CN103795610B - The instant messaging dispatching method of a kind of Consumer's Experience justice and device - Google Patents

Abstract

The invention discloses the instant messaging dispatching method that a kind of Consumer's Experience is fair, comprise the steps: (1) normalization quality score according to each subservice, calculate the attention rate increment of each subservice of present period；According to this attention rate increment and user's attention rate of previous period, calculate user's attention rate of present period；(2) choose present period user's attention rate the highest subservice representative business as user, make this represent the representative quality of service scoring as user of the normalization quality score of business；(3) subservice representing the worst user of quality of service scoring is scheduling.A kind of instant messaging dispatching device, including data queue's Delay maintenance module, Consumer's Experience estimation module and subservice data stream scheduling module.Can there is multiple instant message user in the present invention, each user uses under the scene of one or more subservice in a network, estimates Consumer's Experience, and realizes Consumer's Experience justice by data stream scheduling.

Description

The instant messaging dispatching method of a kind of Consumer's Experience justice and device

Technical field

The present invention relates to the communications field, be specifically related to instant messaging dispatching method and realization thereof in a kind of wireless network Device.

Background technology

Instant messaging is one of the most widely used business in current mobile Internet.Its number of users is huge, As a example by Skype, its simultaneously online user number be up to 7,000 ten thousand.Instant messaging refer to comprise video conversation, voice dialogue with And the mixed type particular service of three kinds of instant message subservices (hereinafter referred to as subservice) such as text.With common data services (as Ftp and email) different, instant messaging is conversational interactive service, has requirement of real-time.But be accepted after permissible The real time business obtaining dedicated channel or bandwidth safeguard is different, and instant messaging user needs to compete shared channel width, do not has There is Quality Ensuring Mechanism.Therefore, dispatching technique is particularly important for instant messaging quality and Consumer's Experience.

Existing concurrently with under the network scenarios of multiple instant message user, design makes user obtain close service experience tune Degree is a technological challenge, and its reason is as follows:

(1) Consumer's Experience is not only affected by number of users in Internet resources quantity and network, and by users Used subservice kind sum purpose affects；

(2) user can be interacted by many seed, but variety classes subservice is for same user, one species Subservice all differs for the impact of different user, and Consumer's Experience is difficult to measure and compare.

(3) subservice is notch cuttype to the demand of system bandwidth.The user of the use variety classes subservice need to bandwidth Ask variant.The feature that same user pays close attention to different subservice at different time causes again it to be dynamic to the demand of system bandwidth Change.Dynamic and demand of differentiation for this type of, simply based on resource fairness dispatching method is unable to reach justice Consumer's Experience.

Z.Han et al. is at document " Delay Sensitive Scheduling Schemes for Heterogeneous QoS over Wireless Networks " in give a kind of can be delay sensitive service-user provide fair services tune Degree method.Its user proposed is satisfied with the factor (USF) tolerance as user's income (experience).But this USF is for each User's any time period only uses the Scenario Design of a subservice, is not suitable for each user and may use multiple subservices Instant message scene.N.Zhou et al. is at document " Low Complexity Cross-Layer Design with Packet Dependent Scheduling for Heterogeneous Traffic in Multiuser OFDM Systems " in carry Gone out one weighting capacity maximum (Maximum Weighted Sum Capacity, MWSC) with user is optimization aim, with number According to the dispatching method that bag is scheduler object.But the weight in the object function of the method is the most relevant with subservice time delay, it is impossible to anti- Reflect the factor size that affects on Consumer's Experience such as subservice kind, user's attention rate, use many sub-industry at instant message user During business, limited resource can be distributed to the subservice that some is little on Consumer's Experience impact by the method, thus causes user true The situation that the subservice just paid close attention to cannot be protected.

The United States Patent (USP) of Patent No. US8,315,213 discloses a kind of for using the user of multiple mobile service to provide ratio The technology of example fair services, claims and is applicable to instant messaging.It is according to subchannel quality and mobile terminal average throughput Amount distributes sub channel resource.But handling capacity can not reflect the quality of three kinds of real-time subservices of instant messaging, this technology It is not suitable for instant messaging scene, because the principal element affecting instant messaging Consumer's Experience is subservice time delay. The United States Patent (USP) of Patent No. US7,957,417 discloses one can be for mobile service user in multiple access wireless network (including instant message user) provides the technology of fair services.Its method is according to the average service condition of active user's resource and not Carry out resource and use prediction, use less user setup high priority for resource so as to get the number of users of service is most, but meeting Resource is made to use the experience of many certain customers the most very poor.

Summary of the invention

The mesh ground of the present invention is to overcome the deficiencies in the prior art, for the feature of instant messaging self, according to user The analysis of behavior characteristics, provides the fair instant messaging dispatching method of a kind of Consumer's Experience and device.

The instant messaging dispatching method that a kind of Consumer's Experience is fair, comprises the steps:

(1) according to the average delay of subservice data queue, the normalizing of each subservice that present period user runs is calculated Change quality score；

(2) according to the normalization quality score of each subservice of present period, the attention rate of each subservice of present period is calculated Increment；User's attention rate according to each subservice of previous period and the attention rate increment of each subservice of present period, calculate and work as User's attention rate of each subservice of front period；

(3) choose the highest subservice of present period user's attention rate as the representative business of user, make this represent business Normalization quality score as user representative quality of service mark；

(4) travel through all users in base station range according to step (1)～(3), obtain the representative business of each user Quality score, chooses and wherein represents the quality of service worst user of scoring as user to be serviced, with user's attention rate as probability The subservice choosing this user to be serviced services, and comments improving the normalization quality of the subservice of user to be serviced Point.

When the subservice of the described step (4) user to be serviced to choosing services, if this subservice comprises two Data stream, then service the data stream that average delay is bigger.

In described step (1), the computing formula of the normalization quality score of each subservice is as follows:

${\tilde{Q}}_i^{*}\left(n\right)=\left\{\begin{array}{cc}\max \{\mathrm{AD}+\mathrm{MS},1\},& *=V\\ 93.2-0.02{\tilde{d}}_S+0.11({\tilde{d}}_S-177.3)H({\tilde{d}}_S-177.3),& *=A\\ 4.73-\ln \left({10}^{-3}{\tilde{d}}_T\right)& *=T\end{array}\right.$

$\mathrm{AD}=3.92-3.24\×{10}^{-4}({\tilde{d}}_S+{\tilde{d}}_I)$

$\mathrm{MS}=\left\{\begin{array}{cc}-1.38\×{10}^{-3}({\tilde{d}}_S-{\tilde{d}}_I),& \mathrm{if}{\tilde{d}}_S\≥{\tilde{d}}_I\\ 1.09\×{10}^{-3}({\tilde{d}}_S-{\tilde{d}}_I),& \mathrm{if}{\tilde{d}}_S\≤{\tilde{d}}_I\end{array}\right.$

$H({\tilde{d}}_S-177.3)=\left\{\begin{array}{cc}0,& \mathrm{if}{\tilde{d}}_S-177.3<0\\ 1,& \mathrm{if}{\tilde{d}}_S-177.3>0\end{array}\right.$

Wherein,For the normalization quality score of n period user i-th subservice, { V, A, T} represent this son to * ∈ The type of business, V represents video, and A represents voice, and T represents text.For the average delay of voice data queue,For figure As the average delay of data queue,Average delay for text data queue.

Wherein, in video subservice quality score calculates, AD is the quality score decrement that absolute time delay is brought, and MS is figure As the quality score decrement brought asynchronous with audio frequency.

Normalization quality score according to each subservice of present period in described step (2), calculates present period user The weighted quality scoring of each subservice；The sequence of the weighted quality scoring according to each subservice of present period, utilizes ROC algorithm meter Calculate the attention rate increment of each subservice of present period user.

The computing formula of described weighted quality scoring is as follows:

$Q_i^{+}\left(n\right)=\left\{\begin{array}{cc}{\tilde{Q}}_i^{*}\left(n\right),& \mathrm{if}0\&le;{\tilde{Q}}_i^{*}\left(n\right)<B\\ B+{\mathrm{\&beta;}}^{*}({\tilde{Q}}_i^{*}\left(n\right)-B),& \mathrm{if\; B}\&le;{\tilde{Q}}_i^{*}\left(n\right)\&le;1\end{array}\right.$

Wherein:Weighted quality for the i-th subservice of n period user is marked,For n period user The normalization quality score of i-th subservice；* { V, A, T} represent the type of subservice, β to ∈^*Represent the phase of subservice type Close weights, 0 ＜ β^T＜ β^A＜ β^V＜ 1；B is normalization quality score thresholding, i.e. meets that user's communication requirements is acceptable minimum returns One change quality score, 0 < B < 1.

In described step (2), the computing formula of user's attention rate of each subservice is as follows:

a_i(n)=γ a_i(n-1)+(1-γ)Δa_i(n)

Wherein: a_i(n) and a_i(n-1) it is respectively n period and n-1 period user user's attention rate to i-th subservice, γ is the user's attention rate of the previous period weighing factor to user's attention rate of present period, 0≤γ≤1；Δa_iN () is n The period user attention rate increment to i-th subservice.

The instant messaging dispatching device that a kind of Consumer's Experience is fair, including:

Data queue's Delay maintenance module, for storing the information of the data queue of each subservice, calculates and exports The average delay of data queue；

Consumer's Experience estimation module, for according to described average delay, calculates and exports the representative of present period user Quality of service is marked；

Subservice data stream scheduling module, represents quality of service and marks in being used for choosing present period base station range The user of difference, is scheduling a subservice of this user with user's attention rate for probability.

Described Consumer's Experience estimation module includes:

Normalization quality score calculating sub module, for according to described average delay, calculates and to export present period each The normalization quality score of subservice；

User's attention rate safeguards submodule, for the normalization quality score according to each subservice of present period, calculates and works as The weighted quality scoring of the front each subservice of period user；And then utilize ROC algorithm to calculate the pass of each subservice of present period user Note degree increment；User's attention rate according to each subservice of previous period and the attention rate increment of each subservice of present period, meter Calculate and export user's attention rate of each subservice of present period；

User representative quality of service score calculation submodule, chooses the subservice conduct that present period user's attention rate is the highest The representative business of user, and from normalization quality score calculating sub module, read the normalization quality score of correspondence, make this generation The normalization quality score of table business is marked as the representative quality of service of this user and exports.

The beneficial effects of the present invention is: (1) present invention can exist each use of multiple instant message user in a network Family uses under the scene of one or more subservices, estimates Consumer's Experience, and it is fair to realize Consumer's Experience by data stream scheduling. (2) dispatching method of the present invention is accustomed to according to feature and the user behavior of instant messaging itself, makes scheduling by calculating Decision-making.Thus without user feedback, it is achieved complexity is low, and engineering practicability is strong.(3) in the data cached queue of user side, and on Under conditions of reporting data queue's Delay, present disclosure applies equally to the scheduling of instant messaging uplink.(4) this Bright being applicable to all has star topology, takes central authorities to control the network scenarios of scheduling.

Accompanying drawing explanation

Fig. 1 is the schematic diagram of technical solution of the present invention.

Fig. 2 is the flow chart of the instant messaging dispatching method of Consumer's Experience of the present invention justice.

Fig. 3 is the structure chart of the instant messaging dispatching device of Consumer's Experience of the present invention justice.

Fig. 4 is the Performance Simulation Results figure of the inventive method.

Detailed description of the invention

In order to more specifically describe the present invention, below in conjunction with the accompanying drawings and detailed description of the invention is to technical scheme And relative theory is described in detail.

The present invention is applicable to the scheduling of the instant messaging of cellular cell downlink.User and base station Y-connection. All instant messaging users compete current available bandwidth (remaining bandwidth after i.e. meeting by guarantee service-user).Base station is made For central controller Resources allocation, it is achieved the centralized scheduling of all instant messaging users.Any time period only one of which number Transmission opportunity is obtained according to stream.In described instant messaging three seed, video subservice comprises image stream and voice flow two Individual independent data stream, voice subservice comprises an audio data stream, and text subservice comprises a textstream.

The technical scheme is that based on the fact that all subservices of (1) instant messaging are in real time or class Real time business, owing to user's own physiological limits, user can only pay close attention to a subservice in any time period.(2) user is for i.e. Time message subservice use be generally of tendentiousness, be inclined to use order for video, voice, word.(3) user is to subservice Degree of concern change with the quality of this subservice, can increase and confront user's attention rate of measured subservice, reduce confrontation User's attention rate of the subservice of amount difference.

Technical solution of the present invention, based on the time-delay analysis to user subservice data queue, follows the tracks of user's subservice quality, Estimate user's user's attention rate to subservice, determine user representative business, by being scheduled to experience worst user with transmission Chance, it is achieved the experience of user is fair.

The principle of technical solution of the present invention is as follows: (1) each subservice is all paid close attention to by its user with certain probability, claims this general Rate is user's attention rate of this subservice, and user's attention rate sum of an all subservice of user's any time period is 1.Time initial Section, user's attention rate sequence center of gravity (Randed Order Centroid, ROC) algorithm of each subservice of a user is asked , statistical significance is all expectations meeting this ordering values.Subservice sequence needed for ROC algorithm, is according to user The tendentiousness using subservice obtains, and i.e. video is front, and voice is placed in the middle, and word is last.The user of (2) subservices is closed Note degree changes with the quality of this subservice, and when quality is good, its user's attention rate increases, and time of poor quality, its user's attention rate reduces. Current user's attention rate is that previous period user's attention rate is the highest with weighted sum (3) user's attention rate of current attention rate increment Subservice be referred to as the user representative business of this user of this period, its quality is referred to as user representative quality of service, represents this user Experiencing, quality score is the highest, experiences the best.(4) transmission opportunity is represented business matter to active user by the dispatching method of the present invention The subservice of the user (user the most to be serviced) of amount scoring minimum (i.e. Consumer's Experience is worst), so that the user of all users Experience tends to close.(5) transmission opportunity is given a subservice of user to be serviced by the dispatching method of the present invention, and this subservice is Choosing according to probability from all subservices of this user, the selected probability of subservice is paid close attention to equal to the user of this subservice Degree.Choosing subservice according to probability had both made user representative business obtain the probability maximum of transmission opportunity, avoided again because user is closed The long lasting effect that the accidental estimation difference of note degree is brought.

The instant messaging dispatching method that the Consumer's Experience of the present invention is fair, as it is shown in figure 1, comprise the steps:

A. data queue's Delay is safeguarded；

B. Consumer's Experience is estimated；

C. the scheduling that Consumer's Experience based on user representative quality of service is fair.

In described step A, base station is that each data queue safeguards its Delay.The renewal of data queue's average delay is adopted With sliding window method, in the record queue of base station head of the queue packet time extend to before the time delay of p-1 packet, and it is flat to do sums All obtain data queue's average delay.This Delay of base station maintenance, and after system sends packet each time, update this letter Breath.

In described step B, calculating by recursive fashion and obtain subservice active user's attention rate, its numerical value is to use the previous period Family attention rate and the weighted sum of current attention rate increment.The tendency that initial user attention rate uses according to user's antithetical phrase class of business Property sequence, with ROC algorithm obtain.The sequence of this tendentiousness uses the kind of subservice based on user behavior custom and user.Pay close attention to The sequence that degree increment is marked according to whole subservice weighted quality, calculates with ROC algorithm.The scoring of subservice weighted quality is sub-industry Business normalization quality score without departing from thresholding part be weighted with such subservice weight beyond thresholding part sum.Son Business normalization quality score is according to the data obtained queue average delay information in step A, in conjunction with each data queue ownership Subservice kind (video subservice comprises Liang Ge data queue), marks by the quality score method of every seed, and Quality score normalization gained is reached as high as with it.Choose the representative as this user of each user's attention rate the highest subservice Business, the normalization quality score of user representative business is as the tolerance of Consumer's Experience.

In described step C, according to gained user representative quality of service scoring in step B, choose the minimum user of scoring and make For user to be serviced, the data stream choosing a subservice with user's attention rate for probability in its current all subservice is carried out Scheduling.If selected subservice comprises 2 data streams (video subservice), then choose the data stream that wherein average delay is big It is scheduling.

The instant messaging dispatching device that a kind of Consumer's Experience is fair, including:

Data queue's Delay maintenance module, for storing the information of the data queue of each subservice, calculates and exports The average delay of data queue；

Consumer's Experience estimation module, for according to average delay, calculates and exports the representative business matter of present period user Amount scoring；

Subservice data stream scheduling module, represents quality of service and marks in being used for choosing present period base station range The user of difference, is scheduling a subservice of this user with user's attention rate for probability.

Consumer's Experience estimation module includes:

Normalization quality score calculating sub module, for according to average delay, calculates and exports each subservice of present period Normalization quality score；

User's attention rate safeguards submodule, for the normalization quality score according to each subservice of present period, calculates and works as The weighted quality scoring of the front each subservice of period user；And then utilize ROC algorithm to calculate the pass of each subservice of present period user Note degree increment；User's attention rate according to each subservice of previous period and the attention rate increment of each subservice of present period, meter Calculate and export user's attention rate of each subservice of present period；

User representative quality of service score calculation submodule, chooses the subservice conduct that present period user's attention rate is the highest The representative business of user, and from normalization quality score calculating sub module, read the normalization quality score of correspondence, make this generation The normalization quality score of table business is marked as the representative quality of service of this user and exports.

Embodiment one

As in figure 2 it is shown, the instant messaging dispatching method of Consumer's Experience of the present invention justice comprises the steps:

Step 21: initialize

The data packet count device of each data queue is made zero.By the average queuing delay record value zero setting of each data queue. Initial user attention rate is calculated according to table 1.In table 1, video traffic user is designated as user V；Speech business user is designated as user A；Depending on Frequently text user is designated as user V+jT；Speech text user is designated as user A+jT；Text user is designated as user jT.Wherein, j is for using The quantity of the text subservice that family uses.In table 1, α is user's attention rate symbol；0 represents period 0, i.e. initial time period；Subscript table Showing subservice kind, V represents video, and A represents voice, and T represents text；Subscript l represents the l subservice of user, l=1,2, 3…j.In table, each user only uses a video or voice subservice, therefore under user's attention rate of video or voice subservice It is designated as 1.

Table 1 initial user attention rate computational methods

Step 22: data queue's Delay is safeguarded

Base station is that each data queue safeguards its Delay, and has renewal after packet Successful transmissions in every subsystem Data queue's average delay

${\tilde{d}}^{*}\left(q\right)=\left\{\begin{array}{cc}\underset{w=q-p+1}{\overset{q}{\mathrm{\&Sigma;}}}{d}^{*}\left(w\right),& \mathrm{if\; q}\&GreaterEqual;p\\ \underset{w=1}{\overset{q}{\mathrm{\&Sigma;}}}{d}^{*}\left(w\right),& \mathrm{otherwise}\end{array}\right.---\left(1\right)$

Wherein:Data queue's average delay during head of the queue, d is become for q-th packet in data queue^*(w) be The wait time delay of w packet in data queue, by base station record.P is that sliding window is long, and { V, A, T} represent subservice to * ∈ Kind, represents video class subservice V, voice class subservice A and text class subservice T respectively.

Step 23: subservice normalization quality score calculates

Subservice normalization quality scoreSuch subservice can be used according to the data obtained queue average delay in step 22 Quality score method is marked, and obtains so that it reaches as high as quality score normalization.All kinds of subservice methods of marking and Reach as high as quality score as shown in table 2.

Table 2 instant messaging subservice quality score method

Wherein in video subservice quality score calculates, AD is the quality score decrement that absolute time delay is brought,MS is image and the asynchronous quality score decrement brought of audio frequency, $\mathrm{MS}=\left\{\begin{array}{cc}-1.38\&times;{10}^{-3}({\tilde{d}}_S-{\tilde{d}}_I),& \mathrm{if}{\tilde{d}}_S\&GreaterEqual;{\tilde{d}}_I\\ 1.09\&times;{10}^{-3}({\tilde{d}}_S-{\tilde{d}}_I),& \mathrm{if}{\tilde{d}}_S\&le;{\tilde{d}}_I\end{array}\right.;$ For the average delay of voice data queue, list Position is ms,For the average delay of view data queue, unit is ms.In voice subservice quality score calculates, symbol letter Number $H({\tilde{d}}_S-177.3)=\left\{\begin{array}{cc}0& \mathrm{if}{\tilde{d}}_S-177.3<0\\ 1& \mathrm{if}{\tilde{d}}_S-177.3>0\end{array}\right.,$ For the average delay of voice data queue, unit For ms.In text subservice quality score calculates,For the average delay of text data queue, unit is ms.

Step 24: the scoring of subservice weighted quality updates

Subservice weighted quality is user's subjective assessment to used subservice, contains quality and the use of subservice itself The use of this type of business is inclined to by family, the comparison between a whole subservice of user.The each sub-industry obtained according to step 23 Business normalization quality score, the scoring of present period subservice weighted quality can be calculated by formula (2).

$Q_i^{+}\left(n\right)=\left\{\begin{array}{cc}{\tilde{Q}}_i^{*}\left(n\right)& \mathrm{if}0\&le;{\tilde{Q}}_i^{*}\left(n\right)<B\\ B+{\mathrm{\&beta;}}^{*}({\tilde{Q}}_i^{*}\left(n\right)-B)& \mathrm{if\; B}\&le;{\tilde{Q}}_i^{*}\left(n\right)\&le;1\end{array}\right.---\left(2\right)$

Wherein:Weighted quality for the i-th subservice of n period user is marked,For n period user The normalization quality score of i-th subservice；* { V, A, T} represent the type of subservice, β to ∈^*Represent the phase of subservice type Close weights, 0 ＜ β^T＜ β^A＜ β^V＜ 1；B is normalization quality score thresholding, i.e. meets that user's communication requirements is acceptable minimum returns One change quality score, 0 < B < 1.

Step 25: user's attention rate updates

User's attention rate that user's attention rate is the previous period of the i-th subservice of present period user and present period The weighted sum of attention rate increment, calculate with formula (3).

a_i(n)=γ a_i(n-1)+(1-γ)Δa_i(n) (3)

Wherein: a_i(n) and a_i(n-1) it is respectively n period and n-1 period user user's attention rate to i-th subservice, γ is the user's attention rate of the n-1 period weighing factor to user's attention rate of n period, 0≤γ≤1；Δa_iN () is the n period User's attention rate increment to i-th subservice, can just can bear.

Mark according to step 24 resulting bottle business weighted quality, the attention rate increment Delta of the i-th subservice of n period user a_iN () can be calculated by formula (4).

$\mathrm{\&Delta;}{a}_i\left(n\right)=\frac{1}{\left|S_k\right|}\underset{r=m_i\left(n\right)}{\overset{\left|S_k\right|}{\mathrm{\&Sigma;}}}\frac{1}{r}---\left(4\right)$

Wherein: m_iN () is the sequence of the weighted quality scoring of the i-th subservice of n period user, | S_k| use for user Subservice quantity.

Step 26: user representative quality of service score calculation

The scoring of user representative quality of service is the normalization quality score of the highest subservice of user's attention rate.N period user User representative quality of serviceScoring can be obtained by formula (5):

$\hat{Q}\left(n\right)\stackrel{\mathrm{\&Delta;}}{=}{\tilde{Q}}_{i^{\&prime;}}\left(n\right),i^{\&prime;}=\arg \underset{1\&le;i\&le;\left|S_k\right|}{\max }{a}_i\left(n\right)---\left(5\right)$

Wherein:And a_iN () is respectively normalization quality score and the user of the i-th subservice of n period user Attention rate, i' is the sequence number of the subservice that n period user attention rate is the highest.

Step 27: subservice data stream scheduling

Mark based on step 26 gained user representative quality of service, first choose the scoring of user representative quality of service minimum User is as user k to be serviced_sel(n), as shown in Equation (6)

$k_{\mathrm{sel}}\left(n\right)=\arg \underset{1\&le;k\&le;K}{\min }\hat{Q}\left(n\right)---\left(6\right)$

Wherein, k represents user's sequence number, and in base station range, total number of users is K.

Selected user obtains subservice serial number i of transmission opportunity_selN (), is chosen by formula (7).

$\underset{i=0}{\overset{i_{\mathrm{sel}}\left(n\right)-1}{\mathrm{\&Sigma;}}}{a}_{k_{\mathrm{sel}}\left(n\right),i}\left(n\right)\&le;h<\underset{i=0}{\overset{i_{\mathrm{sel}}\left(n\right)}{\mathrm{\&Sigma;}}}{a}_{k_{\mathrm{sel}}\left(n\right),i}\left(n\right)---\left(7\right)$

Wherein: h is that algorithm randomly generates, and numerical value is between 0～1.If selected i-th subservice comprises 2 Data stream (video class subservice), then the data stream choosing wherein average delay big is scheduling.

By above step it can be seen that the average delay of record user of the present invention all subservices data queue, use low The method of complexity estimates user's attention rate of subservice, determines user representative business, thus obtains user representative quality of service Scoring is as the tolerance of Consumer's Experience.Without user feedback, Consumer's Experience i.e. can be estimated easily, and by promoting worst use The mode that family is experienced realizes experiencing between instant message user justice.

Embodiment two

The instant message dispatching method of the Consumer's Experience justice that the present invention provides can use the device shown in Fig. 3 to realize.Dress Put and include:

Data queue's Delay maintenance module, inputs the instant message data queue information for being scheduled, including each The data stream kind of queue, affiliated subservice and owning user.The case propagation delays of monitoring subservice data queue, updates sliding window Interior delay data, calculates and updates and output data queue average delay data；

Consumer's Experience estimation module, from data queue's Delay maintenance module read data queue average delay data, Estimate and export the user representative quality of service score data of current all instant message users；

Subservice data stream scheduling module, reads the user representative business of instant message user from Consumer's Experience estimation module Quality scoring data, output scheduling instructs.

Described Consumer's Experience estimation module farther includes:

Normalization quality score calculating sub module, from data queue's Delay maintenance module read data queue mean time Prolong data, calculate and export subservice normalization quality scoring data；

User's attention rate safeguards submodule, reads subservice normalization quality from normalization quality score calculating sub module and comments Divided data, updates the scoring of subservice weighted quality, updates output user's attention rate data；

User representative quality of service score calculation submodule, reads subservice from normalization quality score calculating sub module and returns One changes quality scoring data, and safeguards that submodule reads user's attention rate data from user's attention rate, calculates and updates and export i.e. Time message user user representative quality of service score data.

This device has the strongest independence, it is not necessary to user participates in, and can add to network scheduling system as standalone module System, it is not necessary to original system is made any amendment.

The beneficial effect of technical scheme of the present invention can be verified by emulation.Such as in following simple scenario: In network, number of users is 20, wherein uses video subservice user (can simultaneously use text subservice) 4 (20%), makes Term phone service-user (can simultaneously use text subservice) 6 (30%), remaining 10 user only uses with text Business.Each user uses text subservice number to be 1～5, determines at random.γ=0.7,Normalization quality score thresholding B=0.6.As shown in Figure 4, compared to traditional (Round Robin, RR) scheduling and minimum mass preferential (Minimum Quality First, MQF) scheduling in turn, 120～ In 840kbps bandwidth range, the fairness index of the present invention (MRSQF) dispatching method is the most excellent.

Obviously, those skilled in the art can carry out various change and the deformation essence without deviating from the present invention to the present invention God and scope.So, if these amendments of the present invention and deformation belong to the scope of the claims in the present invention and equivalent technologies thereof Within, then the present invention is also intended to comprise these changes and deformation.

Claims (4)

1. the instant messaging dispatching method that Consumer's Experience is fair, comprises the steps:

(1) according to the average delay of subservice data queue, calculate, according to below equation, each sub-industry that present period user runs The normalization quality score of business；

$Q_i^{*}\left(n\right)=\left\{\begin{array}{cc}max\{AD+MS,1\},& *=V\\ 93.2-0.024d_S+0.11(d_S-177.3)H(d_S-177.3),& *=A\\ 4.73-\ln \left({10}^{-3}{d}_T\right),& *=T\end{array}\right.$

AD=3.92-3.24 × 10^-4(d_S+d_I)

$MS=\left\{\begin{array}{cc}-1.38\&times;{10}^{-3}(d_S-d_I),& if{d}_S\&GreaterEqual;d_I\\ 1.09\&times;{10}^{-3}(d_S-d_I),& if{d}_S\&le;d_I\end{array}\right.$

$H(d_S-177.3)=\left\{\begin{array}{cc}0,& if{d}_S-177.3<0\\ 1,& if{d}_S-177.3>0\end{array}\right.$

Wherein,For the normalization quality score of n period user i-th subservice, { V, A, T} represent this subservice to * ∈ Type, V represents video, and A represents voice, and T represents text, d_SFor the average delay of voice data queue, d_IFor view data queue Average delay, d_TAverage delay for text data queue；

(2) according to the normalization quality score of each subservice of present period, each son of present period user is calculated by below equation The weighted quality scoring of business, and then the sequence marked according to the weighted quality of each subservice of present period, utilize ROC algorithm meter Calculate the attention rate increment of each subservice of present period；User's attention rate and present period according to each subservice of previous period are each The attention rate increment of subservice, calculates user's attention rate of each subservice of present period；

$Q_i^{+}\left(n\right)=\left\{\begin{array}{cc}{\tilde{Q}}_i^{*}\left(n\right),& if0\&le;{\tilde{Q}}_i^{*}\left(n\right)<B\\ B+{\mathrm{\&beta;}}^{*}\left({\tilde{Q}}_i^{*}\right(n)-B),& ifB\&le;{\tilde{Q}}_i^{*}\left(n\right)\&le;1\end{array}\right.$

Wherein: Q⁺ _iN () is the weighted quality scoring of the i-th subservice of n period user,I-th for n period user The normalization quality score of subservice, β^*Represent the relevant weights of subservice type；B is normalization quality score thresholding；

Described ROC algorithm is sequence centroid algorithm, and subservice is used by the sequence that this algorithm is used according to user Tendentiousness obtains, and i.e. video is front, and voice is placed in the middle, and word is last；

(3) choose present period user's attention rate the highest subservice representative business as user, make this represent returning of business One changes the quality score representative quality of service scoring as user；

(4) travel through all users in base station range according to step (1)～(3), obtain the representative quality of service of each user Scoring, chooses and wherein represents the quality of service worst user of scoring as user to be serviced, choose for probability with user's attention rate One subservice of this user to be serviced services, to improve the normalization quality score of the subservice of user to be serviced.

Instant messaging dispatching method the most according to claim 1, it is characterised in that: described step (4) is to choosing The subservice of user to be serviced when servicing, if this subservice comprises two data streams, then bigger to average delay number Service according to stream.

Instant messaging dispatching method the most according to claim 1, it is characterised in that: described step (2) Zhong Gezi The computing formula of user's attention rate of business is as follows:

a_i(n)=γ a_i(n-1)+(1-γ)Δa_i(n)

Wherein: a_i(n) and a_i(n-1) being respectively n period and n-1 period user user's attention rate to i-th subservice, γ is The user's attention rate of the previous period weighing factor to user's attention rate of present period；Δa_iN () is that n period user is to i-th The attention rate increment of individual subservice.

4. the instant messaging dispatching device that a Consumer's Experience is fair, it is characterised in that including:

Data queue's Delay maintenance module, for storing the information of the data queue of each subservice, calculates and exports data The average delay of queue；

Consumer's Experience estimation module, for according to described average delay, calculates and exports the representative business of present period user Quality score；

Subservice data stream scheduling module, represents quality of service scoring worst in choosing present period base station range User, is scheduling a subservice of this user with user's attention rate for probability；

Described Consumer's Experience estimation module includes:

Normalization quality score calculating sub module, for according to described average delay, calculates and exports each sub-industry of present period The normalization quality score of business；

User's attention rate safeguards submodule, for the normalization quality score according to each subservice of present period, when calculating current The weighted quality scoring of the section each subservice of user；And then utilize ROC algorithm to calculate the attention rate of each subservice of present period user Increment；User's attention rate according to each subservice of previous period and the attention rate increment of each subservice of present period, calculate also User's attention rate of the output each subservice of present period；

User representative quality of service score calculation submodule, chooses the highest subservice of present period user's attention rate as user Representative business, and read from normalization quality score calculating sub module correspondence normalization quality score, make this represent industry The normalization quality score of business is marked as the representative quality of service of this user and exports；

Described ROC algorithm is sequence centroid algorithm, and subservice is used by the sequence that this algorithm is used according to user Tendentiousness obtains, and i.e. video is front, and voice is placed in the middle, and word is last.