CN100469187C - A satisfaction-based multi-user scheduling method in the multi-antenna system - Google Patents

Abstract

The related multi-user schedule method in multi-antenna system is characteristics in that: the perception layer maps user business quality feature and scheduler result as user satisfaction degree weighting coefficient, the scheduler schedules according to former user satisfaction degree weighting coefficient and user channel information from physical layer, wherein the update period of user satisfaction degree weighting coefficient is more than the work period of scheduler. Compared with prior art, this invention integrates both user quality request and channel quality, applies punishing mechanism to avoid extra occupancy of some user, and belongs to better method on using system resource.

Description

In a kind of multiaerial system based on the multi-user dispatching method of satisfaction

Technical field:

The invention belongs to mobile communication multi-subscriber dispatching technical field, particularly in the multiaerial system based on the multi-user dispatching method of satisfaction.

Background technology:

Appearance and maturation along with BWA (IEEE 802.16), the super three generations's wireless communication system broadband wireless communication systems such as (FUTUREBeyond3G) of China, how more effectively to utilize hard-won bandwidth, provide abundant multimedia service for the user better, become an important research direction, and dispatching method one of them key issue just.About the quality of dispatching method the final performance of giving the user that embodies of wireless communication system, even if physical layer can provide the bandwidth of tens of megabits, if the outstanding dispatching method of neither one is reasonably distributed to the user of different demands with them, also can only be waste to bandwidth.Multi-subscriber dispatching particularly different service quality (QoS) demand user's scheduling is the problem of present moving communicating field broad research, also is the effective ways that improve wireless resource utility efficiency and improve telecommunication service quality.

According to U.S.'s " international electronics and The Institution of Electrical Engineers's radio communication periodical " (IEEE Wireless Communication, Hossam Fattah, Oct, 2002, the 76-83 page or leaf) introduces, for different application, the service quality of different business can be divided into (guaranteed) of assurance and transmit (best effort) two big classes as possible.For guaranteeing the class business, system provides the throughput that is higher than certain value on the statistical significance or is lower than the time delay of certain thresholding; For transmitting the class business as possible, system does not provide any service promise, only " does the best " and goes service.The QoS parameter of typical services comprises constant rate of speed (CBR), Real Time-Variable Bit Rate (rt-VBR), Non Real Time-Variable Bit Rate (nrt-VBR) and assigned rate (UBR) etc. not.Existing dispatching method is mostly with maximization system throughput, maximization system equitable degree, or both certain is compromise is purpose.Yet for the system of a reality, fairness or system throughput may not be best criterions.The scheduling mode of maximization system throughput is not considered the quality of services for users demand, can't provide satisfied service for the user; The mode of maximization system fairness is ignored user's channel conditions, can bring the loss of bigger system throughput, compromise with certain of maximum throughput and maximization fairness is criterion, in multiple service quality user and the system that deposits, often also can't ensure all kinds of different service quality users' service satisfaction simultaneously.

" the automobile-used technology autumn session of international electronics and The Institution of Electrical Engineers " (Vehicular TechnologyConference (Fall), Los Angeles, CA, USA, 2004,4878-4882 pages) a kind of dispatching method based on subchannel of introducing is used for random beam forming (RBF) system, and the channel fluctuation of being introduced by this random beam forming technology has increased the chance that the relatively poor user of channel quality obtains to dispatch, and is that particle is dispatched and improved resource utilization with the subchannel.But when having the user of multiple QoS requirement in the system, dispatching method wherein can't make its satisfied communication service for various users provide.

Summary of the invention:

The present invention proposes in a kind of multiaerial system the multi-user dispatching method based on satisfaction, for multiple different service quality demand user provides its satisfied service that makes, avoid CU to exceed the resource of its demand by the difference of dispatching cycle and satisfaction update cycle simultaneously, guarantee the higher system throughput, make full use of system resource.

Based on the multi-user dispatching method of satisfaction, the base station comprises physical layer block, media interviews controls (MAC) layer module, network layer module in the multiaerial system of the present invention, and wherein MAC layer module comprises scheduler and MAC layer data face mould piece; The base station assembling is no less than two transmitting antennas; User terminal is equipped with reception antenna, and its receiver carries out channel estimating and feedback of channel information is returned the base station physical layer; The base station obtains the channel information of down channel by its physical layer; The base station dispatcher utilization is dispatched from each user's that physical layer obtains descending channel information and quality of services for users parameter, obtains for the described QoS parameter of each user from application layer by network layer channel resource allocation; The MAC layer data face mould piece of base station sends the physical layer block that data are sent to the base station according to the result of scheduling; Physical layer block adopts the random beam forming technology that user's transmission data are multiplied each other with the column vector of taking out from a unitary matrix and it is sent to each antenna transmission to travelling carriage again; It is characterized in that: the base station is provided with cognitive layer module, and it obtains the customer service QoS requirement by network layer module from application layer, comprises professional maximum rate demand, minimum-rate demand and maximum delay demand; And according to customer service satisfaction weight coefficient formula, go out each user's customer service satisfaction weight coefficient by professional maximum rate demand, minimum-rate demand, maximum delay demand and from user's Mean Speed and average time-delay calculation that scheduler obtains, and leave in the customer service satisfaction weight coefficient table of cognitive layer module; The every time slot work of scheduler once, every two or more time slots upgrade a customer service satisfaction weight coefficient from cognitive layer module, and the customer service satisfaction weight coefficient table that described renewal customer service satisfaction weight coefficient is about to cognitive layer module copies in the interior customer service satisfaction weight coefficient table of scheduler; Each time slot of scheduler upgrades a subscriber channel information from physical layer, and this subscriber channel information can be user's received signal to noise ratio or subscriber channel capacity; The weight of different business is set when system initialization; Each time slot of scheduler calculates the product of its professional weight, customer service satisfaction weight coefficient and channel information respectively to all users, and selects the user of the user of this product maximum as the acquisition scheduling; After scheduler is selected to obtain the user of scheduling, recomputate this user's customer service satisfaction weight coefficient and it is deposited in the customer service satisfaction weight coefficient table of scheduler inside, scheduler is added up the Mean Speed and the average delay of each customer service and is deposited in the customer service Mean Speed and average delay table in the scheduler simultaneously;

Described customer service satisfaction weight coefficient is calculated by following formula:

w _i，k(n)＝P(1+Δ-s _i，k) (F1)

$P\left(a\right)=\left\{\begin{array}{cc}a& a\&GreaterEqual;0\\ 0& a<0\end{array}\right.---\left(F2\right)$

$s_{i,k}=\left\{\begin{array}{cc}\left(\frac{{\stackrel{\&OverBar;}{r}}_{i,k}}{r_{i,k,\mathrm{tar}}}\right)\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_{i,k}}{d_{i,k,\mathrm{tar}}}\right)}^2)& {\stackrel{\&OverBar;}{r}}_{i,k}>r_{i,k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_{i,k}<d_{i,k,\mathrm{tar}}\\ {\left(\frac{{\stackrel{\&OverBar;}{r}}_{i,k}-r_{i,k,\min }}{r_{i,k,\mathrm{tar}}-r_{i,k,\min }}\right)}^2\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_{i,k}}{d_{i,k,\mathrm{tar}}}\right)}^2)& r_{i,k,\min }\&le;{\stackrel{\&OverBar;}{r}}_{i,k}\&le;r_{i,k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_{i,k}<d_{i,k,\mathrm{tar}}\\ 0& {\stackrel{\&OverBar;}{r}}_{i,k}<r_{i,k,\min }\mathrm{or}{\stackrel{\&OverBar;}{d}}_{i,k}\&GreaterEqual;d_{i,k,\mathrm{tar}}\end{array}\right.---\left(F3\right)$

W among the formula F1 _{I, k}(n) be the customer service satisfaction weight coefficient of n time slot services i user k, Δ gets 0.01, and wherein function P (a) is suc as formula shown in the F2; S among the formula F1 _{I, k}Be k the user's of professional i customer service satisfaction, it defines by shown in the formula F3; Among the formula F3

Be the Mean Speed that this service-user obtains, r _{I, k, tar}Be the maximum rate demand of this service-user, r _{I, k, min}The 0th, the minimum-rate demand of this service-user, d _{I, k}Be the average time interval that this service-user obtains scheduling, d _{I, k}Be this service-user maximum delay demand.

Below pass through to analyze the principle of the inventive method institute foundation, and advantage of the present invention is described compared with prior art.

The user that different service quality requires, to time delay, the requirement difference of speed as assigned rate business not such as file transfer protocol (FTP) (FTP), does not require minimum speed limit, and is also not strict to the requirement of time delay; And real-time variable Rate business such as video conference is responsive to time delay, in case certain grouping exceeds delay requirement, then abandons this grouping, and simultaneously, such business also has requirement to speed, and this requirement comprises two parameters: maximum rate and minimum-rate.Usually, because the existence of video coding, minimum-rate can be lower than maximum rate, in case the service speed that should business obtains is lower than minimum-rate, then can cause tangible pause and distortion, makes the user feel not " satisfaction ".

The present invention defines customer service satisfaction s _{I, k}For:

$s_{i,k}=\left\{\begin{array}{cc}\left(\frac{{\stackrel{\&OverBar;}{r}}_{i,k}}{r_{i,k,\mathrm{tar}}}\right)\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_{i,k}}{d_{i,k,\mathrm{tar}}}\right)}^2)& {\stackrel{\&OverBar;}{r}}_{i,k}>r_{i,k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_{i,k}<d_{i,k,\mathrm{tar}}\\ {\left(\frac{{\stackrel{\&OverBar;}{r}}_{i,k}-r_{i,k,\min }}{r_{i,k,\mathrm{tar}}-r_{i,k,\min }}\right)}^2\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_{i,k}}{d_{i,k,\mathrm{tar}}}\right)}^2)& r_{i,k,\min }\&le;{\stackrel{\&OverBar;}{r}}_{i,k}\&le;r_{i,k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_{i,k}<d_{i,k,\mathrm{tar}}\\ 0& {\stackrel{\&OverBar;}{r}}_{i,k}<r_{i,k,\min }\mathrm{or}{\stackrel{\&OverBar;}{d}}_{i,k}\&GreaterEqual;d_{i,k,\mathrm{tar}}\end{array}\right.---\left(F3\right)$

s _{1, k}K user's of 〉=0 expression i kind business customer service satisfaction, when the user is dissatisfied to this business sensation, s _{I, k}=0, as user s during to professional approbation _{I, k}=1, when customer service had obtained to exceed the speed of its demand, this value showed as one and surpasses 1 positive number, and tackle it and punish this moment, promptly no longer dispatch this customer service within a certain period of time, it surpasses 1 part, and the punishment that needs of bright this user of multilist is also many more more.User's average scheduling time delay is more near its maximum delay requirement, and this user's customer service satisfaction is more little.

For assigned rate business not, if do not have minimum-rate requirement, then r _{I, k, min}=0, then formula F3 deteriorates to:

$s_{i,k}=\left\{\begin{array}{cc}0& {\stackrel{\&OverBar;}{d}}_{i,k}\&GreaterEqual;d_{i,k,\mathrm{tar}}\\ {\left(\frac{{\stackrel{\&OverBar;}{r}}_{i,k}}{r_{i,k,\mathrm{tar}}}\right)}^{\mathrm{\&Omega;}}\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_{i,k}}{d_{i,k,\mathrm{tar}}}\right)}^{\mathrm{\&Psi;}})& \mathrm{else}\end{array}\right.---\left(F4\right)$

The average satisfaction notion of employing system among the present invention, definition:

$S=\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}\frac{W_i}{K_i}\underset{k=1}{\overset{K_i}{\mathrm{\&Sigma;}}}{s}_{i,k}---\left(F5\right)$

I represents the class of business that exists in the system, W _iBe the weight of i kind business, this weight is relevant for the service class that the client provides with operator.K _iThe number of i kind service-user in the expression system, s _{I, k}I.e. k user's of i kind business professional satisfaction.

S ∈ [0,1], when all users were satisfied with in the system, S should equal 1, otherwise then was 0.The target of scheduling maximizes the average satisfaction of system exactly; unite simultaneously and consider quality of services for users demand and its channel quality; for distributing, the user makes its satisfied resource; and prevent that by penalty mechanism the specific user from taking the bandwidth that exceeds its demand; thereby realize the reasonable distribution of system resource, improve resource utilization ratio.

The application problem of multi-user dispatching method in the random beam forming system based on satisfaction summed up as follows:

$\max \left(\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}\frac{W_i}{K_i}\underset{k=1}{\overset{K_i}{\mathrm{\&Sigma;}}}{s}_{i,k}\right)---\left(F6\right)$

$s.t.\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}\underset{k=1}{\overset{K_i}{\mathrm{\&Sigma;}}}{x}_{n,i,k}=M_t,\&ForAll;n\&Element;T$

$\underset{i=1}{\overset{I}{\mathrm{\&Sigma;}}}{W}_i=1---\left(F7\right)$

N is a timeslot number in the formula (F7), and T is the set of timeslot number in the system operation time.M _tIt is the sub-fluxion of parallel emission in the random beam forming system.In the formula (F6), W _iBe the weight of i kind business, K _iBe the quantity of i kind service-user.x _{N, i, k}Be scheduling result, when k the user of n time slot services i obtains to dispatch then x _{N, i, k}=1, in the optimization aim, W _iAnd K _iIn specific system, can be constant, s by access control and priority level initializing _{I, k}It is k user's of i kind business professional satisfaction, its value is relevant with the scheduling time delay with the Mean Speed that the user obtains, simultaneously, for when satisfying the QoS of customer requirement, reduce the loss of the system throughput that the user because of dispatch service quality requirement height link performance difference causes as far as possible, when scheduler is selected the user, also should consider the channel quality that the user is current.

The cost function of user's scheduling is expressed as among the present invention:

$F_{i,k}\left(n\right)=W_{i,k}*w_{i,k}*{\hat{R}}_{i,k}\left(n\right)---\left(F8\right)$

W wherein _{I, k}K the customer service satisfaction weight coefficient of the professional i that promptly cognitive layer provides.

Be the channel capacity of physical layer by the i kind service-user k of based on feedback link acquisition.W _{I, k}Be the type of service weight of the user k of i kind business, be defaulted as 1, promptly weight is identical between different business.

In the random beam forming system, suppose that there is N the base station _tIndividual transmitting antenna, each antenna channel statistics of fading is independent.Total N user in the system, every user's reception antenna number is 1.User's received signal is $r_k\left(t\right)=\sqrt{\mathrm{\&rho;}}{H}_k\left(t\right)s\left(t\right)+n\left(t\right),t\&Element;[(n-1)T,\mathrm{nT}]$

S (t)=B (n) (1/M _t) ^1/2X (t) is the signal that the base station sends, M _tBe the parallel sub-fluxion of base station, B (n)=[b wherein ₁(n), b ₂(n) ..., b _Mt(n)], b _i(n) take from a unitary matrix, constant power emission between son stream.H _k(n)=[h _{1, k}(n), h _{2, k}(n), ..., h _{N, k}(n)], be channel matrix, h wherein _{I, k}(n) be the channel gain of i the antenna in base station to k mobile station receiver, establish it for independent identically distributed multiple Gaussian random variable, be distributed as CN (0,1), n (n) is a white complex gaussian noise, is distributed as CN (0,1).Suppose that passage remains unchanged in a time slot.If E{s (t) ^HS (t) }=1 antenna reception place average signal-to-noise ratio just in time is ρ.After the vectorial weighting through beam forming, the equivalent received signal to noise ratio is for this reason:

${\hat{r}}_{i,k}=\mathrm{\&rho;}\&CenterDot;{\left|\right|{\tilde{h}}_{i,k}\left(n\right)\left|\right|}^2=\mathrm{\&rho;}\&CenterDot;{\left|\right|H_k\left(n\right)b_i\left(n\right)\left|\right|}^2---\left(F9\right)$

Then this receiver user maximum rate that can obtain is

${\hat{R}}_{i,k}\left(t\right)={\log }_2(1+{\hat{r}}_{i,k})---\left(F10\right)$

The customer service satisfaction weight coefficient w of i kind service-user k among the formula F8 _{I, k}For

w _i，k(n)＝P(1+Δ-s _i，k) (F11)

$P\left(a\right)=\left\{\begin{array}{cc}a& a\&GreaterEqual;0\\ 0& a<0\end{array}\right.---\left(F12\right)$

s _{I, k}It is the customer service satisfaction of i kind service-user k.Punishment thresholding when △ is the weight renewal when the customer service satisfaction exceeds this 1+ △, thinks that the user has obtained to exceed the service of its demand, thereby this user's weight is changed to zero in this dispatching cycle, thereby it is punished.s _{I, k}In

Be the Mean Speed of the business of i kind service-user k, adopt the low circulation method of index of weighting to calculate at this through the scheduling acquisition:

${\stackrel{\&OverBar;}{r}}_{i,k}(n+1)=(1-\frac{1}{t_c}){\stackrel{\&OverBar;}{r}}_{i,k}\left(n\right)+\frac{1}{t_c}{R}_{i,k}^{\&prime;}\left(n\right)---\left(F13\right)$

Be the speed of the user k scheduling acquisition of n time slot services i.t _cBe the time window of scheduler, be constant.

In formula (F3), the customer service satisfaction is described can be divided into two factors, and first factor is the speed satisfaction factor, and second is the time delay satisfaction factor.The customer service satisfaction is low more, then this user's w _{I, k}(t) big more.

The dispatching criterion of scheduler is:

${(i,k)}^{*}=\underset{i,k}{\arg \max }\left(F_{i,k}\left(n\right)\right)---\left(F14\right)$

Select the user of current cost function maximum during promptly each the scheduling.

During scheduler work, user's time delay factor, Mean Speed, targeted rate and current channel capacity all will influence the result of scheduling.After the user has obtained to exceed the scheduling of resource of its demand, be that its customer service satisfaction exceeds at 1 o'clock, P among the formula F12 (a) is that penalty can be 0, then before next customer service satisfaction weight coefficient upgrades, this user can not obtain scheduling, this is a kind of punishment to the resource user that obtains to exceed demand, thereby has avoided too much resource to be occupied for a long time by the user of some priority level height (channel good or delay requirement strictness etc.).Therefore the update cycle of customer service satisfaction weight coefficient is slow than the scheduler work period.

In the process of scheduler schedules work, comprise its channel capacity in user's the cost function, promptly when two users' customer service satisfaction weight coefficient was close, the user that channel is good can obtain scheduling.Simultaneously because the high long-term occupying system resources of user of minority QoS requirement has been avoided in the existence of punitive measures, thus can be in maximization system satisfaction during scheduling the loss of minimizing system throughput.When the real-time variable Rate customer service satisfaction of statistics, the speed that obtains as the user is lower than the minimum speed that guarantees and judges that promptly this user is dissatisfied.Increasing methods such as maximization system throughput scheduling, direct ratio equity dispatching, minimax equity dispatching and can't guarantee the demand of user along with number of users to time delay, when the user increases to a certain degree the time, the variable Rate service-user can be because the scheduling time delay exceed its tolerance boundary and cause the user dissatisfied in real time.Multi-user dispatching method based on the customer service satisfaction is then different, when the resource of user's request in the system has exceeded the total resources of system, variable Rate is professional in real time can be because the sensitivity of its time delay makes satisfaction descend, user's cost function value formula this moment (F8) raises, and makes the easier acquisition scheduling of delay sensitive user; Simultaneously, because the existence of Penalty time, can not make the whole system resource of real-time variable Rate CU again, thereby guaranteed service to the insensitive not assigned rate of time delay service-user, therefore from system's satisfaction, the multi-user dispatching method that the present invention is based on satisfaction is better than the maximization system and gulps down methods such as rate scheduling, direct ratio equity dispatching, the scheduling of minimax equitable degree.

Guarantee that if desired real-time variable Rate business is in number of users service quality more for a long time, and give real-time variable Rate business revenue outer weight, the multi-user dispatching method that the present invention is based on satisfaction so will be partial to select unsatisfied real-time variable Rate service-user when scheduling, thereby will further guarantee these quality of services for users.By combining with penalty mechanism of the present invention, can avoid real-time variable Rate service-user to seize all system resources again, guarantee that other service-users obtain service.

Based on the multi-user dispatching method of satisfaction, the dispatching criterion owing to adopting based on satisfaction can ensure quality of services for users in the multiaerial system of the present invention; By adopting the strategy different dispatching cycle, realize having avoided the part high-priority users to take multi-system resource to obtaining to exceed demand user's punishment with the satisfaction update cycle; By adopting the strategy that comprises the subscriber channel capacity in the scheduling cost function, to the consideration of the availability of frequency spectrum, improved system throughput when having realized scheduling.By the embodiments of the invention enumerated below as can be seen, the present invention has lost less throughput than the dispatching method of maximization system throughput, but significantly better than the direct ratio equity dispatching method, simultaneously aspect system's satisfaction, significantly better than additive method.

Description of drawings:

Fig. 1 is the scheduler architecture schematic diagram among the present invention.

Fig. 2 is dispatching cycle and the customer service satisfaction weight coefficient update cycle graph of a relation among the present invention.

Fig. 3 is descending multi-user system user number and the throughput curve that adopts satisfaction scheduling and random beam forming.

Fig. 4 is descending multi-user system user number and the system's satisfaction curve that adopts satisfaction scheduling and random beam forming.

Fig. 5 is the descending multi-user system of scheduling of employing satisfaction and random beam forming adds right scheduling to real-time variable Rate business system user number and a throughput curve.

Fig. 6 is the descending multi-user system of scheduling of employing satisfaction and random beam forming adds right scheduling to real-time variable Rate business system user number and a satisfaction curve.

Embodiment:

Embodiment below in conjunction with description of drawings this method.

Embodiment 1:

Present embodiment is an example with Chinese Future plan B3G FDD system framework, and the base station comprises physical layer block, medium Access Layer module, network layer module, and wherein the medium Access Layer comprises scheduler and medium Access Layer data surface module and is provided with 8 transmit antennas; User terminal is equipped with reception antenna, and comprise physical layer block, medium Access Layer module, network layer module, application layer module, its physical layer comprises receiver and sending module, and is equipped with 1 reception antenna, and receiver carries out channel estimating and feedback of channel information is returned the base station physical layer; Channel is non-frequency selectivity (establishing through technology planarizations such as the OFDMs) fading channel in a broadband, and bandwidth is 20MHz; Each user has only a kind of business, and this business is file transfer protocol (FTP) (FTP) business or video conference business; The maximum bandwidth of ftp business demand is 10Mbps, maximum delay is 1s, the real time flow medium business is an example with now typical video conference, the maximum bandwidth demand is 2Mbps, and minimum-rate is 1Mbps, and maximum delay is 50ms, the actual bandwidth demand of business is mapped as the throughput of service needed, then the ftp business throughput need be 10/20=0.5 (bps/Hz), and variable Rate business objective speed is 2/20=0.1 (bps/Hz) in real time, and minimum-rate is mapped as 1/20=0.05 (bps/Hz).The frame length 8.375ms of system, every frame contain 10 time slots, promptly every slot length 0.8375ms, and then the maximum of ftp business scheduling time delay is mapped as 1200 time slots, and the maximum of streaming media service scheduling time delay is mapped as 60 time slots; All customer service weights W _k=1, promptly different user operational authority heavy phase together; The every time slot work of scheduler once, work period T _sBe 1 time slot 0.8375ms, the every frame update of customer service satisfaction weight coefficient once, i.e. customer service satisfaction update cycle T _wBe 10 time slot 8.375ms.

Fig. 1 has provided the scheduler architecture schematic diagram among the present invention:

Application layer 1 mails to user service data 3 network layer module 4 of base station 2; Cognitive layer module 6 obtained customer service quality of service characteristics 5 from network layer module 4, obtains user's statistical information 11 of scheduler from Scheduler module 9, and calculates customer service satisfaction weight coefficient 7; Scheduler module 9 in media interviews controls (MAC) layer 20 is dispatched according to the customer service satisfaction weight coefficient 7 that obtains from cognitive layer module 6 with from the subscriber channel information 19 that physical layer block 14 obtains, and scheduling result 10 is exported to MAC layer data face mould piece 12; MAC layer data face mould piece 12 will be sent to physical layer block 14 by the MAC layer data 13 that network layer data 8 generates according to scheduling result 10; Physical layer block 14 sends signal by antenna for base station 15, and receiver user 17 receives base station signal by user antenna 16, and channel feedback information 18 is sent to the physical layer block 14 of base station 2.

The down link workflow is as follows:

The first step: application layer module 1 mails to user service data 3 network layer module 4 of base station 2.

Second step: cognitive layer module 6 obtained the quality of service characteristics information 5 of customer service from application layer module 1 by network layer module 1.The maximum rate r that comprises the user _{K, tar}, minimum-rate r _{K, min}, maximum delay d _{K, tar}

The 3rd step: cognitive floor module 6 number divided by the customer service satisfaction weight coefficient update cycle, if remainder is 0, then recomputates all customer service satisfaction weight coefficients with current time slots, and cognitive layer module 6 comprises from Scheduler module acquisition user statistical information 11

With

, according to:

Customer service satisfaction formula

$s_k=\left\{\begin{array}{cc}\left(\frac{\stackrel{\&OverBar;}{r}}{r_{k,\mathrm{tar}}}\right)\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_k}{d_k}\right)}^2)& {\stackrel{\&OverBar;}{r}}_k\&GreaterEqual;r_{k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_k\&le;d_{k,\mathrm{tar}}\\ {\left(\frac{\stackrel{\&OverBar;}{r}-r_{k,\min }}{r_{k,\mathrm{tar}}-r_{k,\min }}\right)}^2\&times;(1-{\left(\frac{{\stackrel{\&OverBar;}{d}}_k}{d_k}\right)}^2)& r_{k,\min }\&le;{\stackrel{\&OverBar;}{r}}_k\&le;r_{k,\mathrm{tar}}\mathrm{and}{\stackrel{\&OverBar;}{d}}_k<d_{k,\mathrm{tar}}\\ 0& {\stackrel{\&OverBar;}{r}}_k<r_{k,\min }\mathrm{or}{\stackrel{\&OverBar;}{d}}_k>d_{k,\mathrm{tar}}\end{array}\right.---\left(F3\right)$

Calculate customer service satisfaction s _k, pass through w again _k(n)=P (1+ △-s _k) calculate customer service satisfaction weight coefficient, and store in the customer service satisfaction weight coefficient table.△=0.01 wherein, P (a) is:

$P\left(a\right)=\left\{\begin{array}{cc}a& a\&GreaterEqual;0\\ 0& a<0\end{array}\right.$

Formula (F2)

The 4th step: Scheduler module 9 with current time slots number divided by the customer service satisfaction weight coefficient update cycle, if remainder is 0, then from the customer service satisfaction weight coefficient table of cognitive layer module 6, duplicate each customer service satisfaction weight coefficient in the customer service satisfaction weight coefficient table of scheduler 9 inside.

The 5th step: the customer service satisfaction weight coefficient w that Scheduler module 9 provides by cognitive layer module 6 _k(n) and the physical layer block 14 subscriber channel information of reporting 19 dispatch.By $F\left(n\right)=W_k*w_k*{\hat{R}}_k\left(n\right)$ Generate user's cost function value.Scheduler is selected the user of this acquisition scheduling of conduct of cost function value maximum among all users, scheduling result 10 is sent to MAC layer data face mould piece 12, upgrade the professional satisfaction weight coefficient of acquisition dispatched users immediately and store in the customer service satisfaction weight coefficient table of scheduler, add up all users' average delay simultaneously And Mean Speed

The 6th step: MAC layer data face mould piece 12 is according to scheduling result 10, and the network layer data 8 that network layer module 4 is sent here converts MAC layer data 13 to and is sent to physical layer block 14.

The 7th step: physical layer block 14 is passed through antenna for base station 15 with the MAC layer data by many, adopts the random beam forming technology to send.The transmission signal is s (t)=b (t) (1/M _t) ^1/2X (t), t ∈ [(n-1) T, nT] wherein x (n) is the signal that the user sends, b (n) takes from a unitary matrix, i.e. beam shaping matrix, M _tSend sub-fluxion for random beam forming is parallel, present embodiment is 1.

The 8th step: receiver user module 17 receives base station signal by user antenna module 16, receives data and is $r_k\left(t\right)=\sqrt{\mathrm{\&rho;}}{H}_k\left(t\right)s\left(t\right)+n_k\left(t\right),t\&Element;[(n-1)T,\mathrm{nT}],$ H wherein _k(t)=[h _{1, k}(t), h _{2, k}(t), ..., h _{N, k}(t)] be channel matrix, h _{I, k}(t) be the channel gain of i the antenna in base station, n to k receiver user _k(t) be the additivity white complex gaussian noise; Receiver user calculates through the equivalent received signal to noise ratio behind the beam forming vector weighting: ${\hat{r}}_k\left(t\right)=\mathrm{\&rho;}\&CenterDot;{\left|\right|{\tilde{h}}_k\left(t\right)\left|\right|}^2=\mathrm{\&rho;}\&CenterDot;{\left|\right|H_k\left(t\right)b\left(n\right)\left|\right|}^2,$ And the equivalent received signal to noise ratio is fed back to the physical layer block 14 of base station 2 as channel feedback information 18.

The 9th step: base station physical layer module 14 used formula ${\hat{R}}_k\left(n\right)={\log }_2(1+{\hat{r}}_k\left(n\right))$ The channel capacity that calculates the user offers Scheduler module 9 as subscriber channel information 19.

The purpose that cognitive layer exists is that the scheduling result with the QoS parameter of application layer and scheduler is mapped as customer service satisfaction weight coefficient and offers scheduler, and the work complexity of simplifying scheduler guarantees that simultaneously the user obtains higher satisfaction.

Fig. 2 has provided dispatching cycle and customer service satisfaction weight coefficient update cycle graph of a relation.As shown in Figure 2: the unidirectional arrow of time shaft top is customer service satisfaction weight coefficient updated time w, and the unidirectional arrow of time shaft below is scheduler work s constantly, the customer service satisfaction weight coefficient update cycle T of cognitive layer _wWorking time slot period T greater than scheduler _s, and T _wBe T _sIntegral multiple (in the present embodiment being 10 times), its objective is that the difference of utilizing two cycles punishes the user who obtains to exceed QoS requirement, after the each scheduling of scheduler is finished, upgrade the customer service satisfaction weight coefficient w that obtains scheduling immediately _k,, upgrade this satisfaction does not obtain the user of scheduling in the update cycle satisfaction weight coefficient by cognitive layer entity at customer service satisfaction weight coefficient updated time w; Because the satisfaction weight coefficient update cycle is greater than the scheduler schedules cycle, when the user at a T _wMiddle satisfaction s _kDuring 〉=1+ △, then at T _wIn the several time slots of residue in this user scheduling of will not reentrying, realized punishment thus, thereby prevented the long-term occupying system resources of a few users the user who obtains to exceed the satisfaction thresholding, improved resource utilization ratio.

It is the FTP user that Fig. 3, Fig. 4 have provided in the present embodiment system 75% user, and 25% is real-time variable Rate service-user, when two kinds of business have equal weight, increases the situation of change of system throughput and system's satisfaction with number of users.Employing has certain reduction based on the more maximum Carrier interference ratio of system throughput rate curve c (MAX-CI) the dispatching method curve d of the multi-user dispatching method of satisfaction among Fig. 3, but still a higher level, at number of users more for a long time apparently higher than direct ratio justice (PF) dispatching method curve b, compared with minimax justice (MMF) the dispatching method curve a of maximization equitable degree, then improve a lot simultaneously.

See system's satisfaction again, at number of users seldom the time, the resource in the system is enough abundant, so the result of four kinds of method scheduling can both satisfy the quality of services for users demand among Fig. 4.Along with number of users increases, the resource of user's request is system's resource that can provide head and shoulders above, and each side's genealogy of law system satisfaction is all reducing.When the user reaches 80 in system, except a significantly turnover is all arranged based on three kinds of methods the multi-user dispatching method curve e of satisfaction (direct ratio justice method curve f, max carrier to interference method curve g, minimax justice method curve h), then reach at 120 o'clock and adopt significantly based on the multi-user dispatching method of satisfaction and descend at number of users.Therefore the multi-user dispatching method that the present invention is based on satisfaction can provide satisfied service for more users when guaranteeing service quality.

Guarantee real-time variable Rate business if desired in number of users service quality more for a long time, then need to the outer weight of real-time variable Rate business revenue, for this reason with the W in the formula (F8) _kGive different assignment, get the W of ftp business _k=1, the W of real-time variable Rate business _kBe respectively 1.1,1.2.

The descending multi-user system that adopts satisfaction scheduling and random beam forming adds the system user number of right scheduling and throughput curve as shown in Figure 5 to real-time variable Rate business.Wherein curve i is unweighted multi-user dispatching method throughput based on satisfaction, curve j is the throughput to the professional weighting 1.1 of real-time variable Rate, curve k is the throughput to the professional weighting 1.2 of real-time variable Rate, as seen, after having increased extra weight for the user of real-time variable Rate business, cause the throughput of system that certain decline is arranged.This is because of the increase along with the user, a large amount of user satisfaction is very low, owing to given extra weight to real-time variable Rate business, then scheduler is when selecting the user, be partial to select unsatisfied real-time variable Rate user, thereby dwindled the selectable user set, for the such system of random beam forming, this just means reducing of multi-user diversity gain, and brings the reduction of system throughput.Along with the increase of number of users in the system, in real time the variable Rate number of users is also increasing, and promptly multi-user diversity gain increases, thus at number of users near 200 o'clock, the scheduling mode system throughput of two kinds of weightings all has increase.

Fig. 6 is the descending multi-user system of scheduling of employing satisfaction and random beam forming adds right scheduling to real-time variable Rate business system user number and a satisfaction curve.Wherein curve m is unweighted multi-user dispatching method system satisfaction based on satisfaction, and curve n is the system's satisfaction to the professional weighting 1.1 of real-time variable Rate, curve.It is system's satisfaction to the professional weighting 1.2 of real-time variable Rate, because the variable Rate business has obtained extra weight in real time, so at number of users more for a long time, unsatisfied real-time variable Rate user has obtained more dispatcher meeting, and having avoided real-time variable Rate user is 0 because of the satisfaction that the scheduling time delay causes.

Along with the increase of number of users, the variable Rate business can obtain to make its satisfied Mean Speed and scheduling time delay all the time in real time, thereby has guaranteed that system's satisfaction is a higher level.Make system's satisfaction unexpected reduction can not occur to real-time variable Rate user's weighting, but along with the increase of number of users descends slowly.Add different weights for real-time variable Rate business and can make it obtain different dispatching priorities, weight is high more, and the service of acquisition is guaranteed that more the influence to ftp business is also big more simultaneously.

This shows, employing is to the multi-user dispatching method based on satisfaction of the professional weighting of real-time variable Rate, service quality that can the real-time variable Rate business of better guarantee, this method can be generalized to the scheduling that some special user is provided weighting, guarantee its communication quality, have Practical significance for system operator.

Claims (1)

1. In the multiaerial system based on the multi-user dispatching method of satisfaction, the base station comprises physical layer block, MAC layer module, network layer module, wherein the MAC layer module comprises scheduler and MAC layer data surface module; The base station assembling is no less than two transmitting antennas; User terminal is equipped with reception antenna, and its receiver carries out channel estimating and feedback of channel information is returned the base station physical layer; The base station obtains the channel information of down channel by its physical layer; The base station dispatcher utilization is dispatched from each user's that physical layer obtains descending channel information and quality of services for users parameter, gives each user with channel resource allocation, and described QoS parameter obtains from application layer by network layer; The MAC layer data surface module of base station sends the physical layer block that data are sent to the base station according to the result of scheduling; Physical layer block adopts the random beam forming technology that user's transmission data are multiplied each other with the column vector of taking out from a unitary matrix and it is sent to each antenna transmission to travelling carriage again; It is characterized in that: the base station is provided with cognitive layer module, and it obtains the customer service QoS requirement by network layer module from application layer, comprises professional maximum rate demand, minimum-rate demand and maximum delay demand; And according to customer service satisfaction weight coefficient formula, go out each user's customer service satisfaction weight coefficient by professional maximum rate demand, minimum-rate demand, maximum delay demand and from user's Mean Speed and average time-delay calculation that scheduler obtains, and leave in the customer service satisfaction weight coefficient table of cognitive layer module; The every time slot of scheduler carries out user's scheduling, per two above time slots upgrade a customer service satisfaction weight coefficient from cognitive layer module, and the customer service satisfaction weight coefficient table that described renewal customer service satisfaction weight coefficient is about to cognitive layer module copies in the interior customer service satisfaction weight coefficient table of scheduler; Each time slot of scheduler upgrades a subscriber channel information from physical layer, and this subscriber channel information can be user's received signal to noise ratio or subscriber channel capacity or channel quality indication; The weight of different business is set when system initialization; Each time slot of scheduler calculates the product of its professional weight, customer service satisfaction weight coefficient and channel information respectively to all users, and selects the user of the user of this product maximum as the acquisition scheduling; After scheduler is selected to obtain the user of scheduling, recomputate this user's customer service satisfaction weight coefficient and it is deposited in the customer service satisfaction weight coefficient table of scheduler inside, scheduler is added up the Mean Speed and the average delay of each customer service simultaneously, and deposits in scheduler interior the customer service Mean Speed and average delay table;

   Described customer service satisfaction weight coefficient is calculated by following formula:

   w _{I, k}(n)=P (1+ Δ-s _{I, k}) formula F1

   $P\left(a\right)=\left\{\begin{array}{cc}a& a\&GreaterEqual;0\\ 0& a<0\end{array}\right.$ Formula F2

   

   Formula F3

   W among the formula F1 _{I, k}(n) be the customer service satisfaction weight coefficient of n time slot services i user k, Δ gets 0.01, and wherein function P (a) is suc as formula shown in the F2; S among the formula F1 _{I, k}Be k the user's of professional i customer service satisfaction, it defines by shown in the formula F3; R among the formula F3 _{I, k}Be the Mean Speed that this service-user obtains, r _{I, k, tar}Be the maximum rate demand of this service-user, r _{I, k, min}The 0th, the minimum-rate demand of this service-user, d _{I, k}Be the average time interval that this service-user obtains scheduling, d _{I, k, tar}Be this service-user maximum delay demand.

Images