CN101414967B - Admission control and resource allocation in a communication system supporting quality of service - Google Patents

Abstract

Provided is a method and an apparatus for admission control in a communication system. An Access Network (AN) element determines available resources. When available resources are sufficient to support the requirements of a requested application flow, the AN admits the application flow. The AN periodically, and on trigger events, updates a measure of available resources. The admission control may operate in coordination with a scheduler applying a compensation factor to each flow type, and a compensation factor for aggregate flows of a given user.

Description

Permission in the communication system of support service quality gets into control and resource allocation

The application is to be application number the dividing an application for the one Chinese patent application of " supporting the permission in the communication system of service quality to get into control and resource allocation " that be No. 200480010998.8 denomination of invention on March 17th, 2004 applying date.

Priority request according to 35U.S.C. § 119

Present patent application requires to transfer assignee of the present invention's the temporary patent application of submitting on March 17th, 2003 the 60/455th that is entitled as " System for Allocating Resources in a Communication System "; No. 906 priority, and therefore incorporate into for referencial use here.

Background

1. Invention field:

The present invention relates to communication system.Especially, these embodiment are to allocate communications resource between a plurality of users of a communication system.

2. Relevant technologies:

Provide plurality of proposals and solved the problem of between a plurality of users, the limited communication resource that individual node provided in the communication system being distributed.A target of this type systematic is at each node place enough resources to be provided, with the demand that satisfies all users and simultaneously with cost minimization.Thus, normally with between each user effectively the target of Resources allocation design this type systematic.

Various systems have realized frequency division multiple access (FDMA) scheme, this scheme with resource concurrence distribute to each user.Communication node in this type systematic is put at any time usually that each user in network sends information or all is limited from the bandwidth that each user receives information.This scheme is usually directed to the different piece of total bandwidth is distributed to unique user.Although for customer requirements and communication node the interference-free system that communicates by letter is arranged, this scheme possibly be effectively, and is lasting when this type of, when interference-free communication is also nonessential, can realize the total bandwidth better utilization.

Other distributes the scheme of the communication resource of single communication node to comprise time division multiple access (TDMA) scheme between a plurality of users.Do not require with single communication node have under lasting, the interference-free situation of communicating by letter the user, these TDMA schemes are effective especially when between a plurality of users, distributing the finite bandwidth resource of single communication node.The TDMA scheme makes whole bandwidth of single communication node specialize in each user's use usually by designated time intervals.In the wireless communication system of employing code division multiple access (CDMA) scheme, this can distribute to each user by designated time intervals with all Code Channel and realize through on the time division multiplexing basis.Communication node realizes that the unique carrier frequency or the channel code that are associated with the user carry out proprietary communication with permission with this user.The TDMA scheme also can use physics contact relaying exchange or packet switching in land line systems, to realize.

Tdma system is distributed to each user with the time interval that equates usually in a looping fashion.This possibly cause the insufficient utilization of certain user to some time interval.Similarly, other user has the demand to the communication resource above the time interval distribute, and makes these users not to be met.The Systems Operator can select to bear increase node bandwidth cost to guarantee the not having user to be not being met, the user who perhaps allows to be not being met continues to keep the state that is not being met.

Thus, need provide a kind of according to a kind of between the user network strategy of allocate communications resource, between the user of communication network effectively with the system and method for allocate communications resource liberally.Consistent with it is; Need the number of users maximization that system served be included, but are not limited to; Various specific requirements in response to system, restriction and/or target are provided, on the basis of every flow process and/or the mechanism of on the basis that adds up to, carrying out resource allocation.Need the optimized permission of resource allocation be got into control and preemption method in addition.

The accompanying drawing summary

Fig. 1 illustrates a kind of communication network according to an embodiment of the invention.

Fig. 2 A illustrates the base station controller of configuration according to one embodiment of present invention and the block diagram of base station equipment.

Fig. 2 B illustrates the block diagram of configured remote stations equipment according to one embodiment of present invention.

Fig. 3 is illustrated in the flow chart of carrying out a kind of dispatching algorithm among the embodiment of the channel scheduler shown in Fig. 2 A.

Fig. 4 is a kind of communication system of supporting multimedia application, and wherein each interapplication communications is represented by an application flows.

Fig. 5 is an application flows formation.

Fig. 6 is a sequential chart, describes the signal timing of the part of an application flows.

Fig. 7 A is a sequential chart, describes the jitter measurement to an application flows.

Fig. 7 B is a sequential chart, is described as handling an application flows, during some time slots, sends continuous IP and divides into groups.

Fig. 8 is a flow chart, is described in a kind of communication system some application flows are dispatched.

Fig. 9 is a flow chart, describes some application flows with different service quality (QoS) requirement are dispatched.

Figure 10 is a system assumption diagram according to an embodiment, and the definition of application flows each application flows consistent with a kind of dispatching algorithm is shown.

Figure 11 is the table according to the type of an embodiment discriminator.

Figure 12 A describes the part according to a kind of dispatching algorithm of an embodiment, comprises the initialization of application programs flow process.

Figure 12 B describes the part according to a kind of dispatching algorithm of an embodiment, comprises the processing according to the type application programs flow process of class.

Figure 12 C describes the part according to a kind of dispatching algorithm of an embodiment, comprises processing to the application flows of pattern I, to the processing of the application flows of pattern II with to the processing of the application flows of pattern III.

Figure 12 D describes the part according to a kind of dispatching algorithm of an embodiment, comprises the processing to the application flows of pattern I.

Figure 12 E describes the part according to a kind of dispatching algorithm of an embodiment, comprises adaptive weighted and with its scheduling that is the basis.

Figure 13 illustrates a kind of base transceiver systems (BTS), is used for realizing that at wireless communication system the adaptive weighted algorithm of a kind of usefulness comes the algorithm of scheduling application flow process.

Figure 14 is a sequential chart, and mapping is such as data transfer rate (L _MAX), reservation of resource (Res (t)) and available resources maximum resources such as (Avail (t)) be as the function of time.

Figure 15 is a sequential chart, and the request of data that the user of mapping from the High Rate Packet Data type system receives and the computational estimation competence L (t) that will keep at time t are as the function of time.

Figure 16 is an information flow chart, describes to be used to support a plurality of schedulers with High Rate Packet Data type system of the application flows that service quality (QoS) requires, and wherein these flow processs are that the compensation of using every flow process is dispatched.

Figure 17 is an information flow chart, describes to be used to support a plurality of schedulers with High Rate Packet Data type system of the application flows that service quality (QoS) requires, and wherein these flow processs are used and added up to compensation to dispatch.

Figure 18 A illustrates a kind of algorithm that in supporting a plurality of High Rate Packet Data type systems with application flows that service quality (QoS) requires, allows to get into control to 18E.

Figure 19 illustrates and is a kind ofly supporting a plurality of algorithms that carry out preemption in the High Rate Packet Data type system of the application flows that service quality (QoS) requires that have.

Figure 20 is a block diagram of supporting a kind of access network (AN) in a plurality of High Rate Packet Data type systems with application flows that service quality (QoS) requires.

Specifically describe

Each embodiment of the present invention is directed to the system and the equipment of Resources allocation between a kind of a plurality of users that in communication network, served by single communication node.At independent discrete transmission intercal, or claim that " service intervals ", unique user take the limited resources of communication node with repelling all other users.Based on weights that are associated with unique user or score, unique user is selected to take this limited resources.Preferably, the change to the weights that are associated with unique user is based on the momentary rate that this unique user can consume these limited resources.

With reference to accompanying drawing, Fig. 1 representes a kind of communication system of exemplary variable bit rate.This type systematic is transferring the U.S. Patent application of submitting on November 3rd, 1997 the 08/963rd that is entitled as " Method and Apparatusfor High Rate Packet Data Transmission " (method and apparatus of High Rate Packet Data transmission) of Qualcomm Co., Ltd; Describe in No. 386 the United States Patent (USP) of on June 3rd, 2003 bulletin (are No. 6574211) at present, and incorporate into as a reference here.Variable rate communication system comprises a plurality of unit 2A-2G.Each unit 2 is by base station 4 services of a correspondence.Each distant station distributes and spreads all over this communication system.In this exemplary embodiment, each distant station 6 is all communicated by letter with base station 4 at the most with transfer of data arbitrarily at interval on the forward link.For example, at time slot n on the forward link, base station 4A is exclusively to distant station 6A transmission data, and base station 4B is exclusively to distant station 6B transmission data, and base station 4C sends data to distant station 6C exclusively.As shown in Figure 1, preferably each base station 4 is transmitted data at any given time to a distant station 6.In other embodiments, can communicate by letter with an above distant station 6 in particular data transmission at interval in base station 4, repels all other distant stations 6 that are associated with base station 4 simultaneously.In addition, data transfer rate is variable, and in one embodiment, data transfer rate is to depend on Carrier interference ratio (C/I) that receiving remote station 6 is measured and the energy per bit noise ratio (E that requires _b/ N ₀).For for simplicity, among Fig. 1 not shown from distant station 6 to the base station 4 back to link.According to an embodiment, distant station 6 is the mobile units that have by the wireless transceiver of wireless data service user operation.

The block diagram of basic subsystem that a kind of exemplary variable rate communication system is shown is shown in Fig. 2 A-2B.All base stations 4 (for for simplicity, a base station 4 only being shown among Fig. 2) in base station controller 10 and packet network interface 24, PSTN (PSTN) 30 and the communication system join.Communicating by letter between the distant station 6 in the base station controller 10 coordinating communication networks and other user who is connected to packet network interface 24 and PSTN30.PSTN30 joins with the user through standard telephone network (not shown among Fig. 2).

Base station controller 10 comprises many selector elements 14, although for for simplicity, only illustrate one among Fig. 2 A.Each selector element 14 all is assigned to control the communication between one or more base stations 4 and the distant station 6.If selector element 14 is not assigned to distant station 6, then notification call processor controls 16 has the needs of paging distant station 6.Call control processor 16 assigns base station 4 to come paging distant station 6 immediately.

Data source 20 comprises the mass data that will be transferred to distant station 6.Data source 20 offers packet network interface 24 with data.Packet network interface 24 receives data, and data are sent to selector element 14.Selector element 14 arrives each base station 4 of communicating by letter with distant station 6 with transfer of data.In this exemplary embodiment, a data formation 40 is all safeguarded in each base station 4, and its storage will be transferred to the data of distant station 6.

Data are that the form with packet is transferred to element in channel 42 from data queue 40.In this exemplary embodiment, on the forward link, " packet " refers to a certain amount of data, and it is at most 1024, and will within one " time slot " (such as 1.667 milliseconds of ≈), be transferred to target remote station 6.For each packet, element in channel 42 is inserted essential control field.In this exemplary embodiment, element in channel 42 is carried out the CRC coding with packet and control field, and inserts one group of code tail bits.Packet, control field, CRC parity check bit and code tail bits are formed formative grouping.In this exemplary embodiment, element in channel 42 is encoded to formatted grouping immediately, and the intersymbol error in the grouping that will encode (or rearrangement).In this exemplary embodiment, covering staggered processing with the Walsh code divides into groups, also with short PNI and PNQ code the grouping of the staggered processing of warp is expanded.Expanded data is provided for RF unit 44, and it carries out quadrature modulation, filtration and amplification to signal.Forward link signal is to transmit in air through the antenna on the forward link 50 46.

At distant station 6, forward link signal is received and is sent to the receiver of front end 62 inside by antenna 60.Receiver to signal filter, amplification, quadrature demodulation and quantification.Digitized signal is provided for demodulator (DEMOD) 64, goes expansion at this place with short PNI and PNQ code, and goes to cover with the Walsh code.Demodulated data are provided for decoder 66, and it carries out the reversing function of the signal processing function that 4 places are done in the base station, particularly is deinterleave, decoding and CRC check function.Decoded data are provided for data receiver 68.

Hardware supports as already pointed out is communicated by letter with other to data, message communicating, voice, video on the forward link and is carried out the transmission of variable bit rate.From the speed change of the data of data queue 40 transmission with the signal strength signal intensity that adapts to distant station 6 and the variation of noisy environment.Preferably, each distant station 6 is all controlled (DRC) signal to data transfer rate of associated base station 4 transmission at each time slot.DRC refers to a kind of controlling mechanism, and distant station is that forward link is confirmed desirable data transfer rate through this mechanism,, receives the data transfer rate of data at the distant station place that is.Distant station sends the desired data rate as data rate request or instruction via DRC message to the base station.Drc signal provides information to base station 4, comprises the identity of distant station 6 and the speed that this distant station will receive data from its data queue that is associated.Thus, the circuit measuring signal strength signal intensity at distant station 6 places is also estimated the noisy environment at distant station 6 places, with the rate information of confirming in drc signal, to transmit.

The drc signal that each distant station 6 is transmitted is advanced to link channel 52 along the back, and 4 places are received through antenna 46 and RF unit 44 in the base station.In this exemplary embodiment, DRC information by demodulation, and is provided for the channel scheduler 12A that is arranged in base station controller 10 in element in channel 42, or offers the channel scheduler 12B that is arranged in base station 4.In first exemplary embodiment, channel scheduler 12B is arranged in base station 4.In alternative embodiment, channel scheduler 12A is arranged in base station controller 10, and is connected to all selector elements 14 of these base station controller 10 inside.

In one embodiment, channel scheduler 12B receives the information that is designated as the data volume (also claiming queue size) that each distant station lines up from data queue 40.The queue size of each distant station that channel scheduler 12B is served based on DRC information and base station 4 is immediately dispatched.If the dispatching algorithm request queue of in alternative embodiment, using is big or small, then channel scheduler 12A can receive the information of queue size from selector element 14.

Each embodiment of the present invention is applicable to other hardware architecture that possibly support variable rate transmission.Can expand the present invention at an easy rate to cover the variable rate transmission of back on link.For example, base station 4 is not based on from the drc signal of distant station 6 confirms that 4 places, base station receive the speed of data, but intensity and the estimated noise environment of measuring the signal that is received from distant station 6 are to confirm to receive from distant station 6 speed of data.Base station 4 is transmitted the speed of the data that will transmit to link in the back from the distant station 6s immediately to each distant station that is associated 6.Base station 4 immediately can similar this paper in regard to the described mode of forward link, come the transmission of back on link dispatched to different data rates on the link based on the back.

Equally, the CDMA scheme is used in the base station 4 of embodiment discussed above, transmits to selected one or several distant stations, repels all the other distant stations that are associated with this base station 4 simultaneously.At any special time, base station 4 is through using a kind of code of distributing to this reception base station 4, and selected one or several distant stations 6 are transmitted.But the present invention is also applicable to other system, and these systems use different TDMA methods to provide data to select several base stations 4 and repel other base station 4 simultaneously, so that transfer resource is carried out optimized distribution.

The transmission of variable bit rate is dispatched on 12 pairs of forward links of channel scheduler.Channel scheduler 12 receives indication will be to the queue size of distant station 6 data quantity transmitted with from the message of distant station 6.Preferably, it is said that channel scheduler 12 logarithms are input into the row scheduling, to realize the data throughout maximization and to observe the aims of systems of fairness constraint simultaneously.

As shown in fig. 1, distant station 6 distributes and spreads all over communication system, and can communicate by letter with zero or a base station 4 on the forward link.In this exemplary embodiment, the forward link data transmission on the channel scheduler 12 coordination whole communication system.Specifically describe in No. the 6th, 335,922, the United States Patent (USP) that a kind of dispatching method that is used for high speed data transfer and equipment were authorized on January 1st, 2002, this patent has transferred assignee of the present invention, and incorporates into as a reference here.

According to an embodiment, channel scheduler 12 realizes in computer system, and this system comprises processor, random-access memory (ram) and is used to store the program storage (not shown) of the instruction that will be carried out by processor.Processor, RAM and program storage can be taken by each function of channel scheduler 12.In other embodiments, processor, RAM and program storage can be the part of the shared computational resource of other function that is used to carry out base station controller 10 places.

Fig. 3 illustrates a kind of embodiment of dispatching algorithm, and 12 pairs of this algorithm controls channel scheduler 4 are dispatched to the transmission of distant station 6 from the base station.As discussed above, a data formation 40 is associated with each distant station 6.Channel scheduler 12 makes each data queue 40 be associated with " weights " in the valuation of step 110 place, is used to select the particular remote station 6 that is associated with base station 4, from service intervals subsequently, receiving data.Channel scheduler 12 is selected single distant station 6, in discrete service intervals, to receive transfer of data.In step 102, channel scheduler is each formation initialization weights that are associated with base station 4.

Channel scheduler 12 circulates to 112 from step 104 by transmission intercal or service intervals.In step 104, channel scheduler 12 whether judge since in previous service intervals being associated between detected another distant station 6 and the base station 4, and have any other formation to add.Channel scheduler 12 is also carried out initialization to the weights that are associated with new formation in step 104.As discussed above, base station 4 receives drc signal by the interval such as rules such as time slots from each distant station 6 that is associated with it.

This drc signal also provides information to supply channel scheduler to use in step 106, thinks that the distant station that is associated with each formation confirms the momentary rate of consumption information (or receive transmit data).According to an embodiment, can be from this distant station 6 of drc signal indication of any distant station 6 transmission with any one the reception data a plurality of effective data rates.

Receive the momentary rate (as indicated in the drc signal that receives recently) of data based on being associated of distant station 6; Channel scheduler 12 is confirmed the length of service intervals in step 108; During this service intervals, data are sent to any specific distant station 6.According to an embodiment,, receive the momentary rate R of data in step 106 _iThe service intervals length L that decision is associated with particular data queue _i

Channel scheduler 12 selects particular data queue to transmit in step 110.The associated data amount that will send be retrieved from data queue 40 immediately, and be provided for element in channel 42, be used to be transferred to the distant station 6 that is associated with this data queue 40.Discuss as following, channel scheduler 12 selects to be used to provide the formation of data in step 110, this formation in ensuing service intervals with comprising that the information of each weights that is associated with each formation transmits.The weights that are associated with the formation of being transmitted obtain upgrading in step 112 immediately.

It will be understood by those skilled in the art that channel scheduler 12 can in all sorts of ways realizes, and can not depart from the present invention.For example, channel scheduler 12 can realize with the computer system that comprises processor, random-access memory (ram) and be used to store the program storage (not shown) of the instruction that will be carried out by processor.In other embodiments, each function of channel scheduler 12 can be incorporated shared computational resource into, and this computational resource of sharing also is used to carry out other function at base station 4 or base station controller 10 places.In addition; The processor that is used to carry out the channel scheduler function can be general purpose microprocessor, digital signal processor (DSP), programmable logic device, application-specific integrated circuit (ASIC) (ASIC), maybe can carry out the miscellaneous equipment of various algorithms described herein, and can not depart from the present invention.

Shown in the embodiment of Fig. 1, distant station 6 moves, and can change the association between the different base station 4.For example, originally 4F receives transfer of data to distant station 6F from the base station.Distant station 6F possibly shift out the unit of base station 4F subsequently and move into the unit of base station 4G.Distant station 6F warns base station 4G with beginning to transmit drc signal, rather than base station 4F.Through no longer receiving drc signal from distant station 6F, the logical deduction distant station 6F at base station 4F place has broken away from and has not prepared to receive transfer of data again.The data queue that is associated with distant station 6F is with being transferred to base station 4G via land line or RF communication link.

Adaptive weighted dispatching algorithm

In addition, when transmitting multimedia service in wireless communication system or other service with various transmission requirements, have a problem, the multimedia service transmission that wherein is called " flow process " causes the paroxysmal traffic.The characteristic of the paroxysmal traffic comprises paroxysmal measuring and average data rate by some variable descriptions.In addition, need satisfy service quality (QoS) requirement of each in each bar flow process in the system.Such as current dispatching method such as equitable proportion (PF) algorithm generally based on by requested data transfer rate (being called the request of data transfer rate control data or " DRC ") to the tolerance that the ratio of throughput (being designated as " T ") provides, select the flow process that will serve.This type of calculates and possibly can't guarantee the QoS that all users are required.Therefore, pure PF algorithm may not provide enough complexities to satisfy the user's of visit multimedia or other application program qos requirement.Need a kind of scheduler that can satisfy these demands of different.

The cdma2000 of considering that support is served like the High Rate Packet Data (HRPD) described in the IS-856 is discussed below the attention.This system uses as an example.The present invention is applicable to other system of the user that can select to serve according to dispatching algorithm.

In the HRPD system, air interface can be supported nearly 4 parallel application flow.First stream carries signaling information, and other three streams can be used for carrying application program or other application program with different service quality (QoS) requirement.

Understand for the purpose of the embodiment that hereinafter appears for clear, following vocabulary is provided.Following vocabulary is not attempted exhaustive.Following vocabulary is not attempted the present invention is limited to this, but for regard to the clear of a kind of embodiment of the communication system of supporting adaptive weighted dispatching algorithm with understand and provide.

Vocabulary

Access network (AN)-the provide network equipment of cellular network and packet switched data network (being often referred to the internet) and AT data connectivity between any two.AN in the HRPD system is equivalent to the base station in the cellular communication system.

The connective equipment of user's data accesses terminal (AT)-is provided to.AT in the HRPD system is corresponding to the mobile radio station in the cellular communication system.AT can be connected to such as computing equipments such as laptop personal computer, and perhaps it can be such as self-contained data equipments such as PDA(Personal Digital Assistant)s.

Application flows-be the specified transmission path from the source to AT of given application flow.Each application flows is identified by source, purpose, traffic overview and service quality overview.

Application flow-corresponding to an application's data communication.Most of application flow all have the service specified quality requirement.

Automatically repetitive requests (ARQ)-transmitter is based on the generation of an incident or do not take place and mechanism that initialization data transmits again.

Available resources (t): on the forward link in the unconfined bandwidth of time t.

The average input data transfer rate of average data rate (r)-given application flows a period of time.

The average retardation that is born on a plurality of groupings of average retardation (AvgD)-from AN to AT or the bit.

Sudden or the density of the packet in sudden (σ)-application programs flow process and the tolerance of time relationship.

Data transfer rate control (DRC)-AT transmits the mechanism of the data transfer rate of being asked to AN.

Dificiency grouping (defpkts)-defined by flow process k at the place that begins of time slot n.It is the also not grouping of transmission on stream that dificiency is divided into groups, and defpkts specifically is defined as the number above the grouping (for example, dividing into groups such as intermediate treatment such as media interviews control (MAC) groupings) of the equal sizes of the delay threshold value of flow process k in the BTS time of staying.

The figure place of dificiency position (defbits)-divide into groups corresponding to dificiency.

Postpone the fixed time that the border-data grouping of transmission is allowed from AN to AT.

Postpone the function on threshold value-delay border or shake border, be used to calculate defpkts.

The compensating factor that delay compensation factors (Ф)-be used for compensating delay is violated.

The compensating factor that DRC compensating factor (β)-worry and the request of data that is associated with the user of application flows require.Being used for application programs appropriately recovers.

Enhancing dithering threshold (dv)-calculate strengthening the jitter compensation function when being used for the shake violation between two that detect flow process IP divides into groups in succession.

Flow process weights (w)-the be applicable to initial weight that each uses the application flows of adaptive weighted dispatching algorithm.Adaptive weight (aw) is the adaptation value of weights.

The transmission airlink of forward link (FL)-from AN to AT.

First grouping in file leader (HOL) grouping-formation.

High Rate Packet Data (HRPD)-with the data, services of high data rate transfers block data communication.Be also referred to as High Data Rate (HDR), in the IS-856 standard that is entitled as " cdma2000 High Rate Packet Data Air InterfaceSpecification " (cdma2000 High Rate Packet Data air-interface standard), specify.

Shake-received continuous grouping time between any two changes.

The border on shake border (j)-in the shake of given application flows.

Strengthen the jitter compensation factor (δ)-the be compensating factor of flow process compensate for jitter violation.

L _Max-BTS can transmit the maximum rate (in cdma2000 1xEV-DO type network, being 2.4 mbit/for example) of data on the forward link.

The statistic of L (t)-violate based on the previous QoS statistic relevant with network load is to the estimation in the forward link capacity of time t reservation.

Normalization dificiency grouping (ndefpkts)-divide into groups with the normalization dificiency that dificiency is divided into groups and this flow process institute rate request is calculated.

The normalization dificiency position of normalization dificiency position (ndefbits)-divide into groups corresponding to the normalization dificiency.

The agreement of Motion Picture Experts Group (MPEG)-transmitting multimedia material.

Unsettled grouping-pend _{K, j}The unsettled byte number of the IP grouping j of flow process k among BTS and the BSC among [n]-time slot n.

The dispatching algorithm that equitable proportion (PF) algorithm-basis is dispatched for the selection factor pair data communication of each AT calculating the ratio of throughput by the data transfer rate of being asked.

Service quality (QoS)-relate to the requirement of block data communication transmission includes but not limited to, delay, request rate and shake.

QoS and network building out function (Ф, γ, α, β, δ)-penalty function as in adaptive weighted dispatching algorithm, using.

Affair quality group (QSG)-the have Application Type group of similar qos requirement.

The compensating factor calculate of the rate compensation factor (α)-violate for compensate for rate.

Service speed (R) or institute's rate request (required_rate)-flow process institute rate request.

Res (t): on the forward link in time t bandwidth reserved.

Resend the formation of application flows that formation (Rx)-be stored as resends and dispatch.

The back is to the transmission airlink of link (RL)-from AT to AN.

Select to measure (Y)-compare the used tolerance of application flows for dispatching decision.

Volume of traffic overview (σ, r)-relate to sudden and tolerance data transfer rate.

Transmit queue (Tx)-be the transmit queue of given BTS application storing flow process.

The tolerance of the stand-by period that the inner file leader IP of stand-by period parameter (γ)-AN divides into groups.

Adaptive weight is applied to the Proportional Fair algorithm

For the forward link of cdma2000 1xEV-DO network has been described equitable proportion (PF) dispatching algorithm, it is based on measuring the flow process that the DRC/T selection will be served.The PF algorithm is designed to provide to each user the transmission time slot of about same number.For strengthening this dispatching algorithm, described herein is a kind of adaptive weighted DRC/T algorithm, and its expansion is also optimized the DRC/T algorithm to satisfy the different QoS requirement of dissimilar application programs.Each multimedia application has concrete separately qos requirement respectively.The target of dispatching algorithm comprises and satisfies various qos requirements.Given adaptive algorithm (being also referred to as self adaptation w*DRC/T algorithm) comprises the forward link of the cdma2000 1xEV-DO network of multimedia application service for application flows wherein among this paper, and the performance advantage of the various DRC/T of being superior to algorithms is provided.Use adaptive algorithm, satisfy the delay and jitter border requirement of delay and jitter sensitive applications on the cdma2000 1xEV-DO network forward link.In addition, the adaptive scheduling algorithm guarantees to satisfy rate requirement and the average retardation that reduces multimedia application.Although the realization that provides multimedia application that the adaptive scheduling algorithm is described as an example, method and apparatus described herein is applicable to other application program that quantizes to require that has qos requirement or be associated with it.

For have the application program that speed and stand-by period require such as network browsing and recreation etc., the adaptive scheduling algorithm provides speed to guarantee and reduces average retardation.Only have the application program of rate requirement for other, can use adaptive weighted dispatching algorithm to satisfy speed and guarantee.When providing these QoS to guarantee, adaptive weighted dispatching algorithm also acts on total throughout is maintained high grade rationally, and realizes approaching the total throughout that when the pure PF dispatching algorithm of use, reached.Pure PF dispatching algorithm refers to use the algorithm of DRC/T calculating.In the flow process extra resource that has QoS to violate, adaptive weighted dispatching algorithm is distributed available resources with fair mode.The various compensation mechanism consistent with it are provided among this paper.

Fig. 4 illustrates the system 800 that supports multimedia application.Notice that once more the present invention is applicable to other system with qos requirement.System 800 comprises the multimedia sources 802 that is coupled to packet data serving node (PDSN) 806.PDSN806 also is coupled to base station controller (BSC) 804, and base station controller 804 can comprise a plurality of BSC.BSC804 communicates by letter with each AT 812,814,816,818 etc. via base transceiver systems (BTS) 808,810.System 800 can comprise than more BTS of diagram and AT.Illustrate three flow processs: article one flow process from multimedia sources 802 via PDSN806, BSC804 and BTS808, to AT812; The second flow process from multimedia sources 802 via PDSN806, BSC804 and BTS810, to AT816; Article three, flow process from multimedia sources 802 via PDSN806, BSC804 and BTS810, to AT818.Note the purpose that AT can be many flow processs.In one example, the transmission of Motion Picture Experts Group (MPEG) types of applications is divided into independent flow process with Voice & Video.

Each application flows that will transmit in the system 800 has: the source address that is associated; Destination address; And qos requirement.Just the transmission application programs flow process from the source to the purpose is dispatched immediately.Application flows is passed and is similar to the path shown in Fig. 4.

Each BTS 808,810 is applicable to maintenance flow process formation as shown in Figure 5.Notice that each BTS safeguards the set of queues corresponding to each application flows on its forward link (FL).An application flows is directed to an AT.But, notice that many flow processs can be directed to an AT.Every flow process all has a service quality group associated therewith (QSG) type.Each QSG is defined by one group of qos parameter.Every flow process of a certain given QSG all has concrete value corresponding to each parameter in the group.For example, a QSG can be defined by the group that comprises delay and jitter.Which flow process with this QSG will be specified the requirement for delay and jitter.For every in formation flow process, BTS safeguards the group that comprises following three independent formations: (1) original transmitted formation (Tx); (2) transmit queue (Rx) again; (3) repetitive requests formation (ARQ) automatically.In one embodiment, the ARQ formation can be corresponding to the formation that such as previous decision-making ARQ etc. is the repeat mechanism Stored Procedure of any kind performed between BTS and AT.Multimedia application can comprise having such as video conference etc. and postpones the application program to delay-sensitive that the border requires.Postponing the border is to be transferred to AT from AN to receive the fixed time that is allowed.The average retardation that the IP grouping that adaptive weighted algorithm is devoted to satisfy the requirement of delay border and reduce this type of application program is stood.For the application program that has speed and average retardation requirement simultaneously, adaptive weighted algorithm is devoted to satisfy rate requirement and is reduced average retardation.

For considering that such as another of the application program of some type such as multimedia video application program item is " shake " that is stood between the grouping continuous in the multimedia transmission.Shake refers to that the grouping time between any two that is received changes.When continuous wave is a little earlier or slightly late when reaching receiver, promptly shake.In radio communication, this type of waveform transmits logical one or 0 usually, and they are decoded at the receiver place immediately.The time that is defined as shake changes the visual effect that makes received transmission and produces distortion.The delay that adaptive weighted dispatching algorithm reduces worst case changes, and the delay between the dividing into groups in succession of the application program of delay-sensitive is changed.

When satisfying each user's qos requirement, adaptive algorithm also is designed to satisfy when some application flows " unanimity " rate requirement of those flow processs.If application flows is sent data according to preassigned traffic overview, then be called " consistent ".If it is inconsistent to have the flow process of rate requirement, that is, the data that they send are preassigned more than institute in its traffic overview, and then this algorithm has higher priority the flow process of lower data rate.Although in the context of cdma2000 1xEV-DO, describe adaptive weighted algorithm among this paper, these notions and method are also applicable to the wireless network of other type.

With regard to the multimedia application flow process, every flow process is defined by the following: (1) traffic overview; (2) QoS overview; (3) Internet Protocol (IP) source address; (4) IP destination address.Flow process also can comprise: (5) L4 protocol type; (6) L4 port numbers; (7) L4 destination slogan, wherein L4 refers to transmission control protocol (TCP)/UDP (UDP) layer in the protocol stack.For example, can mpeg audio and the MPEG video corresponding to a MPEG application program be regarded as two independent flow processs.

Every flow process is all specified by traffic overview, and come under observation or shaping consistent to guarantee it with this traffic overview.Traffic overview is by the variable of the paroxysmal tolerance of expression that is designated as σ and be designated as the flow process average data rate definition of r.Therefore, (σ r) describes every flow process by traffic overview.The QoS overview is by at least one definition in following each parameter: (1) is designated as the delay border of " D ", and its definition IP divides into groups from being transferred to the time that reception allows.For the multimedia application flow process, but system specified delay border.For such as some other application flows such as network browsings, system can specify average retardation (AvgD) to replace or the complementary delay border; (2) be designated as the shake border of " j ", it is defined in the maximum between any two admissible time variation of grouping that the AT place receives; (3) be designated as the service speed (request rate) of " R " or " req_rate ".

For definition postpones border D, with reference to figure 6, it is the sequential chart that comprises each an AN element and an AT.The multimedia flow process is transferred to AT from multimedia sources (not shown) via PDSN, BSC and BTS.An IP is grouped in time t ₀Send from PDSN, and at time t ₃Receive at the AT place.Parameter D definition is from time t ₀To time t ₃Maximum can allow the time, that is, D specifies t ₃-t ₀Restriction.

Be definition shake border j, with reference to figure 7A, it is the sequential chart that comprises a several A N element and an AT.First is grouped in time t ₁Send from PDSN, and at time t ₁' receive at the AT place.Second is grouped in time t ₂Send from PDSN, and at time t ₂' receive at the AT place.The shake border j definition maximum between any two admissible variation of grouping continuously wherein changes being given as (t ₂'-t ₁')-(t ₂-t ₁).Fig. 7 B further is described in detail in the plurality of continuous IP that transmits on some time slots and divides into groups.

In one embodiment, the QoS overview is classified into group, is called QoS dispatch group (QSG).Table 1 has been listed these classifications.

Table 1

Index	Postpone border (D)	Shake border (j)	Service speed (R)	Average retardation (AvgD)	The application program example
						1	X	X	X	-	The MPEG meeting, VoIP, video flowing
2	-	-	X	X	Network browsing
						3	-	-	X	-	FTP
4	-	-	-	-	The best effort type

Fig. 8 illustrates the processing according to adaptive weighted dispatching algorithm flow.Flow process 900,902,904 and 906 is handled by the scheduling unit that is labeled as " S " 908.Scheduling unit 908 is used a kind of adaptive weighted dispatching algorithm, wherein is that every flow process is used the QSG overview.The identification of QSG overview is used for calculating like following detailed description ground the variable of adaptive weight.Scheduling unit 908 outputs to selected AT with being about to scheduled transmission 910.

Description is called the PF dispatching algorithm of DRC/T algorithm, and wherein each divides into groups to be classified into m formation, for example Q1, Q2 ..., Qm.Make DRC that [k, n] is the DRC by the mobile radio station request, corresponding to the flow process k of time slot n.Scheduler selects to have the flow process of the highest selectance value, wherein:

$Y[k,n+1]=\mathrm{DRC}[k,n+1]/T_k[n+1],\∀k,\∀n.---\left(1\right)$

Y [k, n+1] is the selection tolerance of formation Qk in the time slot (n+1), and

$T_k[n+1]=(1-\frac{1}{t_c})T_k\left[n\right]+\frac{1}{t_c}R[k,n],---\left(2\right)$

$\frac{1}{t_c}\≤1$ , and

0<1/>t _c (3)

As used herein, t _cIt is the used time constant of calculating mean value.

Self adaptation w*DRC/T algorithm

In one embodiment, be called the adaptive weighted dispatching algorithm of " self adaptation w*DRC/T " algorithm, distribute an initial weight to every flow process.Suppose that the initial weight of distributing to flow process k is designated as w _k, be DRC [k, n] by the DRC corresponding to time slot n flow process k of AT request.Self adaptation w*DRC/T algorithm is a tolerance below every flow process k calculates at each time slot n:

Y _k[n]＝aw _k[n]*DRC _k[n]/T _k[n] (4)

Here, the throughput T of flow process k and time slot n _k[n] is as defined for DRC/T in the PF algorithm.As employed in adaptive weighted dispatching algorithm, aw _k[n] is the adaptive weight of flow process k at time slot n.Self adaptation w*DRC/T dispatching algorithm is with several modes work, and wherein pattern is defined by QSG.Flow process k is at the adaptive weight aw of time slot n _k[n] is based on that the scheduler pattern that is described below and one group of selected strategy or mechanism calculated.Attention will be every flow process calculation equation (4), wherein will calculate adaptive weighting according to the formula that is specific to every flow process.In other words, this dispatching algorithm is considered the QoS overview of given flow process, and uses this QoS overview to construct the calculating to the adaptive weight of this flow process.In this way, have the different flow processs that different QoS requires and to have the adaptive weight that calculates differently.Next this dispatching algorithm selects to have maximum Y _kThe flow process of [n] value is to serve in time slot n.

Self adaptation w*DRC/T scheduler is with following pattern work:

Pattern I [aw*DRC/T] (r, d, j): the delay and jitter border is had strict demand and requires certain minimum-rate for the delay and jitter sensitive applications designs.

Pattern II [aw*DRC/T] (r, d): the application program that is used to have average retardation and rate requirement.

Pattern III [aw*DRC/T] is (r): be used for the only application program of assigned rate requirement.

Pattern IV [DRC/T]: be used for not specifying any QoS plan but by the flow process of DRC/T algorithm service.

Based on qos requirement, can use the self adaptation w*DRC/T algorithm of AD HOC to given flow process.Also but flow application model II is to increase the throughput that scheduler gives this flow process.For example, pattern II can be used for the FTP application program, to increase the throughput of corresponding application flows potentially.

Below provide the packet applications program (that is QSG) an example:

Group I: be strict with the application program that is similar to IP phone (VoIP) with postponing to change to have to postponing the border.Notice that this type of application program also usually has rate requirement.Use scheduler pattern I.

Group II: have the multimedia conferencing application program of strict demand with postponing to change to postponing the border.In instant these application programs some are adaptive, for the purpose of the high-quality of unanimity, guarantee that service speed remains desirable.Use scheduler pattern I.

Group III: have the video stream application program of requirement for postponing border, speed and delay variation.Use scheduler pattern I.

Group IV: network browsing application program with speed and (on average) delay requirement---use scheduler pattern II.

Group V: FTP application program with rate requirement---use scheduler pattern III.Perhaps, use scheduler pattern II with undemanding delay restriction.

Group VI: best effort type application program---use no adaptive weighted PF algorithm, that is, and the DRC/T algorithm.

Notice that db transaction, recreation and other application program also can be categorized in the suitable group according to qos requirement respectively.

Fig. 9 illustrates the adaptive weighted scheduler with a plurality of grades, and a plurality of grades include, but not limited to grade I and grade II.Grade I scheduler have a plurality of scheduler S1, S2, S3 ..., Sm, wherein m refers to the sum organized.A certain operational modes of each the grade I scheduler operation self adaptation w*DRC/T dispatching algorithm among Fig. 9, and from this group, select a flow process.At first, grade I scheduler calculates the part of Y, particularly is throughput T, and the rate compensation factor-alpha.Next, grade II scheduler is considered each bar flow process, and to grade I scheduler enough inputs is provided, to be accomplished selecting the calculating of tolerance Y by grade I scheduler.In case be that all unsettled flow processs are accomplished the calculating to Y, grade I scheduler carries out valuation to the Y value, and selects to have the flow process of the highest Y value.Each grade I scheduler carries out valuation to one group of flow process with similar qos requirement.The selected flow process of each grade I scheduler is provided for grade II scheduler immediately, for comparing with flow process from other group.Grade II scheduler is considered every group of selected flow process, and selects to have the flow process of maximum metric (aw*DRC/T) or Y value.When scheduler need select a flow process to serve, for each time slot repeats this process.Alternative embodiment can be used the scheduler of single grade, maybe can use than more grade shown in Fig. 9.Alternative embodiment can comprise the grade I scheduler of different numbers, and its middle grade I scheduler is corresponding to process organization.

Generally speaking, be the function that is given as several parameters with the calculating of adaptive weight, and be given as:

a＝f(Ф，γ，α，β，δ)。(5)

The delay compensation function is designated as Ф.The stand-by period constant is designated as γ.The rate compensation function is designated as α.The DRC penalty function is designated as β.Strengthen the jitter compensation factor and be designated as δ.Note it not being all to have actual value for all parameters of all multimedia services.For example, when only the qos requirement of given flow process was the data transfer rate of appointment, variable α was designated so, and (rate parameter will have actual value) and all other constants will be set as and equal 1.In this case, adaptive weight will only comprise rate parameter in calculating.In one embodiment, adaptive weighting is calculated as follows:

a＝Ф*γ*α*β*δ (6)

Wherein operator is a multiplication sign.The details that the various compensation term that can comprise in the adaptive weight calculating are provided below is discussed.

For the application program of pattern I, the QoS overview is specified all indicated parameters in the equality (6).Adaptive weight calculates to be considered owing to postponing the delay compensation that threshold-violating produces; The delay compensation that produces owing to the stand-by period threshold-violating; Violate rate compensation and because the reinforcement jitter compensation that the violation of reinforcement dithering threshold produces produces owing to speed.This notion improves the weights of violating the flow process of specifying qos requirement.Triggered in case violate qos requirement, this type of flow process promptly is given to trust divides.Trust to divide be to multiply by that the suitable value of delay compensation function realizes through weights with flow process.This result also will multiply by rate compensation and strengthen jitter compensation.

On the contrary, when flow process showed as in the service that reception exceeds the quata, this flow process can be punished.Any flow process of punishing in can ining all sorts of ways.According to a kind of method, can come directly punishment flow process through reducing the flow process weights.According to another kind of method, can also increase other (that is, those do not reach the flow process of desired QoS) user's who falls behind the next punishment indirectly of weights flow process simultaneously through keeping these flow process weights.

Various computing relay compensation are arranged to consider to violate the mechanism that postpones threshold value.The delay threshold value of supposing flow process k is designated as dth_ Ф _k, flow process k is designated as Ф owing to postponing the delay compensation that threshold-violating produces in time slot n _k[n].Be computing relay compensation Ф _k[n] considers the grouping in all three formations of every flow process (that is, Tx, RTx and ARQ).

For every flow process, also Ф is specified minimum and maximum threshold value, with guarantee a flow process can not consume several in succession time slot and make the deficient time slot of other flow process.Design guarantees that also the flow process delay compensation item that produces owing to the delay threshold-violating is the same with minimum threshold at least good like this.Make Ф _{Thres, min, k}And Ф _{Thres, max, k}Be minimum and max-thresholds for every flow process k appointment.These (for all k and all n) cause:

${\mathrm{\&phi;}}_{\mathrm{thres},\min ,k}\&le;{\mathrm{\&phi;}}_k\left[n\right]\&le;{\mathrm{\&phi;}}_{\mathrm{thres},\max ,k},\&ForAll;k,\&ForAll;n.---\left(7\right)$

Derive to the calculating of delay compensation to give a definition using:

D [n]: be defined in time slot n the set of process that begins to locate to stand to postpone threshold-violating (that is, each this type of flow process time slot n begin locate all to have at least one and surpass the grouping that this flow process postpones threshold value).

Defpkts _k[n]: be defined in " dificiency " grouping that begins to locate flow process k of time slot n.It is the also not grouping of transmission on stream that dificiency is divided into groups, and defpkts specifically is defined as the number that (MAC) of equal size that in BTS time of staying surpasses the delay threshold value of flow process k divides into groups.

Required_rate _k: the request rate of definition flow process k.

Ndefpkts _k: the normalization dificiency grouping number of definition flow process k specifically is defined as:

${\mathrm{ndefpkts}}_k=\frac{{\mathrm{defpkts}}_k}{\mathrm{required}\_{\mathrm{rate}}_k}---\left(8\right)$

Notice that the grouping among BTS, BSC and the PDSN possibly be unequal size, therefore, the number of calculating dificiency position rather than grouping here is good.

If the HOL of flow process is grouped in time period of stopping in the BTS formation greater than preassigned threshold value, then available below mechanism this flow process is compensated.The stand-by period threshold value of using for this purpose should be more than or equal to being the employed threshold value of calculating Ф.For flow process k, the stand-by period threshold value is designated as dth_ γ _k, wherein the stand-by period threshold value receives dth_ γ _k>=dth_ φ _k,

The constraint of k.For the HOL that selects flow process divides into groups, consider that at first the HOL of Tx, RTx and ARQ formation from flow process divides into groups, and based on the stand-by period selection at BTS place one of them, that is, be chosen in that waited for maximum duration among the BTS.Make γ _k[n] is for beginning to locate the stand-by period compensation of flow process k, S at time slot n _k[n] is grouped in the time that stops in the BTS formation for the HOL of the flow process k at time slot n place.For every flow process k, also specify minimum threshold S _{Thres, min, k}With max-thresholds S _{Thres, max, k}, satisfy S _{Thres, min, k}≤S _k[n]≤S _{Thresh, max, k},

According to an embodiment, when flow process stands to postpone threshold-violating or stand-by period application delay compensation during threshold-violating.This mechanism is applied to adaptive weight with the DRC data rate request.Make β _k[n] is the DRC adjustment function of flow process k among the time slot n.For every flow process k specifies minimum threshold β _{Min, thres, k}With max-thresholds β _{Max, thres, k}, satisfy β _{Min, thres, k}≤β k [n]≤β _{Max, thres, k}

Although for such as some application programs such as video/audio meetings, the delay that above-mentioned compensation mechanism helps to reduce in the flow process changes, and comprises more effectively postponing to change (shake) control and reducing further that to postpone variation possibly be desirable.The delay that following mechanism is divided into groups between any two through the minimizing flow process in succession changes, and provides effective delay to change control.Having the divide into groups flow process of size of big IP from then on benefits in the compensation mechanism more.

Suppose that (k j) is the time of advent of the IP grouping j of BSC porch flow process k at.Make dt (k, j) IP divided into groups from the time that BTS leaves for this reason, that is, all segmentations that this IP divides into groups till this time are all transmitted by the forward link schedulers at BTS place.Make pend _{K, j}[n] is the byte total length of the IP grouping j of BTS and the flow process k of BSC place.Suppose dv _{K, target}Be flow process k IP grouping target delay variation (shake) between any two in succession, dv _{K, thres}The predefined enhancing dithering threshold of flow process for this reason, and make dv _{K, thres}<dv _{K, target}In one embodiment, change and surpass dv when divide into groups between any two delay of IP in succession _{K, thres}The time, this algorithm is flow process k and triggers enhancing delay compensating for variations mechanism.

Figure 10 provides the system assumption diagram corresponding to an embodiment.Each application flows is described by traffic overview, QoS overview and DRC request (that is requested data rate).Each traffic overview comprises paroxysmal tolerance and average data rate.Each QoS overview type of comprising type and bound of parameter.Type type can be a kind of among pattern I, pattern II, pattern III or the pattern IV.The border of border specified delay, shake and the data transfer rate of being asked.Can specify average retardation rather than postpone the border such as some application programs such as network browsings.The delay threshold value of preference pattern I is less than the shake border; For pattern II, select to postpone threshold value less than average retardation.Select to strengthen dithering threshold less than the shake border.Alternative embodiment can be applied to each application flows with more or less information, and wherein qos requirement can be specific to network and configuration.

Figure 11 is the table of specifying qos requirement and qos parameter for each type type.As shown in the figure, pattern I is corresponding to the strictest requirement, and pattern IV is corresponding to the best effort type of not specifying any qos requirement.Alternative embodiment can comprise other qos requirement, qos parameter and/or pattern.

Figure 12 A as the effective part of application flows, illustrates the processing of application programs flow process and to the scheduling of this application flows to 12E.It shown in Figure 12 A the flow chart that single application flows is carried out initialization and setting.This process is used for the mechanism of each compensating parameter with selection in step 1100 beginning.Compensating parameter includes, but are not limited to: postpone (Ф); The unsettled time (γ); DRC (β), shake (δ) and speed (α).Select threshold value in step 1102 for the compensating parameter that is suitable for.Notice that compensating parameter can comprise any parameter that has the application flows of importance for AN.The algorithm of weights in the middle of step 1104 is selected to be used to calculate, wherein middle weights are used to calculate the adaptive weight that dispatching office is used.In step 1106 scale parameter (C) and priority factor (Z) are set, the two all is used to calculate adaptive weight.Step 1108 for this reason application flows initial weight is set.Qos requirement in step 1110 pair this application flows is carried out valuation.If except the speed that DRC requires to be identified, do not have the qos requirement of appointment, then be suitable for implied terms.Implied terms is aforesaid " best effort type ".In this case, default treatment all compensating factors that will be used for this application flows are made as and equal 1.For present embodiment, in this case, to the suitable multiplication operator of the calculating of equality (6), therefore each factor is made as 1 ignores those factors effectively, that is, those factors do not influence weighting.Notice that alternative embodiment can realize other mechanism and function, and therefore suitable other mechanism is ignored concrete or all compensating factors.

The best effort type is handled in step 1112 and is continued with 1116.The scheduling factor calculation of gained is calculated consistent with equitable proportion.If application flows has qos requirement, then handle and advance to step 1114.Step 1114 is handled with 1116 indications and in ensuing accompanying drawing, is continued.

Figure 12 B continues the processing of Figure 12 A from step 1114.Processing in step 1120 pair current time slots begins.Judge the class type of application flows in step 1122.At step 1128 tupe I, at step 1126 tupe II, at step 1124 tupe III.Qos parameter at step 1128 monitoring mode I; Qos parameter at step 1124 monitoring mode III.At step 1130,1140 and 1150 places QoS is violated immediately and check that this further details in Figure 12 C and 12D.

For the application program of pattern I, II or III, the processing of application flows continues in the step 1130 of Figure 12 C.In step 1132, this algorithm is periodically kept watch on speed and is violated.Note periodically carrying out rate compensation and calculate, and in a plurality of time slots thereafter, use.Violate if detect speed, then handle advancing to step 1138 with computation rate compensating factor (α) in step 1134.Otherwise, in step 1136 the rate compensation factor (α) is made as and equals 1.Handle advancing to step 1160 immediately, this will further detail in Figure 12 E.

For the application program of pattern I or II, the processing of application flows continues in the step 1140 of Figure 12 C.Delay and jitter is violated at each time slot in this method of step 1142 and to be kept watch on.Postpone and/or shake is violated if detect, then handle advancing to 1148 with according at the selected machine-processed computing relay compensating factor of initialization (Ф) in step 1144.For the flow process of the pattern I of enhancing jitter compensation, so also calculate and strengthen the jitter compensation factor (δ) with request.For the pattern I flow process of the enhancing jitter compensation that does not have request with for the flow process of pattern II, δ is made as 1.Otherwise, in step 1146 delay compensation factors (Ф) is made as and equals 1, and δ is made as equals 1.Handle advancing to step 1160 immediately, this further details in Figure 12 E.Attention is for the application flows of pattern I or pattern II, can be in tandem or violate inspection concurrently.In other words, can carry out speed violation and delay/shake in time continuously or concurrently and violate inspection.

For the application program of pattern I, the processing of application flows continues in the step 1150 of Figure 12 D.Keep watch on for the stand-by period violation in this method of step 1152.If detect the stand-by period violation, then handle advancing to step 1158 to calculate stand-by period compensating factor (γ) according to the mechanism of selecting in initialization in step 1154.Otherwise, in step 1156 stand-by period compensating factor (γ) is made as and equals 1.Handle advancing to step 1160 immediately, this further details in Figure 12 E.Attention is for the application flows of pattern I, can be one in front and one in back or violate inspection concurrently.In other words, can carry out the inspection that speed is violated, delay/shake is violated and the stand-by period violates in time continuously or concurrently, shown in step 1170 and 1172.

Figure 12 E illustrates the processing of commentaries on classics from step 1160 and 1116.According to qos parameter and compensating factor, calculate adaptive weight in step 1162, be given as for application flows:

aw＝f(Ф，γ，α，β，δ) (9)

Be calculated as follows the scheduling factor or scheduling tolerance in step 1164:

The scheduling factor=aw* (DRC)/T (10)

In step 1166, according to be the scheduling factor that each effective application program circuit calculate gained, dispatch immediately by the application programs flow process for this dispatching algorithm.

Figure 13 illustrates the BTS1200 that is fit to the application schedules algorithm according to an embodiment.BTS1200 comprises scheduling unit 1202, application flows processor unit 1206, qos parameter valuation 1204, adaptive weight computing unit 1212 and CPU1208, more than each all is coupled to communication bus 1210.Scheduling unit 1202 is through preparing the scheduling factor, selecting to dispatch between each effective application flows according to these scheduling factors subsequently for each application flows.Be incorporated in the dispatching algorithm for the strategy and the target of fixed system.1204 couples of QoS of qos parameter valuation violate and keep watch on, and to scheduling unit 1202 and weight calculation unit 1212 information are provided.The application flows processing execution is handled, and includes, but not limited to groupings is directed to target AT, receives the QoS information that is used to dispatch, also this information is offered qos parameter valuation 1204 from target AT.BTS1200 also comprises memory 1214, is used to store average information and safeguards the data that are used for calculating mean value, flow process formation etc.Violating inspection carries out at the BTS place.An embodiment continues the byte number that sends for every flow process is counted, and this information is used for the inspection that speed is violated.When each packet arrives BSC place, it is all stamped timestamp.As long as divide into groups still to rest on AN, BSC or BTS, the time just continues cumulative.BTS is used for detection threshold with this time and violates, and immediately according to flow process computing relay, stand-by period or enhancing jitter compensation function.

Allow to get into control

Allow to get into control and refer to allow the decision process of the user of request msg service in getting into.When new user request when having the data, services such as application program of qos requirement, AN has determined whether that available resources support this use.Allow the entering process to consider institute's application requested, current use and QoS and network statistics.If AN confirms to support this new user that the corresponding application program flow process promptly is allowed to so.Otherwise if current do not have available resources, then application flows is rejected or is placed in the formation change with wait state.Notice that in fact new user can be the current effective application flows with request extra-service, that is, and the additional application flow process.

Except allowing entering control, and, can realize a kind of preemption processes that is used to stop effective application flows, wherein current operational circumstances carried out the decision-making that preemption is made in valuation as its part.In this case, with regard to QoS violation and data transfer rate each current flow process is carried out valuation.

The permission that this joint provides a kind of adaptive each sector gets into control algolithm.This permission gets into the control algolithm decision and in given radio multimedium network, whether allows (or preemption) flow process.Therefore admissible in the given network (each type) flow process quantity possibly can be confirmed.Some embodiment that given permission gets into control algolithm among this paper comprise and carry out simultaneously between the user and keep watch on and with this information of being about to is applied to allow to get into and/or preemption is made a strategic decision mechanism with indoor QoS.These embodiment are designed to guarantee flow process and the user for allow entering, and every flow process and each user's qos requirement is met.These mechanism help to coordinate to allow to get into control algolithm and hierarchy dispatching algorithm.

Scheduling and permission get into the part that control is forward link in the wireless network (FL) QoS management, and wherein this type of management is a complicated problems.QoS management is important consideration in design and the operation of communication network.Application flows is classified according to waiting like the defined criterion of system.In one embodiment, classification is according to qos requirement.At first, allow to get into control and confirm under current operational circumstances, can to allow the flow process quantity that gets into.This flow process quantity is divided into the flow process quantity of each type immediately.System operates to satisfy the qos requirement of each flow process of entering that allowed immediately.The quantity of noting flow process can dynamically change in time and with type of application.For example, in the very first time, access network (AN) can support every type of application program is permitted first situation of concrete quantity flow process.In second time, AN can support second situation at least one type in types of applications program permission varying number flow process.

Scheduler (that is dispatching algorithm) is realized a kind of fairness policy in the flow process that is allowed.Scheduler is also attempted the flow process that has QoS to violate is appropriately recovered.Operator's income and interests are the validity that depends on used dispatching algorithm.Algorithm more effective and feature rich provides the chance that increases these interests.

With regard to allowing entering control, an embodiment realizes a kind of method based on the factor of reservation.Attention gets in the control algolithm based on the permission that the method for subscribing usually is used in cable network.In wireless network, each user's channel situation continues to change, and therefore also continues to change like the being seen forward link capacity of BTS scheduler.Based on the fixing link capacity of algorithm hypothesis of wired reservation factor, and therefore can not be directly applied for wireless network.

For wireless network, an embodiment gets into control algolithm, the wherein a plurality of application flows with qos requirement of network support for the FL management provides the permission of a kind of adaptive reservation factor base (ASF).ASF in the wireless network allows to get into control through monitoring qos statistic and network statistics, comes dynamically to upgrade the reservation factor.Can use various mechanism to carry out update functions.Therefore can utilize the adaptive reservation factor to take the action of correction property.In addition, ASF also is used to realize the preemption method.

Calculate ASF, AS (t) at each time t.This process is that AS (t) confirms minimum threshold AS _{Min_prespecified}And max-thresholds AS _{Max_prespecified}, so that 1≤AS _{Min_prespecified}≤AS (t)≤AS _{Max_prespecified}<∞,

When initial, to ASF apportioning cost S _Initial, so that AS (0)=S _Initial

Figure 14 is a sequential chart, surveys and draws out maximum data rate, reserved bandwidth and available bandwidth as the function of time.BTS can transmit the maximum rate (L of data on the forward link _Max) upper bound of resource allocation is provided.Effective application flows is carried out valuation to confirm reserved bandwidth Res (t).Use QoS to violate the statistic relevant, carry out the calculating and forward link capacity L (t) estimation of adaptive reservation factor hoping at time t to keep with offered load.Notice that L (t)≤Res (t) is possible.For example, suppose that the flow process that is allowed experiences extraordinary channel situation when they are allowed to get into.Now, the channel situation variation of some flow processs and some flow processs do not reach the QoS assurance that is associated.In this case, system possibly hope allowing more to become more conservative on the multipaths, and establishes Avail (t)=0.On the other hand, if L (t)>Res (t), then system can establish Avail (t)=L (t)-Res (t).Value L (t) is based on previous QoS and violates the statistic relevant with network load, to the estimation of the forward link capacity of hoping at time t to keep, and according to L _MaxCalculate with ASF, as follows:

$L\left(t\right)=\frac{L_{\max }}{\mathrm{AS}\left(t\right)},\&ForAll;t,---\left(11\right)$

It is constrained to:

$L\left(t\right)\&le;\frac{L_{\mathrm{Max}}}{{\mathrm{AS}}_{\mathrm{Min}\_\mathrm{Prespecified}}}<L_{\mathrm{Max}}$ And (12)

$L\left(t\right)\&GreaterEqual;\frac{L_{\max }}{{\mathrm{AS}}_{\max \_\mathrm{prespecified}}}>0.---\left(13\right)$

Available bandwidth Avail (t) is calculated as follows:

Avail(t)＝maximum(L(t)-Res(t)，0) (14)

The tolerance of various resources as shown in Figure 14 is to be confirmed by data transfer rate control (DRC) request of data that receives from the user.Each user sends the DRC request of data on link in the back.In the system of cdma2000 1xEV-DO or other HRPD type, the user sends the DRC request of data on the time slot of each RL transmission.Shown in figure 15, change in time from user 1 request of data (DRC1) with from user 2 request of data (DRC2).These request of data and desired QoS confirm reserved bandwidth (Res).Notice that it is as an example that the relation between this DRC value and the Res is provided.Alternative embodiment and situation possibly cause different relationships.

Each application flows all has the designated communication amount overview of Mean Speed and sudden aspect, wherein flow process f _kTraffic overview by (σ _k, r _k) given.Here, r _kBe flow process f _kAverage request speed, σ _kBe paroxysmal tolerance, flow process f wherein _kInstitute's request rate be given as req_rate (f _k)=r _k

According to an embodiment, allow to get into control and just allow to get into flow f _kCarry out valuation.Allowing to get into control at first uses corresponding to flow process f for the user _kThe DRC that observes, u (f _k) rate request to satisfy, the DRC that is wherein observed is less than or equal to this user's average DRC data demand, and is as follows:

req_rate(f _k)≤Avg(DRC(u(f _k)))。(15)

Carry out the calculating AS (t) of ASF immediately, and use it for and be calculated as follows Avail (t):

$\mathrm{Avail}\left(t\right)=\max \mathrm{imum}(\frac{L_{\max }}{\mathrm{AS}\left(t\right)}-\mathrm{Res}\left(t\right),0)---\left(16\right)$

At last, for flow process f _kPermission get into decision-making and consider at time t:

req_rate(f _k)≤Avail(t)。(17)

If flow process f _kBe allowed to get into, then as shown in Figure 14 resource measurement press the following formula renewal:

Res(t)＝Res(t)+req_rate(f _k) 18)

Avail(t)＝Avail(t)-req_rate(f _k) (19)

AN continue as the flow process monitoring qos statistic that allow to get into to some extent, and the relevant statistic of monitoring network.Keep watch on to adapting to the reservation factor feedback is provided.

The adaptive approach of AS (t): the QoS of each sector and network statistics

Figure 18 A provides a kind of to 18E and supports the permission of the system of a plurality of application flows with qos requirement to get into the flow chart of the method 300 of control.In Figure 18 A,, use in step 304 and to allow to get into control program when when judging that diamond 302AN receives the request to new technological process.Otherwise this process is waited for new flow process request.During between noting at this moment, AN continues to keep watch on current operational circumstances, carries out the qos statistic amount of current effective flow process and the network statistics of flow process.Allow to get into control program and determined whether that resource can be in order to support this new technological process.Resource measurement shown in Figure 14 obtains upgrading in step 305.If judging diamond 306, new flow process is allowed to get into, and then handles advancing to 307 with the application self-adapting scheduling process.

The permission of step 304 gets into control program and in Figure 18 B, further details.Judging diamond 308, if flow process f _kRequest rate greater than flow process f _kAverage DRC data demand, then handle advancing to step 312 with refusal flow process f _kGet into.Otherwise, handle to return and judge that diamond 312 is to confirm flow process f _kRequest rate whether greater than available resources Avail at time t.If institute's rate request less than Avail, is allowed to get in this flow process of step 314 so, otherwise in this flow process of step 312 refusal.

Being updated among Figure 18 C of step 305 pair resource measurement further detailed.In step 320, resource measurement Avail and Res obtain upgrading.Obtain upgrading and keeping watch in step 322QoS statistic.In step 324, ASF obtains upgrading based on the result of step 320 and 322.In step 326, the hierarchical resource L of estimation is recomputated.If have new flow process to be asked, then handle and return to handle in the step 304 of Figure 18 A in step 328.If do not have new flow process to be asked, confirm the existence of user in this sector in step 330 for each user so in step 328.Confirm that in step 332 sampling continues the cycle.

Continuing Figure 18 D is discussed, is every process selecting QSG parameter in step 340.Consider two parallel processing paths, wherein first path step 342 speed on the sampling interval processing speed violate.Notice that this speed sampling interval is greater than the sampling duration of calculating gained in step 332.The processing of each effective flow process is specialized in second path.For given flow process, this process confirms to have at the sampling duration ratio of the IP grouping that postpones violation in step 344.In step 346, the ratio that the IP that has shake to violate for the flow process calculating sampling duration of being considered divides into groups.Calculate mark in step 348 immediately at the used time slot of sampling duration flow process.Confirm the mark of time slot that has the flow process of qos requirement at the sampling duration in this process of step 350.In step 352, this process is violated with regard to QoS and is checked, and confirms the QoS packet ID in step 354.

Processing advances to Figure 18 E, and step 360 is the quantity of each QoS grouping calculation process.Calculate mark immediately in this process of step 362 corresponding to the QoS flow process of each qos statistic amount.Immediately in result and the predetermined threshold of step 364 with step 360 and 362.Notice that threshold value can be during operation by dynamic renewal.Adjust ASF in view of the above in step 366.

Figure 18 A provides an embodiment who allows entry control method to 18E.Other details that allows entry control method is discussed hereinafter.Such as AN elements such as BTS is the statistic that every flow process is collected each sector, and this information is used for the permission entering control and the preemption algorithm of each sector.Only corresponding to the user of flow process in this subregion during collect the statistic of each sector.BTS periodically collects the QoS statistic relevant with network.Make T collect these information that time cycle before, wherein Z is a sample index, t=Z*T.

Consider the flow process f among the s of sector _k, u (f wherein _k) be corresponding to flow process f _kThe user.This user is at time t _EnterGet into sector s.S is inner in the sector, at duration [t _Reserve(f _k, s), t _Free(f _k, s)] in flow process reservation of resource for this reason, wherein resource is at time t _ReserveObtain keeping.When user's request applications was served, resource was retained.User u (f _k) in the above duration, at t _{Enter, j}(f _k, s) get into sector s the j time, and at t _{Leave, j}(f _k, s) leave for the j time.Therefore, t _Reserve(f _k, s)≤t _{Enter, first}(f _k, s) and t _Free(f _k, s)>=t _{Leave, last}(f _k, s).Notice that from estimating that this user possibly move on to this sector at certain following time point, can be apprised of via qos signaling agreement AN to be possible for the flow process reservation of resource.Here, t _{Enter, first}(f _k, s) be user u (f _k) get into the time of this sector s the first time, and t _{Leave, last}(f _k, be s) at flow process f _kLifetime during this user leave time of sector s for the last time.

Only when three conditions all were met below time t, this algorithm was included consideration in the time t performance statistics that QoS is relevant with network:

t _reserve(f _k，s)≤t≤t _free(f _k，s) (20)

t _{Entet_latest}(f _k, s)+δ (f _k)≤t≤t _{Leave_latest}(f _k, s) and (21)

IN_IP_PKTS(f _k，t，s)>θ(f _k) (22)

Wherein, δ (f _k) and θ (f _k) be to be flow process f _kPreassigned, IN_IP_PKTS (f _k, t is at time period (max (t-T, t in the s of sector s) _{Enter_latest}(f _k, s)), t) during by BTS forward link schedulers scheduling flow process f for transmission _kThe input IP number of dividing into groups.If last position that IP divides into groups is transmitted in the s of sector, then this is counted the IN_IP_PKTS of this sector.Variable δ (f _k) refer to that the appearance in this sector is regarded as the important time before.In other words, in case the user has stopped δ second in the s of sector, then this process begins resource is carried out valuation.Notice that in case obtain allowing, the user need not require to allow again to leave at once and get into this sector again.

The flow process that Qos and network statistics are used for the service of each request applications immediately allows the valuation of entry criteria.At time period (max (t-T, t _{Enter_latest}(f _k, s)), t) during among the s of sector corresponding to flow process f _kDelay IP divide into groups to be calculated as follows:

$\mathrm{Frac}\_\mathrm{delayed}\_\mathrm{IP}\_\mathrm{Pkts}(f_k,t,s)$

$=\frac{\mathrm{DELAYED}\_\mathrm{IP}\_\mathrm{PKTS}(f_k,t,s)}{\mathrm{IN}\_\mathrm{IP}\_\mathrm{PKTS}(f_k,t,s)}---\left(23\right)$

DELAYED_IP_PKTS (f wherein _k, t s) corresponds to till the time t number that IP that in the sector BTS place of s postponed to surpass the corresponding delay border of flow process fk divides into groups.In case in the s of sector, detect the delay violation that IP divides into groups, promptly cumulative this sector postpones the counting of violation.Flow process f till the time t _kHave the right mark of IP grouping of shake boundary violation to be calculated as follows:

$\mathrm{Frac}\_\mathrm{jitter}\_\mathrm{viol}(f_k,t,s)=\frac{\mathrm{JTR}\_\mathrm{VIOL}\_\mathrm{PKT}\_\mathrm{PAIRS}(f_k,t,s)}{\mathrm{IN}\_\mathrm{IP}\_\mathrm{PKTS}(f_k,t,s)-1},---\left(24\right)$

JTR_VIOL_PKT_PAIRS (f wherein _k, t s) corresponds to flow process f till the time slot t _kThe quantity that (IP divides into groups in succession), it is right that IP divides into groups that the shake boundary violation is arranged.When two of a flow process in succession IP be grouped in when transmitting on the sector, for carrying out this counting in this sector.

At time period (t _{Enter_latest}(f _k, s), t) during flow process f _kSpeed violate,

$\mathrm{Rate}\_\mathrm{viol}(f_k,t,s)=\frac{\mathrm{req}\_\mathrm{rate}\left(f_k\right)-\mathrm{served}\_\mathrm{rate}(f_k,t,s)}{\mathrm{req}\_\mathrm{rate}\left(f_k\right)}---\left(25\right)$

Req_rate (f wherein _k)>Served_rate (f _k, t, s).Otherwise as flow process f _kWhen time t does not have speed to violate, calculate at (t _{Enter_latest}(f _k, s), t) flow process f among the s of time durations sector _kServed_rate (f _k, t, s).

Notice that violate for speed, this process is with time period (t _{Enter_latest}(f _k, s), t) to use, but violate for delay and jitter as the sampling duration, this process is with time period (max (t-T, t _{Enter_latest}(f _k, s)), t) use as the sampling duration.

At time period (max (t-T, t _{Enter_latest}(f _k, s)), t) during flow process f _kEmployed time slot mark is calculated as follows:

$\mathrm{Frac}\_\mathrm{slots}\_\mathrm{flow}(f_k,t,s)=\frac{\mathrm{SERVED}\_\mathrm{SLOTS}(f_k,t,s)}{\mathrm{IN}\_\mathrm{SECTOR}(f_k,t,s)}---\left(26\right)$

SERVED_SLOTS (f wherein _k, t is by (max (t-T, t s) _{Enter_latest}(f _k, s)), t) flow process f during the time period of definition _kThe quantity of the time slot of being served, IN_SECTOR (f _k, t s) is (max (t-T, t _{Enter_latest}(f _k, s)), t) during the sum of time slot.The time period (t-T, the mark of time slot that gives to have among the s of sector the flow process of qos requirement during t) is provided by following formula:

The dynamic flow classification

For every flow process f _k, below four threshold values be preassigned, and be used to carry out the QoS inspection relevant of every flow process with channel situation:

Frac_delayed_IP_pkts_thres (f _k), (fractional delay IP packet threshold (f _k))

Frac_jitter_viol_IP_pkts_thres (f _k), (IP packet threshold (f is violated in the mark shake _k))

Rate_viol_thres (f _k), (speed is violated threshold value (f _k)) and

Frac_slots_viol_thes (f _k), (the mark time slot is violated threshold value (f _k)).

System periodically adapts to ASF after each time T, wherein T is preassigned value.T ought the Z time (that is, when t=Z*T) carrying out the self adaptation inspection, those flow processs that have some resources to obtain keeping among this process consideration sector s promptly, satisfy t in preset time _Reverse(f _k, s)≤t≤t _Free(f _k, flow process s).For each the bar flow process f that comes to organize since then flow process _k, check so that following " the threshold value inspection of every flow process " carried out valuation:

Frac_delayed_IP_Pkts(f _k，t，s)>Frac_delayed_IP_pkts_thres(f _k) (28)

(fractional delay IP (the f that divides into groups _k, t, s)>Fractional delay IP packet threshold (f _k))

Frac_jitter_viol_IP_pkts(f _k)>Frac_jitter_viol_IP_pkts_thres(f _k，t，s)(29)

(IP grouping (f is violated in the mark shake _k)>IP packet threshold (f is violated in the mark shake _k, t, s))

Rate_viol(f _k，t，s)>rate_viol_thres(f _k) (30)

(speed is violated (f _k, t, s)>Speed is violated threshold value (f _k))

Frac_slots_flow(f _k，t，s)>Frac_slots_viol_thes(f _k) (31)

(mark time slot flow process (f _k, t, s)>The mark time slot is violated threshold value (f _k))

If satisfy following two conditions, then flow f _kCarry out above four inspections:

t _{Enter_latest}(f _k, s)+δ (f _k)≤t≤t _{Leave_latest}(f _k, s), and (32)

IN_IP_PKTS(f _k，t，s)>θ((f _k)) (33)

When for a flow process, at least one in condition (32), (33) do not satisfy, but t _Reserve(f _k, s)≤t≤t _Free(f _k, s) being true time, the result queue of the threshold value of every flow process being checked for this flow process is NA.

This process is calculated at the sampling duration; The mark of the time slot that uses for flow process with qos requirement; Also calculate the threshold value of this fractional value, wherein Frac_slot_thres_qos_flows (s) is the upper threshold value of mark of in time period T, distributing to the time slot of the flow process with qos requirement among the s of sector.The value of Frac_slot_thres_qos_flows (s) is used for checking whether following formula is set up:

frac_slots_qos_flows(t，s)>Frac_slots_thres_qos_flows(s)。(34)

This process is categorized into one of following QoS dispatch group (QSG) with current flow process:

QSG I or Q_DJR: have the flow process of delay, shake and rate requirement,

QSG II or Q_RavgD: have the flow process of speed and average retardation requirement,

QSG III or Q_R: flow process with rate requirement.

Consider one group of flow process that belongs to the Q_DJR class.If given flow process f _kFor the NA classification is underproof, and or satisfy:

Frac_delayed_IP_Pkts(f _k，t，s)>Frac_delayed_IP_pkts_thres(f _k)， (35)

Perhaps satisfy

Frac_jitter_viol_IP_pkts(f _k，t，s)>Frac_jitter_viol_IP_pkts_thres(f _k)，(36)

This flow process has been designated as delay_or_jitter_viol (f _k, t, s)=((f is violated in delay or shake to Y _k, t, s)=be).Otherwise this process flow process is for this reason established delay_or_jitter_viol (f _k, t, s)=((f is violated in delay or shake to N _k, t, s)=not).On the other hand, if in the NA classification, be qualified, delay_or_jitter_viol (f so _k, t, s)=NA.

Table 2

QoS state group ID (QS_GID)	The QoS class	delay_or_jitterviol(f k，t，s)	rate_viol(f k，t，s)>rate_viol_thres(f k)
				1	Be used for the Q_DJR flow process	Be	Be
2	Be used for the Q_DJR flow process	Be	Not
				3	Be used for Q_DJR, Q_RavgD, Q_R flow process	Not	Be
4	Be used for Q_DJR, Q_RavgD, Q_R flow process	Not (or NA)	Not (or NA)

At each time t, when carrying out the self adaptation inspection of AS (t), this process is as shown in table 2 classifies to QoS flow process (that is the flow process that, has qos requirement).Every flow process all is assigned with a QoS state group ID (QS_GID).

QS_GID=1: the flow process that the Q_DJR class of speed and delay (or shake) violation is arranged.

QS_GID=2: have delay (or shake) to violate and do not have a flow process of the Q_DJR class that speed violates.

QS_GID=3: do not have delay and jitter to violate but the flow process of the Q_DJR class that speed violates is arranged.

Possibly produce this situation for the self adaptation application program.And the flow process corresponding to Q_R and Q_RavgD class that has speed to violate is assigned in this group.

QS_GID=4: the flow process that does not have QoS (speed, delay and jitter) to violate.Flow process in the NA classification as indicated above also is placed in this group.

The self adaptation of reservation factor

Make N _k(t is in the flow process number of time t corresponding to QSG k s), N (t, s) the flow process sum in the s of sector, there are some resources to obtain keeping at time t.

$N(t,s)=\underset{k}{\mathrm{\&Sigma;}}{N}_k(t,s)---\left(37\right)$

Make M represent QoS state group ID (QS_GID), the result is:

$N_{Q\_\mathrm{DJR}}\left(t\right)=\underset{M=1}{\overset{4}{\mathrm{\&Sigma;}}}{N}_{Q\_\mathrm{DJR}}(t,M)---\left(38\right)$

$N_{Q\_\mathrm{RAvgD}}\left(t\right)=\underset{M=3}{\overset{4}{\mathrm{\&Sigma;}}}{N}_{Q\_\mathrm{RAvgD}}(t,M)---\left(39\right)$

$N_{Q\_R}\left(t\right)=\underset{M=3}{\overset{4}{\mathrm{\&Sigma;}}}{N}_{Q\_R}(t,M)---\left(40\right)$

, in QoS state group M, provide by following formula at time t (in the s of sector) corresponding to the mark of the flow process of the QSG k of flow process k:

$F_k(t,M,s)=\frac{N_k(t,M,s)}{N_k(t,s)}---\left(41\right)$

Have the mark of the flow process of delay (or shake) and speed violation to provide at time t by following formula:

Frac_flows_DJR_viol(t，s)＝F _{Q_DJR}(t，M＝1，s) (42)

There is delay (or shake) to violate at time t, but do not have the mark of the flow process of speed violation to provide by following formula:

Frac_flows_DJ_viol(t，s)＝F _{Q_DJR}(t，M＝2，s) (43)

Only have the mark of the flow process of speed violation to provide at time t by following formula:

$\mathrm{Frac}\_\mathrm{flows}\_R\_\mathrm{only}\_\mathrm{viol}(t,s)=$

$\frac{N_{Q\_\mathrm{DJR}}(t,M=3,s)+N_{Q\_\mathrm{RAvgD}}(t,M=3,s)+N_{Q\_R}(t,M=3,s)}{N_{Q\_\mathrm{DJR}}(t,s)+N_{Q\_\mathrm{RAvgD}}(t,s)+N_{Q\_R}(t,s)}---\left(44\right)$

Not having QoS to violate (or in NA classification) flow process mark is provided by following formula:

$\mathrm{Frac}\_\mathrm{flows}\_\mathrm{no}\_\mathrm{na}\_\mathrm{viol}(t,s)=$

$\frac{N_{Q\_\mathrm{DJR}}(t,M=4,s)+N_{Q\_\mathrm{RAvgD}}(t,M=4,s)+N_{Q\_R}(t,M=4,s)}{N_{Q\_\mathrm{DJR}}(t,s)+N_{Q\_\mathrm{RAvgD}}(t,s)+N_{Q\_R}(t,s)}---\left(45\right)$

The self adaptation of S (t) is periodically carried out after each time period T, and T is preassigned.From adaptive purpose, can consider above each group as follows.

Table 3

QSG	The mark of the flow process of QS_GID=1	The mark of QID=2	The mark of QID=3	The mark of QID=4
					QSG_DJR	Frac_flows_DJR_viol(t，s)	Frac_flowsDJ_viol(t，s)	(t s) combines with Frac_flows_R_only_viol	(t s) combines with Frac_flows_no_na_viol
QSG_RavgD	Inapplicable	Do not use	(t s) combines with Frac_flows_R_only_viol	(t s) combines with Frac_flows_no_na_viol
					QSG_R	Inapplicable	Inapplicable	(t s) combines with Frac_flows_R_only_viol	(t s) combines with Frac_flows_no_na_viol

With lower threshold value is preassigned, and uses in the adaptive approach shown in following.

Frac_flows_thres_DJR: have on the mark of the delay (or shake) and the flow process of speed violation

Threshold value

Frac_flows_thres_DJ: have delay (or shake) to violate and do not have the threshold value on the mark of the flow process that speed violates

Frac_flows_thres_R: have speed to violate the threshold value on the mark of flow process of (and do not have postpone or shake is violated)

Frac_flows_thres_ok_qos: the threshold value on the mark of the flow process that no QoS violates

D is after carrying out the self adaptation inspection for AS (t), this process continues to carry out by following order immediately:

Step 1: if Frac_flows_DJR_viol (t, s) >=Frac_flows_thres_DJR:

AS (t ⁺)=f _Qos* AS (t)+x _QosThereby, AS (t ⁺)>=AS (t).Here, f _QosAnd x _QosBe preassigned.Otherwise,

Step 2: if Frac_flows_DJ_viol (t, s) >=Frac_flows_thres_DJ:

AS (t then ⁺)=f _{Delay_jitter}* AS (t)+x _{Delay_jitter}Thereby, AS (t ⁺)>=AS (t).Here, f _{Delay_jitter}And x _{Delay_jitter}Be preassigned.Otherwise,

Step 3: if Frac_flows_R_only_viol (t, s) >=Frac_flows_thres_R:

AS (t then ⁺)=f _Rate* AS (t)+x _RateThereby, AS (t ⁺)>=AS (t).Here, f _RateAnd x _RateBe preassigned.Otherwise,

Step 4: if Frac_flows_no_na_viol (t, s) >=Frac_flows_thres_ok_qos:

AS (t then ⁺)=f _{All_qos_flows}* AS (t)+x _{All_qos_flows}Thereby, AS (t ⁺)>=AS (t).Here, f _{All_qos_flows}And x _{All_qos_flows}Be preassigned.Otherwise,

Step 5: if Frac_flows_no_na_viol (t, s) >=Frac_flows_thres_ok_qos, and if Frac_slots_qos_flows (t, s) Frac_thres_slots_qos_flows:

AS (t then ⁺)=f _Ok* AS (t)+x _OkThereby, AS (t ⁺)≤AS (t).Here, f _OkAnd x _OkBe preassigned.Otherwise,

Step 6: if Frac_flows_no_na_viol (t, s) >=Frac_flows_thres_ok_qos, and if Frac_slots_qos_flows (t, s) >=Frac_thres_slots_qos_flows:

AS (t then ⁺)=AS (t).

The preemption scheme

Figure 19 illustrates a kind of preemption method 400 according to an embodiment.Method 400 is by judging whether diamond 402 definite ASF obtain increasing and beginning.When detecting the increase of ASF, handle advancing to step 404 to define the flow process of flank speed violation number of times.In other words, when ASF increased, preemption method 400 began to identify those flow processs of wanting preemption.In the present embodiment, has the optimal candidate that flow process that speed violates is represented as preemption.Alternative embodiment can give other flow process priority, and can dynamically change priority scheme.

If judging that diamond 406 reaches the maximum P of preemption _MAX(following detailed description) then handles advancing to step 408 and violate the flow process of number of times to occupy high latency earlier.Otherwise handle to return and judge diamond 402.When after flow process of step 408 preemption, handle to advance to and judge that diamond 410 is to determine whether a plurality of flow processs that high latency is violated number of times that have.For many flow processs, handle advancing to step 412 is used maximum time slots with preemption flow process.Usually, this flow process will have very low data transfer rate, and therefore consume maximum time slots in preset time during the section.Handle to return immediately and judge diamond 402.

In one embodiment, use this preemption method 400, wherein P_max puts to allow preoccupied maximum flow process number at any given time.The flow process subclass of the condition of two preemption groups shown in the table 3 is satisfied in consideration.Particularly, preemption group 1 comprises the flow process that belongs to QSG_R or QSG_RavgD, and has

Rate_viol (f _k, t, s)>Rate_viol_thres ((f _k) and (46)

Frac_slot_flow(f _k，t，s)>frac_thres_slots_flow(f _k)。(47)

Preemption group 2 comprises the flow process that belongs to Q_DJR QSG, and said flow process contains

Frac_delayed_IP_Pkts (f _k, t, s)>Frac_delayed_IP_pkts_thres (f _k) and (48)

Frac_slots_fow(f _k，t，s)>frac_thres_slots_flow(f _k)。(49)

Table 4

Preemption group ID	QSG	QoS violates	The time slot threshold-violating
				1	Q_R，Q_RavgD	rate_viol(f k，t，s)>rate_viol_thres((f k)	Frac_slot_flow(f k，t，s)>frac_thres_slots_flow(f k)
2	Q_DJR	Frac_delayed_IP_Pkts(f k，t，s)>Frac_delayed_IP_pktsthres(f k)	Frac_slots_flow(f k，t，s)>frac_thres_slots_flow(f k)

Step 1: if at certain time point, AS (t) obtains increasing (as in the adaptive approach of AS (t)), and then this process checks that whether one or several flow processs are arranged is qualified preemptions.Note, not having flow process by preemption as AS (t) when not obtaining increasing.

Step 2: consider flow process subclass corresponding to preemption group 1.Outside these flow processs, select to have the P_max bar flow process of maximum rate_viol value.If draw is arranged, then preemption has the flow process of big Frac_slots_flow value.If P_max bar flow process by preemption, is so no longer violated the more flow process of preemption for speed.

Step 3: consider flow process subclass corresponding to preemption group 2.These flow processs have the delay and jitter requirement, and Frac_delayed_IP_Pkts (f is arranged _k, t, s)>Frac_delayed_IP_pkts_thres (f _k) and Frac_slots_flow (f _k, t, s)>Frac_thres_slots_flow (f _k).Outside these flow processs, select P_max to deduct the flow process number with maximum Frac_delayed_IP_Pkts of preemption in the step 2.If draw is arranged, then preemption has the flow process of big Frac_slots_flow value.

With the QoS between indoor and user

A mobile subscriber possibly have many flow processs simultaneously, that is, and and a plurality of application programs.As given among this paper, the user can specify the following:

Every flow process about its whether responsive indication to delay and jitter.If it is responsive to delay and jitter, then want specified delay and shake border.

Each user's general objective speed (ATR).This is the targeted rate that the scheduler of forward link hierarchy is intended to give this user.

Allow to get into control

Give the individual user of R (t) who fixes on time t, be designated as U ₁, U ₂..., U _{R (t)}, with num_flows (U _j, t) see U as time t user _jThe flow process number.Suppose, at time t user U _jThere is (k-1) bar flow process to be allowed to get into, i.e. num_flows (U _j, t)=k-1.For decision allows user U at time t _jA new technological process f _{K, j}Get into, this process is used user U _jThe DRC that is observed check below:

$\underset{c=1}{\overset{k}{\mathrm{\&Sigma;}}}\mathrm{req}\_\mathrm{rate}\left(f_{k,j}\right)\&le;\mathrm{Avg}\left(\mathrm{DRC}\left(U_j\right)\right).---\left(50\right)$

When carrying out the self adaptation of foregoing ASF, the timeslot number that flow process is got is also included in consideration with its corresponding DRC.This process is calculated as follows AS (t) and Avail (t):

$\mathrm{Avail}\left(t\right)=\max \mathrm{imum}(\frac{L_{\max }}{\mathrm{AS}\left(t\right)}-\mathrm{Res}\left(t\right),0)---\left(51\right)$

If following formula is set up, then this process allows flow process f at time t _{K, j}Get into:

req_rate(f _k，j)≤Avail(t) (52)

If this flow process is allowed to get into, then do following the renewal:

Res(t)＝Res(t)+req_rate(f _k，j)， (53)

Avail(t)＝Avail(t)-req_rate(f _k，j)， (54)

ATR(U _j，t)＝ATR(U _j，t)+req_rate(f _k，j)， (55)

This process continues as all flow processs that are allowed to get into and user's monitoring qos statistic, and the relevant statistic of monitoring network.Use these information to continue to adapt to the reservation factor.Calculate and use ASF AS (t) immediately.

The scheduler of hierarchy

Every flow process and each user's compensation:

The responsive flow process of each delay and jitter all is assigned with a dithering threshold.For user U _kEach flow process fx responsive to delay and jitter, this process is calculated corresponding delay and jitter compensation Ф.If having no, flow process surpasses the grouping that postpones threshold value in a formation, so

φ(fx(U _k))＝1。(56)

Otherwise, calculate

$\mathrm{\&phi;}\left(\mathrm{fx}\left(U_k\right)\right)=C_{\mathrm{delay}}\left(\mathrm{fx}\left(U_k\right)\right)*\frac{\mathrm{ndefpkts}\left(\mathrm{fx}(U_k,n)\right)}{{\mathrm{ndefpkts}}_{\min }\left(n\right)}---\left(57\right)$

Here,

$\mathrm{ndefpkts}(\mathrm{fx}(U_k,n)=\frac{\mathrm{defpkts}\left(\mathrm{fx}(U_k,n)\right)}{\mathrm{req}\_\mathrm{rate}\left(\mathrm{fx}\left(U_k\right)\right)}---\left(58\right)$

For user U _kEvery flow process fx,

Ndefpkts _Min(n)=minimum value (59) of ndefpkts

Consideration is in (all users') all flow processs of time slot n,

Defpkts (fx (U _k, n)=at time slot n user U _kFlow process fx violate the number that its unsettled MAC that postpones threshold value divides into groups.(60)

For user U at time slot n _kRate compensation, the definition,

$\mathrm{\&alpha;}(U_k,n)=\frac{\mathrm{ATR}\left(U_k\right)}{\mathrm{ASR}(U_k,n_{\mathrm{prev}}\left(n\right))}---\left(61\right)$

Here,

ASR (U _k, n _Prev(n))=time slot n _PrevMiddle user U _kTotal service speed, and (62)

n _prev≤n。(63)

Timeslot number n _PrevBe that n goes up or last time slot before when keeping watch on speed from the purpose of dispatching algorithm.

This process is user U _kBe defined in the total delay compensation of any time slot n.Consider that this user has all flow processs that postpone requirement, and check that file leader (HOL) MAC of the flow process of each these delay-sensitive divides into groups.If have no one in system the time of staying surpass it and postpone threshold value, then:

agg_delay_comp(U _k，n)＝1。(64)

Otherwise the user considers to have HOL to be grouped in the system time of staying to be longer than the flow process subclass that postpones threshold value for this reason.For this user,, HOL MAC is grouped in the user U that lingering period in the system surpasses the delay threshold value of these flow processs for being arranged _kAll flow process fx, calculate:

$\mathrm{agg}\_\mathrm{delay}\_\mathrm{comp}(U_k,n)=\underset{x}{\mathrm{\&Sigma;}}w\left(\mathrm{fx}\left(U_k\right)\right)*\mathrm{\&phi;}(\mathrm{fx}\left(U_k\right),n),---\left(65\right)$

Here, w (fx (U _k)) be to distribute to user U _kThe initial weight of flow process fx.

Adaptive weight calculates

For user for this reason calculates adaptive weight, this process postpones the inspection of threshold-violating at each time slot for each has each bar flow process of the user of a delay-sensitive flow process at least.This process is each such user U _kCalculate:

agg_delay_comp(U _k，n) (66)

If agg_delay_comp is (U _k, n)>1, then this process is user U as follows _kCalculate adaptive weight:

aw ^t(U _k，n)＝agg_delay_comp(U _k，n)*aw ^t(U _k，n-1) (67)

On the other hand, if agg_delay_comp is (U _k, n)=1, then this process is calculated:

aw ^t(U _k，n)＝α(U _k，n _prev(n))*w(U _k) (68)

Here, n _Prev, k (n) is as ASR (U _k, n _Prev(n)) (and so a (U comes under observation _k, n _Prev(n)) by being calculated) time on n or last time slot before.This process continue by following formula for this reason the user calculate final adaptive weight:

aw(U _k，n)＝Z(U _k，n)*aw ^t(U _k，n) (69)

Here, if the user is U _kHas no grouping in the RTx of any delay-sensitive flow process or the DARQ formation, then Z (U _k, n)=1.Otherwise, Z (U _k, n)=C (U _k).Here, C (U _k) be preassigned constant.

User and process selecting method

This process has at least the user of a grouping to calculate the tolerance among the following time slot n in its formation for each,

Y(U _k，n)＝aw(U _k，n)*DRC(U _k，n)/T(U _k，n) (70)

Here, $T(U_k,n)=\underset{\mathrm{Fz}}{\mathrm{\&Sigma;}}T(\mathrm{Fz}(U_k,n)$ Be user U _kAverage service rate (that is, comprising the flow process that all are corresponding).This process selects to have maximum Y (U _k, the n) user of value.In case selected a user with this scheduler, this process is this user and selects a stream, serves according to each following scheme.

Consider to assign to the flow process in following each group:

Group 1:QSG_ delay _ shake.The VoIP flow process.

Group 2:QSG_ delay _ shake.The video conference flow process.

Group 3:QSG_ delay _ shake.The flow process of video flowing.

Group 4:QSG_ speed _ average _ postpone.Flow process with speed and average retardation requirement.

Group 5:QSG_ speed.The flow process that only has rate requirement.

In carrying out dispatching algorithm described herein, can follow following steps.

Step 1: the flow process of considering all reserve of selected user in this time slot.

Step 2: consider the flow process of this user corresponding to group 1 and 2.Selecting HOL to divide into groups to have violated it postpones threshold value and postpones a flow process on border near this flow process.If find flow process, promptly serve this flow process.Otherwise, advance to step 3.

Step 3: consider corresponding to group 3 and the HOL flow process that postpones threshold value of dividing into groups to surpass wherein, and select wherein HOL to divide into groups near the flow process that postpones the border.If find flow process, promptly serve this flow process.Otherwise, advance to next procedure.

Step 4: consider corresponding to group 4 and the HOL flow process that postpones threshold value of dividing into groups to surpass wherein, and select wherein HOL to divide into groups near the flow process that postpones the border.If find flow process, promptly serve this flow process.Otherwise, advance to next procedure.

Step 5: select a reserve flow process 1 to 4 and serve from organizing.Flow process with minimum group number is with priority.If selected flow process, then serve it.Otherwise, advance to next procedure.

Step 6: consider the reserve flow process of this user corresponding to group 5.Selection has that of maximum institute request rate/service speed value.Serve this flow process.If have no one to be selected, then advance to next procedure.

Step 7: the best effort type flow process of serving this user.If have more than one, selection has that of minimum service rate value.

Figure 16 illustrates a kind of two-level scheduler device of using every flow process and each user's compensation.Scheduler shown in Figure 16 and Figure 17 is between the user who is used for described in this paper and with the scheduler of the hierarchy of indoor QoS compensation.As shown in Figure 16, the first order that is expressed as grade 1 comprise a plurality of scheduling elements or node S1, S2 ..., SM, wherein different QSG of each node processing.In this example, M is the QSG packet count that will handle.For example, the application flows of scheduling node S1 process IP phone (VoIP) type.Although provide IP phone as an example, the application flows of any QSG1 of being categorized into will obtain at the S1 place handling.These flow processs have specifies the delay and jitter border that is used to assess qos requirement.The flow process of a plurality of application program IP phone types obtains handling at scheduling element S1 place.Similarly, each scheduling element is the QSG handling process of an appointment.Notice that alternative embodiment can provide the application flows of a plurality of QSG to a scheduling element.Note, handle same QSG group and can use a plurality of scheduling elements.

The grade I of the scheduler shown in Figure 16 calculates the part of the compensation of every flow process.A plurality of application flows of each user are shown in Figure 16.Grade II scheduling element is accomplished the calculating of the compensation of every flow process.

Figure 17 illustrates has scheduling node S1, S2..., SzScheduler.In this step, z is user's a number.Notice that user's number is dynamic, so the number of current scheduling node can dynamically change.Each scheduling node S1, S2..., SzAll apply flexibly in receiving many flow processs from a given user.Scheduling node S1For user 1 (U1) receives flow process F1 to Fk.Here k is the sum of the application flows of the current user of being 1 processing.Use is by the compensation of every flow process of grade I among Figure 16 and the calculating of grade II scheduler, and the grade II scheduler among Figure 17 is total user's compensation for each user calculates.Grade II scheduler is chosen in the user that will serve in the time slot according to above with regard to the described adaptive weighted DRC/T algorithm of user choosing method immediately.Grade II scheduler is selected between the weighted value that receives from grade I scheduler subsequently.As indicated, W (Uk) is an initial weight of distributing to user Uk, and ATR (Uk) is the general objective speed of user Uk.In case selected the user, promptly according to above-mentioned process selecting method, selected the flow process that to serve in this time slot for this user corresponding to this user's grade I scheduler.

A kind of forward link schedulers can allow each user in a time slot, to specify the price of bearing at given DRC value place for the flow process that will serve.In case specified price, the adaptive frame structure scheduler moves to satisfy the qos requirement of dissimilar application programs.Corresponding scheduling mechanism allows the service supplier increasing interests and satisfying between the qos requirement target of application program and seek good balance.This scheduling mechanism also provides terminal use's Dynamic Cost Control, and can be used for having the application program of speed and/or average retardation requirement or being used for streaming application, or the like.Embodiment provides a kind of price to select, and wherein every flow process is specified the price when its each time slot when serving.This price depends on the DRC value that the user asks for this flow process in this time slot.Flow process j (that is, the user of browsing process j) is ready that at time slot m the price of paying is designated as c [j, m, DRC [j, m]].DRC [j, m] is illustrated in this user's of service among the time slot m speed here.The user is designated price statically, such as being each DRC value designated price in advance.Perhaps, the user is designated price dynamically, such as during the lifetime of application program, changing price.This allows user that price is had control to a certain degree, with in response to the channel situation that changes and reach the QoS of expectation.The operator can use a such scheduler together with as between the user and with the given scheduler of indoor QoS.This allows the operator to specify at least two types pricing scheme.For between the user and with indoor QoS scheduler, the operator can specify static pricing scheme (based on the services level protocol), and allows the Dynamic Pricing scheme based on the scheduler of adaptive frame structure at the same time.The user can select different flow processs are used different schemes.

An embodiment will be divided into several frames the time, and violate the specified price of statistic and each user according to DRC value, qos requirement, QoS and be the decision-making that works as dispatcher of each time slot.Frame structure provides the order of the Subscriber Queue that should serve in taking turns basically.Network determines in the scheduling will take turns which bar flow process/user of service to reach desired target for this at given time slot every the wheel.Frame structure is promptly taken turns the order of service procedure at each, continue to change, and be called as the algorithm based on AFS.

To give a definition some symbols that use in the computational process are made an explanation.A given N formation (formation of every flow process) supposes that then its qos requirement is met if with speed r [j] service procedure j.Also be every flow process j preassignment initial weight w [j] and time scale ts [j].This process is intended to guarantee that for flow process j provides speed flow process j comes under observation at the time slot (that is, at each m*ts [j] time slot, wherein m is an integer) of the integral multiple of each time scale.

Make start [j] for begin to receive the time slot that consideration will be served at first in one takes turns as flow process j.When finishing to time slot z, system hopes to serve S [j, z]=individual bit of r [j] * (z-start [j]), wherein for certain integer m z=m*ts [j] is arranged.Use a kind of scheduling mechanism, this system can balance hope to distribute to the timeslot number of a given flow process and the bit number that hope is served for this stream.

In addition, other parameter that is used for the AFS scheduler is following:

Slots_alloc [j, n]: the timeslot number of distributing to formation (flow process) j at the n wheel.

Slot_served [j, n]: when the timeslot number of formation (flow process) j when the n wheel obtains serving.

S_r [j, n]: till during to the end of n wheel will be the bit number of flow process j service.

Round_len [n]: n wheel timeslot number length.

Round_len_thres: a length of taking turns is the upper bound with this threshold value.

B [n]: the tabulation of pool queue when the beginning of time slot n.

R _{Out_round}[j, n]: take turns bit number at n for formation j service by scheduler.

R _Out[j, n, g]: the bit number of during the time interval [n, g], serving, wherein g>=n for formation j.

The description that provides more than the use, the dificiency bit that begins formation j at the n wheel is provided by following formula:

$\mathrm{def}\_\mathrm{bits}\_r[j,n]=\max (S\_r[j,n-1]-\underset{k=1}{\overset{n-1}{\mathrm{\&Sigma;}}}{R}_{\mathrm{out}\_\mathrm{round}}[j,k],0),\&ForAll;n,\&ForAll;j---\left(71\right)$

When the dificiency bit is correct time, corresponding flow process falls behind in service, and will obtain compensation.On the other hand, the suffered extra-service of flow process does not catch a packet with not showing, but catchs a packet indirectly, because this flow process can not be compensated, and other flow process that in service, falls behind will be compensated.

In addition, the normalization dificiency bit at the beginning flow process j of n wheel is provided by following formula:

$\mathrm{ndef}\_\mathrm{bits}\_r[j,n]=\frac{\mathrm{def}\_\mathrm{bits}\_r[j,n]}{S\_r[j,n-1]},\&ForAll;j,\&ForAll;n.---\left(72\right)$

Dificiency time slot at the beginning formation j of n wheel is provided by following formula:

$\mathrm{def}\_\mathrm{slots}\_r[j,n]=\max (\underset{k=1}{\overset{n-1}{\mathrm{\&Sigma;}}}\mathrm{slots}\_\mathrm{alloc}[j,k]-\underset{k=1}{\overset{n-1}{\mathrm{\&Sigma;}}}\mathrm{slots}\_\mathrm{served}[j,k],0),\&ForAll;n,\&ForAll;j---\left(73\right)$

This procedure definition is following at the normalization dificiency time slot of the beginning formation j of n wheel:

$\mathrm{ndef}\_\mathrm{slots}\_r\left[j,n\right]=\frac{\mathrm{def}\_\mathrm{slots}\_r[j,n]}{\underset{k=0}{\overset{n-1}{\mathrm{\&Sigma;}}}\mathrm{slots}\_\mathrm{alloc}\left[j,k\right]},\&ForAll;j,\&ForAll;n.---\left(74\right)$

Make that lslot [n] is last time slot of n wheel, fslot [n] is first time slot of n wheel.Suppose that aw [j, n] expression n takes turns (self adaptation) weights of distributing to flow process j.The timeslot number of flow process j is distributed in this weights decision at the n wheel.

DRC value to the user asked sorts.Particularly, if DRC ₁[B, S] is superior to DRC ₂[B, S], (B/S) so ₁>(B/S) ₂Here B is the bit number of each grouping, and S is a timeslot number.

Take turns scheduling for AFS scheduler every, this process all is that each takes turns the state variable that provides more than the calculating, is that every flow process is calculated weights each beginning of taking turns subsequently, to take turns the time slot that distributes certain quantity to this flow process at this.For this purpose, this process is used a kind of adaptive weight computing mechanism, and uses a kind of each service rule of taking turns to take turns the calculating frame structure for each.

The adaptive weight computing mechanism

Make ndef_bits_r _{Thres, min}Threshold value for preassigned ndef_bits_r.This process n wheel begin define an array ndefbits_set [n] as follows:

ndefbits_set[n]＝{k：ndef_bits_r _k≥ndef_bits_r _thres，min}。(75)

Make S_I [n] be the flow process number in this array that begins at the n wheel.Similarly, ndef_slots_r _{Thres, min}It is the threshold value of predefined ndef_slots_r.This procedure definition is at the ndefslots_set [n] of the beginning of n wheel as follows:

ndefslot_set[n]＝{k：ndef_slots_r _k≥ndef_slots_r _thres，min}。(76)

Make S_II [n] be the flow process number in this array of beginning of n wheel.

For flow process j arbitrarily, in the beginning of n wheel, definition time slot corresponding penalty function is following:

If ndef_bits_r _j>=ndef_bits_r _{Thres, min}

Slot_comp [j, n]=slot_comp_I [j, n] * slot_comp_II [j, n] then, (77)

If ndef_bits_r _j<ndef_bits_r _{Thres, min},

Then, slot_comp [j, n]=1.(78)

Here,

If ndef_bits_r _j>=ndef_bits_r _{Thres, min},

Then $\mathrm{Slot}\_\mathrm{Comp}\_I[j,n]=\frac{\mathrm{Ndef}\_\mathrm{Bits}\_r_j}{\mathrm{Ndef}\_\mathrm{Bits}\_r_{\mathrm{Avg}}\left[n\right]},---\left(79\right)$

For all k:k ∈ ndefbits_set [n], $\mathrm{Ndef}\_\mathrm{Bits}\_r_{\mathrm{Avg}}\left[n\right]=\frac{\underset{k}{\mathrm{\&Sigma;}}\mathrm{Ndef}\_\mathrm{Bits}\_r_j}{S\_I\left[n\right]}---\left(80\right)$

If ndef_bits_r _j<ndef_bits_r _{Thres, min},

Slot_comp_I [j, n]=1 then.(81)

For every flow process j, define two threshold values, slot_comp_I _{Thres, min}[j] and slot_comp_I _{Thres, max}[j], so that:

$\mathrm{slot}\_\mathrm{comp}\_I_{\mathrm{thres},\min }\left[j\right]\&le;\mathrm{slot}\_\mathrm{comp}\_I[j,n]\&le;\mathrm{slot}\_\mathrm{comp}\_I_{\mathrm{thres},\max }\left[j\right],\&ForAll;j,\&ForAll;n---\left(82\right)$

It is right to use these threshold values to be communicated with adaptive weight computing mechanism described herein, can prevent any flow process in one takes turns, to consume a large amount of time slots partially, and this flow process that surpasses given restriction is not punished simultaneously.

If ndef_slots_r _j>=ndef_slots_r _{Thres, min},

Then $\mathrm{Slot}\_\mathrm{Comp}\_\mathrm{II}[j,n]=\frac{\mathrm{Ndef}\_\mathrm{Slots}\_r_j}{\mathrm{Ndef}\_\mathrm{Slots}\_r_{\mathrm{Avg}}\left[n\right]}---\left(83\right)$

If ndef_slots_r _j<ndef_slots_r _{Thres, min},

Slot_comp_II [j, n]=1 then.(84)

For every flow process j, define two threshold values, slot_comp_II _{Thres, min}[j] and slot_comp_II _{Thres, max}[j], so that:

$\mathrm{slot}\_\mathrm{comp}\_{\mathrm{II}}_{\mathrm{thres},\min }\left[j\right]\&le;\mathrm{slot}\_\mathrm{comp}\_\mathrm{II}[j,n]\&le;\mathrm{slot}\_\mathrm{comp}\_{\mathrm{II}}_{\mathrm{thres},\max }\left[j\right],\&ForAll;j,\&ForAll;n---\left(85\right)$

In each beginning of taking turns, flow process is divided into as in the table 5 given four groups.

Table 5

Group	ndef_bits_r flow≥ndef_bits_r thres，min	ndef_slots_r flow≥ndef_slots_r thres，min
			Group I	Be	Not
Group II	Not	Be
			Group III	Be	Be
Group IV	Not	Not

Take turns the beginning of n any one, belong to the flow process j of group II or IV, use slot_comp [j, n]=1 for each.Beginning at the n wheel belongs to the flow process j that organizes I for each, uses following formula:

slot_comp[j，n]＝slot_comp_I[j，n]。(86)

If belong to group III at the beginning flow process j of n wheel, then use following formula:

slot_comp[j，n]＝slot_comp_I[j，n]*slot_comp_II[j，n] (87)

This process is the adaptive weight of calculation process j immediately, and is wherein following at the adaptive weight of n wheel non-empty queue:

$\mathrm{aw}[j,n]=\mathrm{slot}\_\mathrm{comp}[j,n]*w\left[j\right]*\frac{r\left[j\right]}{r_{\min ,n}},\&ForAll;j\&Element;B\left[n\right]---\left(88\right)$

Here, r _{Min, n}=min{r [k]; K:k ∈ B [n] }.For every flow process j, definition threshold value aw _{Thres, max}[j] also guarantees with this threshold value:

$\mathrm{aw}[j,n]\&le;{\mathrm{aw}}_{\mathrm{thres},\max }\left[j\right],\&ForAll;j,\&ForAll;n.---\left(89\right)$

Next, these adaptive weights are applied to calculating timeslot number and the every length of taking turns that will distribute to every flow process, so that:

$\mathrm{sloc}\_\mathrm{alloc}[j,n]\&Proportional;\mathrm{aw}[j,n],\&ForAll;j,j:j\&Element;B\left[n\right]---\left(90\right)$

$\underset{j:j\&Element;B\left[n\right]}{\mathrm{\&Sigma;}}\mathrm{slot}\_\mathrm{alloc}[j,n]\&le;\mathrm{round}\_\mathrm{len}\left[n\right]\&le;\mathrm{round}\_\mathrm{len}\_\mathrm{thres},\&ForAll;n.---\left(91\right)$

Every scheduling rule of taking turns

In case distributed every flow process, then calculated some time slots and this length of taking turns in taking turns.Next step is to be chosen in the flow process that will serve in any given time slot.In any given time slot m of n wheel,, then need not to select the new technological process that to serve if the selected grouping of one of time slot is formerly still being served.On the other hand, if in this time slot m of n wheel, have no grouping to be served, then select the flow process that to serve as follows.Each time slot m of the new technological process that need select for scheduler to serve, and for each flow process j calculates the selection tolerance of time slot m of the n wheel of following flow process j,

$\mathrm{YY}[j,n,m]=c[j,m,\mathrm{DRC}[j,m]]*\mathrm{wait}\_\mathrm{comp}[j,n,m]*\mathrm{DRC}[j,m],\&ForAll;n,---\left(92\right)$

So that j ∈ B [n] and slots_served [j, n] θ (j) * slots_alloc [j, n], here, θ (j) be for every flow process j preassigned, wait_comp be give flow process with improve its postpone the border etc. to be compensated.Make that wait [j, n, m] is the stand-by period that the file leader at the beginning flow process j of the time slot m of n wheel divides into groups.

$\mathrm{wait}\_\mathrm{comp}[j,n,m]=\frac{\mathrm{wait}\left[j,n,m\right]}{\mathrm{wait}\_\mathrm{avg}[n,m]},\&ForAll;j,j:j\&Element;B_j---\left(93\right)$

And in the beginning of the time slot m of n wheel, flow process j has at least a grouping unsettled.Next, calculate for flow process k:

$\mathrm{wait}\_\mathrm{avg}[n,m]=\frac{\underset{k}{\mathrm{\&Sigma;}}\mathrm{wait}[k,n,m]}{\mathrm{wait}\_\mathrm{num}[n,m]},---\left(94\right)$

So that k ∈ B [n], and the number of these flow processs is wait_num [n, m].For every flow process j, to wait_comp _{Thres, min}[j] and wait_comp _{Thres, max}[j] these two threshold values are carried out assignment, and are used to guarantee

$\mathrm{wait}\_{\mathrm{comp}}_{\mathrm{thres},\min }\left[j\right]\&le;\mathrm{wait}\_\mathrm{comp}[j,n,m]\&le;\mathrm{wait}\_{\mathrm{comp}}_{\mathrm{thres},\max }\left[j\right],\&ForAll;m,\&ForAll;n,\&ForAll;j.---\left(95\right)$

Having the peaked flow process of selecting tolerance YY is selected thus the AFS scheduler and in any given time slot, serves.

According to the AN element of an embodiment shown in Figure 20.AN element 500 receives the application flows data and handles these data to feed to user's transmission.500 pairs of a plurality of application flows of AN element are dispatched, and wherein every flow process all has qos requirement.Notice that as indicated above, application flows can comprise best effort type flow process.AN element 500 comprises flow process taxon 502, is applicable to identification be associated with flow process traffic overview and QoS overview and with flow process type of being mapped to (or QSG) flow process.Flow process grader unit 502 is coupled to scheduler 504, allows to get into control unit 510 and QoS monitor 506.Scheduler 504 can be realized any in the various dispatching algorithms including, but not limited to equitable proportion (PF) algorithm and adaptive weighted PF algorithm.Allow getting into control unit 510 allows to get into controlling schemes and is applied to the received application flows of AN500 a kind of.Allow to get into control unit 510 based on QoS and network statistics, each new technological process of being asked is carried out valuation, and determined whether that enough resources can be in order to support this new technological process.Adaptive unit is coupled to and allows to get into control unit, and ASF obtains upgrading there.Adaptive unit 512 is applicable to the current effective application flows is carried out the preemption decision-making.The performance of a certain given flow process relevant data rate, used time slot and other QoS and network statistics flow process is considered in preemption.The QoS monitor is applicable to the qos requirement of the application flows that supervision receives.Notice that AN element 500 receives many flow processs usually, and between them, select to feed to user's transmission.Scheduler 504 receives the information that whether is allowed to get into about new technological process from allowing to get into control unit 510.Scheduler 504 receives qos statistic amount and out of Memory from QoS monitor 506, and wherein scheduler 504 is used these QoS information and selected the flow process transmitted.

Given among this paper is that the permission that in wireless communication system, is used for application flows gets into the method and apparatus of control, preemption and scheduling.Allow to get into the data transfer rate that control considers that new technological process is asked, and this is compared with available resources.In case be allowed to get into, flow process promptly is provided for scheduler, scheduler is applicable to the analysis of carrying out every flow process and each user, is chosen in the user of transmitting in the time period of each time slot or appointment.

Those skilled in the art can understand, and information and signal can be represented with various method and technology.For example, running through the citable data of above description, instruction, order, information, signal, bit, symbol and chip can be represented by voltage, electric current, electromagnetic wave, magnetic field or magnetic particle, light field or light particle or its combination in any.

In addition, those skilled in the art also will understand, and can be implemented as electronic hardware, computer software or both combinations in conjunction with the described various illustrative logical blocks of each embodiment that is explained among this paper, module, circuit and algorithm steps.For this interchangeability of hardware and software clearly is described, more than with regard to its function general description various example components, frame, module, circuit and step.This type of function be embodied as hardware still be software be depend on the certain applications program with the restriction whole system design limit.Those skilled in the art can be the described function of each certain applications program accomplished in various ways, but this type of realizes determining should not be interpreted into to cause to depart from scope of the present invention.

Any combination that can use general processor, digital signal processor (DSP), application-specific integrated circuit (ASIC) (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components or its to be designed to carry out various functions described herein in conjunction with the described various illustrative logical blocks of each embodiment disclosed herein, module and circuit realizes or carries out.General processor can be a microprocessor, and still, alternatively, processor can be any conventional processors, controller, microcontroller or state machine.Processor can be embodied as the combination of some computing equipments, the for example combination of one or more microprocessors of DSP and microprocessor, a plurality of microprocessor, combination DSP core, or any other this type of configuration.

Can be embodied directly in hardware, in the software module that processor carries out or in the two the combination in conjunction with the step of described method of each embodiment disclosed herein or algorithm and to specialize.Software module can reside in the storage medium of RAM memory, flash memory, ROM memory, eprom memory, eeprom memory, register, hard disk, moveable magnetic disc, CD-ROM or any other form well known in the prior art.Exemplary storage medium is coupled to processor, so that processor can be from this read information, and information is write in this storage medium.Perhaps, storage medium can be integrated in the processor.Processor and storage medium can reside among the ASIC.ASIC can reside in the user terminal.Perhaps, processor and storage medium can be used as discrete assembly and reside in the user terminal.

Any technical staff the above description of the embodiment that is disclosed is provided, so that can make or use the present invention in this area.To those skilled in the art, will be conspicuous to the various modifications of these embodiment, and defined universal principle and can not depart from the spirit and scope of the present invention applicable to other embodiment among this paper.Therefore, do not attempt to limit the invention to the embodiment shown in this paper, but it is met and principle disclosed herein and novel characteristics the most wide in range corresponding to scope.

Claims (3)

1. method of allocating resources in the communication system of supporting Internet Protocol IP application program; Said communication system comprises an access network AN and a plurality of access terminal AT; Among the said AT each is all sent a requested data rate to said AN; Said communication system support has the application flows of qos requirement to said AT, and said method comprises:

For the increase in the adaptive reservation, preemption has first application flows that first kind QoS violates;

Confirm that said first kind QoS violation has arrived the preemption maximum; And

Arrived the preemption maximum if confirm said first kind QoS violation, then preemption has second application flows that second type of QoS violates,

The preemption priority of wherein said first and second application flows is dynamically to change.

2. the method for claim 1 is characterized in that, preemption second application flows also comprises:

Select said second application flows based on the employed timeslot number of the transmission of said second flow process.

3. the equipment of a Resources allocation in the communication system of supporting Internet Protocol IP application program; Said communication system comprises an access network AN and a plurality of access terminal AT; Among the said AT each is all sent a requested data rate to said AN; Said communication system support has the application flows of qos requirement to said AT, and said equipment comprises:

Be used to the device that the increase preemption in the adaptive reservation has first application flows of first kind QoS violation;

Be used for confirming that said first kind QoS violation has arrived the peaked device of preemption; And

Arrived the preemption maximum if be used for confirming said first kind QoS violation, then preemption has the device of second application flows of second type of QoS violation,

The preemption priority of wherein said first and second application flows is dynamically to change.

Images