth trend and take different adjustable strategies：
If (3-3-3-1) DT (t)=0, illustrate that One Way Delay is reduction trend, that is, network condition is in the trend that improves, but due to
Network congestion seriously creates excessive packet loss, therefore still needs to reduce coding bit rate to alleviate network congestion, expression formula is BR (t+
1)=max { Δ × BR (t), Rmin, wherein Δ=max (Δ4,Δ5),
If (3-3-3-2) DT (t)=1, illustrate that One Way Delay is increase trend, that is, network condition is in variation trend, now should
To reduce coding bit rate by a larger margin, expression formula is BR (t+1)=max { Δ × BR (t), Rmin), Δ=min (Δ4,Δ5)；
Above-mentioned occur heavy congestion, i.e. p ＞ phWhen, Δ is the descending factors with multiplier form, it is ensured that coding bit rate
BR (t+1) at least have dropped the half of initial value BR (t), realizes reducing packet loss thus alleviating network congestion；
Step 4：Repeat the above steps 2～step 3, until wireless video streaming service request terminates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410076993.0A CN103888846B (en) | 2014-03-04 | 2014-03-04 | Wireless video streaming service self-adaption rate control method based on QoE |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410076993.0A CN103888846B (en) | 2014-03-04 | 2014-03-04 | Wireless video streaming service self-adaption rate control method based on QoE |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103888846A CN103888846A (en) | 2014-06-25 |
| CN103888846B true CN103888846B (en) | 2017-02-08 |
Family
ID=50957534
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410076993.0A Active CN103888846B (en) | 2014-03-04 | 2014-03-04 | Wireless video streaming service self-adaption rate control method based on QoE |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103888846B (en) |
Families Citing this family (39)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105611322A (en) * | 2014-11-14 | 2016-05-25 | 台湾艾特维股份有限公司 | Video bandwidth adjustment device and adjustment method thereof |
| CN104519325B (en) * | 2014-12-29 | 2018-12-07 | 大唐移动通信设备有限公司 | A kind of adaptive support method of wireless video monitoring system based on 4G network |
| GB2542984B (en) * | 2015-07-31 | 2020-02-19 | Imagination Tech Ltd | Identifying network congestion based on a processor load and receiving delay |
| CN105049930B (en) * | 2015-08-14 | 2018-03-06 | 浙江大学 | Wireless video streaming business QoE appraisal procedures based on SVMs |
| CN105262699B (en) * | 2015-10-29 | 2018-07-03 | 浙江大华技术股份有限公司 | A kind of network self-adapting code adjustment method and device |
| CN108605114B (en) | 2016-01-25 | 2020-04-21 | 华为技术有限公司 | Control method, control device and network controller |
| EP3413620A4 (en) * | 2016-02-29 | 2019-01-16 | Huawei Technologies Co., Ltd. | Video optimization method, user equipment unit and network equipment unit |
| CN105812097B (en) * | 2016-03-16 | 2019-04-23 | 北京邮电大学 | A kind of AMR code rate self-adapting regulation method based on network state |
| CN105871744A (en) * | 2016-03-31 | 2016-08-17 | 北京小米移动软件有限公司 | Method and device for controlling traffic dynamically |
| CN105828069B (en) * | 2016-04-06 | 2019-01-18 | 上海交通大学 | A kind of encoder self-adapting regulation method based on subjective quality assessment |
| CN105933329B (en) * | 2016-06-12 | 2018-11-20 | 浙江大学 | Video stream traffic code rate adaptive approach based on on-line study |
| CN107770473B (en) * | 2016-08-16 | 2020-05-26 | 大唐移动通信设备有限公司 | Audio and video data transmission control method and device |
| CN108064058B (en) * | 2016-11-07 | 2022-11-01 | 中兴通讯股份有限公司 | Congestion control method and device and base station |
| CN106533963B (en) * | 2017-01-11 | 2019-05-03 | 深圳云视融通科技有限公司 | A kind of method for controlling network congestion of streaming media |
| CN106933585B (en) * | 2017-03-07 | 2020-02-21 | 吉林大学 | Self-adaptive multi-channel interface selection method under distributed cloud environment |
| CN108574840B (en) * | 2017-03-10 | 2020-03-10 | 华为技术有限公司 | Method and device for evaluating video experience quality |
| CN107276910A (en) * | 2017-06-07 | 2017-10-20 | 上海迪爱斯通信设备有限公司 | The real-time adjusting apparatus of video code rate and system, video server |
| CN107888579B (en) * | 2017-11-06 | 2020-06-09 | 浙江大学 | Non-interference mobile video user experience quality index modeling method |
| CN107968944B (en) * | 2017-12-04 | 2020-01-17 | 深圳市瑞科慧联科技有限公司 | Image transmission method and system |
| US10602139B2 (en) * | 2017-12-27 | 2020-03-24 | Omnivision Technologies, Inc. | Embedded multimedia systems with adaptive rate control for power efficient video streaming |
| CN108495182A (en) * | 2018-03-23 | 2018-09-04 | 山西大学 | A kind of audio quality self-adjusting control method |
| CN108848415B (en) * | 2018-05-04 | 2021-01-15 | 北京大学 | Network real-time video transmission method and device for adaptive learning |
| EP3633999A1 (en) | 2018-10-05 | 2020-04-08 | InterDigital CE Patent Holdings | Method to be implemented at a device able to run one adaptive streaming session, and corresponding device |
| CN109120933B (en) * | 2018-10-11 | 2021-05-14 | 广州酷狗计算机科技有限公司 | Method, device and equipment for dynamically adjusting code rate and storage medium |
| CN111245769B (en) * | 2018-11-28 | 2022-09-30 | 深圳技威时代科技有限公司 | Audio and video transmission method and device and storage medium |
| CN111385574B (en) * | 2018-12-28 | 2021-08-10 | 北京字节跳动网络技术有限公司 | Code rate control method and device in video coding, mobile terminal and storage medium |
| CN109905326A (en) * | 2019-03-26 | 2019-06-18 | 武汉大学 | A kind of rate drawdown parameter optimization method based on the Congestion Level SPCC factor |
| CN110266714B (en) | 2019-06-28 | 2020-04-21 | 合肥工业大学 | QoE-driven VR video self-adaptive acquisition and transmission method |
| CN110324621B (en) * | 2019-07-04 | 2021-05-18 | 北京达佳互联信息技术有限公司 | Video encoding method, video encoding device, electronic equipment and storage medium |
| CN113497671B (en) * | 2020-04-02 | 2023-07-18 | 成都鼎桥通信技术有限公司 | Data processing method, device, equipment, storage medium and terminal equipment |
| CN111901642B (en) * | 2020-07-31 | 2022-09-30 | 成都云格致力科技有限公司 | Real-time video code rate self-adaptive control method and system based on reinforcement learning |
| CN112291495B (en) * | 2020-10-16 | 2021-08-03 | 厦门大学 | Wireless video low-delay anti-interference transmission method based on reinforcement learning |
| CN112565884B (en) * | 2020-11-27 | 2024-03-05 | 北京达佳互联信息技术有限公司 | Image processing method, device, terminal, server and storage medium |
| CN112911650A (en) * | 2021-03-28 | 2021-06-04 | 高小翎 | Mobile high-definition video intelligent bidirectional detection bandwidth control system |
| CN113473190B (en) * | 2021-08-05 | 2022-06-28 | 佳源科技股份有限公司 | Audio and video flexible transmission device and transmission method based on reinforcement learning |
| CN113660488B (en) * | 2021-10-18 | 2022-02-11 | 腾讯科技（深圳）有限公司 | Method and device for carrying out flow control on multimedia data and training flow control model |
| CN115002819B (en) * | 2021-10-20 | 2023-05-23 | 荣耀终端有限公司 | Network quality assessment method and device |
| CN116708843B (en) * | 2023-08-03 | 2023-10-31 | 清华大学 | User experience quality feedback regulation system in semantic communication process |
| CN116743609B (en) * | 2023-08-14 | 2023-10-17 | 清华大学 | QoE evaluation method and device for video streaming media based on semantic communication |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102547475A (en) * | 2012-02-21 | 2012-07-04 | 德科仕通信(上海)有限公司 | Method and system for improving quality of service alarming accuracy of Internet protocol (IP) video media stream |
| CN102802089A (en) * | 2012-09-13 | 2012-11-28 | 浙江大学 | Shifting video code rate regulation method based on experience qualitative forecast |
| CN103475951A (en) * | 2013-08-20 | 2013-12-25 | 南京邮电大学 | User-experience-based real-time video transmission rate self-adaption method |
-
2014
- 2014-03-04 CN CN201410076993.0A patent/CN103888846B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102547475A (en) * | 2012-02-21 | 2012-07-04 | 德科仕通信(上海)有限公司 | Method and system for improving quality of service alarming accuracy of Internet protocol (IP) video media stream |
| CN102802089A (en) * | 2012-09-13 | 2012-11-28 | 浙江大学 | Shifting video code rate regulation method based on experience qualitative forecast |
| CN103475951A (en) * | 2013-08-20 | 2013-12-25 | 南京邮电大学 | User-experience-based real-time video transmission rate self-adaption method |
Non-Patent Citations (2)
| Title |
|---|
| An artificial-neural-network-based QoE estimation model for video streaming over wireless networks;Yaqian Kang,Huifang Chen,Lei Xie;《2013 2nd IEEE/CIC International Conference on Communications in China》;20130814;264-269 * |
| Modeling and analysis of Skype video calls:rate control and video quality;Xinggong Zhang et.al;《IEEE Transactions on Multimedia》;20130220;1446-1457 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103888846A (en) | 2014-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103888846B (en) | Wireless video streaming service self-adaption rate control method based on QoE | |
| CN102802089B (en) | Shifting video code rate regulation method based on experience qualitative forecast | |
| Khan et al. | Quality of experience-driven adaptation scheme for video applications over wireless networks | |
| CN106134130B (en) | The system and method that dynamic effective speed for real-time video flow is estimated | |
| CN105828069B (en) | A kind of encoder self-adapting regulation method based on subjective quality assessment | |
| US9419875B2 (en) | Video quality assessment method and apparatus | |
| US20150082345A1 (en) | System for generating enhanced advertizements and methods for use therewith | |
| US8255559B2 (en) | Data streaming through time-varying transport media | |
| Ramakrishnan et al. | SDN based QoE optimization for HTTP-based adaptive video streaming | |
| US20140181266A1 (en) | System, streaming media optimizer and methods for use therewith | |
| US20130304934A1 (en) | Methods and systems for controlling quality of a media session | |
| CN102710374B (en) | Speed control method in wireless streaming media transmission | |
| WO2020220902A1 (en) | Method and apparatus for distributing transmission parameters of video resources | |
| CN103475951A (en) | User-experience-based real-time video transmission rate self-adaption method | |
| TW201212636A (en) | Method and an apparatus for transferring a video stream | |
| US10116715B2 (en) | Adapting encoded bandwidth | |
| CN110312131B (en) | Content self-adaptive online video coding method based on deep learning | |
| CN107087192A (en) | Target bit rate method of adjustment and device | |
| Sobhani et al. | A video bitrate adaptation and prediction mechanism for HTTP adaptive streaming | |
| Kimura et al. | Balancing quality of experience and traffic volume in adaptive bitrate streaming | |
| US20160277467A1 (en) | Adapting Encoded Bandwidth | |
| CN104871502B (en) | The strong solution based on content for dynamic optimization multi-user radio multimedia transmission | |
| Liu et al. | Streaming variable bitrate video over mobile networks with predictable performance | |
| Zhang et al. | A QOE-driven approach to rate adaptation for dynamic adaptive streaming over http | |
| Feng-Hui et al. | QoE issues of OTT services over 5G network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |