CN101111070B - Fairness resource scheduling method based on resource scheduling in broadband wireless access system - Google Patents

Abstract

The present invention relates to the resource dispatch of a broadband wireless access system based on an IEEE802.16, the present invention is characterized in that the reserved bandwidth required by mutual necessity is calculated according to the Qos parameter of the different business, the admission and control is performed to the newly achieved business according to the reserved bandwidth, and the bandwidth is distributed by the means of a secondary dispatching mode according to bandwidth required by each business and the bandwidth required by a system, the reserved bandwidth is distributed according to each business firstly, the secondary dispatch is performed to the insufficient part of the bandwidth required by the means of the weighted recycle transfer method according to the residual bandwidth after a primary dispatch is performed to the system, the secondary dispatch distributes the different priority factors among the different businesses according to the business priority, meanwhile, the distribution performed according to the actual scale of the bandwidth required by the business residue accounting for the bandwidth required by residue also shall be taken into account. The present invention has the advantage of the equity, the high efficiency, the simplicity and the flexibility on the basis of ensuring the Qos quality.

Description

In the BWA based on the fairness resource scheduling method of scheduling of resource

Technical field

The present invention designs the method for scheduling of resource in a kind of IEEE 802.16 broadband wireless accesss (BWA) system, belongs to wireless communication technology field.

Background technology

IEEE 802.16 standards are air interface specification of wireless MAN, and crucial meaning and wide application prospect are arranged in the broadband wireless access in future.802.16 standard not only allows non line of sight to connect, supports large-capacity user, and can provide carrier-class QoS assurance, the application of the multimedia service that support voice, video, data etc. are abundant comprehensively.

The effective QoS guarantee system must carry out reasonable distribution to Radio Resource, and especially at the complex environment under the wireless communications environment, the QoS system of wireless access system faces a lot of challenges.802.16 standard has been implemented a series of mechanism at the MAC layer.Resource scheduling algorithm is an indispensable ring in the BWA QoS guarantee system.And 802.16 standards only propose to carry out scheduling of resource by base station or user according to each professional QoS request and priority, do not stipulate concrete resource scheduling algorithm.And the existing resources dispatching algorithm is not all considered the fairness problem of algorithm, under the very heavy situation of high-priority service load, causes low priority traffice to can not get bandwidth for a long time and " dying of hunger " easily.This algorithm carries out reasonable distribution to resource, thereby improves the fairness of algorithm by adopting certain balancing under the prerequisite that at first guarantees each QoS.

There is the business of four kinds of different grades of service in IEEE 802.16 systems, be respectively that business (unsolicitedgrant service is authorized in non-application, UGS), real-time polling service (real-time polling service, rtPS), non-real-time polling service (non-real-time polling service, nrtPS), the business of doing one's best (best effort service, BE).

The existing resources dispatching algorithm is based on the dispatching algorithm of strict priority, and its concrete realization flow is as follows:

Schedule level one algorithm: between the business of each different priorities kind, adopt SPQ (strict priority queue: strategy strict-priority), satisfy the bandwidth request of the high business of priority earlier, also have the bandwidth request that satisfies low priority traffice under the remaining situation more successively in bandwidth;

Second-level dispatching algorithm: between the business of same priority kind, adopt different allocated bandwidth strategies:, adopt EDF (earliest deadline first) strategy for the rtPS business according to business features; For the nrtPS business, adopt WFQ (weightedfair queue) strategy; For the BE business, adopt RR (round robin) strategy or divide equally band-width tactics.

Because the SPQ strategy is used in first order scheduling, under the very heavy situation of high-priority service load, cause easily high-priority service too much take the low priority traffice resource, thereby make low priority traffice can not get necessary bandwidth and the situation of " dying of hunger " occurs, can't satisfy the quality of service requirement of low priority traffice, concerning low priority traffice, lose fairness;

Resource reservation is exactly that system reserves the Qos demand that a part of resource (as bandwidth, buffering area or processor ability or the like) guarantees business when service access.Thought based on resource reservation, when service access, just for reserving its a part of bandwidth, and between each business, implement flexible and distribution justice by certain algorithm to remaining bandwidth, so both can effectively guarantee professional QoS demand, realize the fairness of allocated bandwidth between each business again.

Summary of the invention:

The objective of the invention is to propose the fairness resource scheduling algorithm in a kind of IEEE of being applied in 802.16 BWA based on resource reservation, can under the prerequisite that guarantees each QoS, reduce the situation that high-priority service too much takies the low priority traffice bandwidth, improve fairness, obtain better system performance.

Characteristic of the present invention is, has following steps successively:

Step (1) after new business arrives, should be calculated as the bandwidth reserved that satisfies different qos parameters based on the base station of the IEEE802.16 BWA qos parameter according to different business, was made as r _Res, comprising the bandwidth and the required bandwidth reserved that newly connects that connect and reserve.If ∑ r _Res≤ r _Total, then allow to insert, otherwise refusal inserts, r _TotalBe overall system bandwidth;

Authorize business for non-application, r _Res=r _Max, r _MaxBe maximum input data transfer rate, set point;

To the business of doing one's best, r _Res=0;

For real-time polling service, $r_{\mathrm{res}}=\frac{r_{\max }+r_{\min }}{2}{P}_0{e}^{-\mathrm{\λ}(t-T_S)},$

Wherein, r _MinBe minimum input data transfer rate, set point;

P ₀Be packet loss thresholding, set point;

λ is the arrival rate of the bag of enforcement poll business, given value;

Be maximum delay, T _SBe the time of a frame, be set point;

For non-real-time polling service, r _Res=r _Min

Step (2), the base station is according to IEEE802.16 medium specified MAC layer mechanism, periodically to assign mode system bandwidth is distributed, and in the media interviews control information UL-MAP of descending sub frame, notify each subscriber station allocation result according to the existing bandwidth of each professional desired bandwidth and system:

The first step: schedule level one

Earlier system's bandwidth reserved is satisfied the bandwidth request of priority business from high to low, if professional desired bandwidth r _ReqLess than bandwidth reserved, then make the bandwidth r of schedule level one _{Allo_1}=r _Req, otherwise make r _{Allo_1}=r _Res, the highest business of described priority is meant that non-application authorizes business, all the other are followed successively by real time polling, non real-time poll and Best-Effort service;

Second step: second-level dispatching

Again the remaining bandwidth after the schedule level one is distributed to each business by weighting loop jump method, each professional weight factor is: k _i=λ _iD _i

Wherein, λ _iFor each professional residue requires bandwidth to account for the ratio of total surplus requirement bandwidth: promptly ${\mathrm{\λ}}_i=(r_{\mathrm{req},i}-r_{\mathrm{res},i})/\underset{j=1}{\overset{n}{\mathrm{\Σ}}}(r_{\mathrm{req},j}-r_{\mathrm{res},j}),$ If r _{Req, i}＜r _{Res, i}, make r _{Req, i}-r _{Res, i}=0;

J=1,2 ..., n, for requiring to distribute the professional number of remaining bandwidth,

d _iBe priority factors, set point, the priority factors of real time business should should be set different priority factors for each non-real-time service greater than the priority factors of non-real-time service;

The resulting bandwidth of second-level dispatching is: r _{Allo_2}=(r _Total-∑ r _{Allo_1}) k _i/ ∑ k _i,

Distribute through two-stage, each professional bandwidth that obtains is r _Allo=r _{Allo_1}+ r _{Allo_2}

This dispatching method based on resource reservation has following advantage:

1.QoS guarantee: admit the strategy of the distribution bandwidth reserved in control and the schedule level one, professional qos parameter can be guaranteed;

2. fairness: adopt weighting loop jump WRR algorithm in the second-level dispatching, improved scheduling fairness;

3. efficient: owing to calculate the bandwidth reserved r of rtPS business _ResThe time, the delay requirement of rtPS business is also taken into account, rather than with maximum data input rate r _MaxAs criterion, therefore under the condition of guaranteed qos, reduced the waste of bandwidth, can insert more business, thereby improve bandwidth availability ratio;

4. simplicity: compare with the fair DFPQ of the priority query scheduling algorithm of deficit, this algorithm is more directly perceived, and it is all simpler to understand still operation;

5. flexibility: can be to different business according to the priority factors in the different WRR algorithm of situations such as load setting, this has embodied the flexibility of algorithm.

Description of drawings

The QoS guarantee system of Fig. 1 .IEEE 802.16 systems;

Fig. 2. existing resource dispatching algorithm flow chart;

Fig. 3. the resource scheduling algorithm flow chart that the present invention proposes;

Each professional variation diagram that obtains bandwidth during Fig. 4 .rtPS traffic growth;

Each professional variation diagram that obtains bandwidth during Fig. 5 .nrtPS traffic growth;

Fig. 6. adopt the fairness index contrast figure of the present invention front and back.

Embodiment

For realizing the technical problem to be solved in the present invention, the present invention adopts a kind of resource scheduling algorithm of introducing resource reservation thought and combining with admission control algorithm, comprises the steps:

(1) when each professional application inserts, subscriber station SS reports professional desired qos parameter to base station BS, BS determines professional required bandwidth reserved according to these parameters, if the bandwidth reserved summation of access service adds that the bandwidth reserved sum of application access service is less than the total bandwidth of system, then allow service access, otherwise refusal inserts;

(2) BS carries out dynamic assignment to system bandwidth before each frame begins, and distribution method is:

Schedule level one: give each traffic assignments bandwidth reserved earlier, thereby each professional service quality is guaranteed;

Second-level dispatching: system's remaining bandwidth is distributed between each business by the WRR algorithm, and the weight factor of WRR is determined by professional institute's bandwidth that requires and professional characteristic itself;

Below the specific embodiment of the present invention is further described.

Fig. 3 has provided the flow chart of new algorithm, specifies as follows:

(1) after new business arrives, at first according to the qos parameter of business, the bandwidth reserved of computing service

Our algorithm is: when a business need connected, BS was calculated as the bandwidth reserved that satisfies these parameters according to its qos parameter, is made as r _Res, if the institute promising connected the reservation bandwidth add that the required bandwidth reserved sum of new connection is less than total bandwidth r _Total, i.e. ∑ r _Res≤ r _Total, then allow to insert.r _ResBe calculated as follows:

1. for the UGS business, because all qos parameters all need satisfy, thereby r _Res=r _Max

2. for rtPS business, r _ResShould be between r _MinAnd r _MaxBetween a value, when BS offers this professional r _ResBandwidth the time, this professional time delay and packet loss are met the demands; Can calculate by certain mathematical method, also can follow and measure according to real system;

3. for the nrtPS business, owing to do not have delay requirement, r _Res=r _Min

4. for the BE business, because without any Qos requirement, r _Res=0;

(2) the MAC layer mechanism of stipulating according to IEEE 802.16 systems, BS will periodically distribute system bandwidth according to the existing bandwidth of each professional desired bandwidth and system, and notify each SS with allocation result in the MAC of descending sub frame control information UL_MAP;

Bandwidth allocation methods is:

A. schedule level one:

Earlier with bandwidth reserved according to priority from high to low order assignment give each business, but if professional desired bandwidth r _ReqLess than bandwidth reserved r _Res, then only need to distribute bandwidth reserved r _ResR even _Req＜r _Res, the bandwidth r that obtains of schedule level one then _{Allo_1}=r _ReqOtherwise r _{Allo_1}=r _Res

B. second-level dispatching:

Remaining bandwidth after the schedule level one is pressed the WRR algorithm assigns give each business;

Each professional weight factor is:

k _i＝λ _i·d _i；

Wherein, λ _iFor each professional residue requires bandwidth to account for the ratio of total surplus requirement bandwidth: promptly ${\mathrm{\λ}}_i=(r_{\mathrm{req},i}-r_{\mathrm{res},i})/\underset{j=1}{\overset{n}{\mathrm{\Σ}}}(r_{\mathrm{req},j}-r_{\mathrm{res},j})$ (if r _{Req, i}＜r _{Res, i}, make r _{Req, i}-r _{Res, i}=0; r _{Req, i}And r _{Res, i}Represent i professional desired bandwidth and bandwidth reserved respectively; r _{Req, j}And r _{Res, j}Represent j professional desired bandwidth and bandwidth reserved respectively)

d _iBe priority factors, how many occurrences can decide according to the traffic carrying capacity of actual each Business Stream of system, should satisfy following two principle simultaneously:

1. the priority factors of real time business should be greater than the priority factors of non-real-time service, to reduce the probability that surpasses the qos requirement packet loss because of time delay;

2. for non-real-time service, also can set different priority factors, obtain bigger bandwidth to guarantee the high business of priority to different business.

The bandwidth that second-level dispatching obtains is: r _{Allo_2}=(r _Total-∑ r _{Allo_1}) k _i/ ∑ k _i

So, to distribute through two-stage, each professional bandwidth that obtains is: r _{Allo, i}=r _{Allo_1, i}+ r _{Allo_2, i}, r wherein _{Allo_1, i}Be i the professional bandwidth that obtains by schedule level one, r _{Allo_2, i}Be i the professional bandwidth that obtains by second-level dispatching.

In order to contrast the effect of the invention process front and back, algorithm of the present invention has been carried out emulation, simulation parameter is as follows: total bandwidth B=10Mbps, frame length t _f=10ms

Type of service	Average bandwidth	Maximum bandwidth	Minimum bandwidth	Bandwidth reserved
					UGS	\	1.6Mbps	\	1.6Mbps
rtPS	4.3Mbps	5.2Mbps	3.4Mbps	4Mbps
					nrtPS	2.8Mbps	3.6Mbps	2Mbps	2Mbps
BE	\	\	1.6Mbps	0

Fig. 4 has represented the situation of the resultant bandwidth of each Business Stream under the situation that the rtPS traffic carrying capacity increases.Here, we call SPQ (strict-priority algorithm) to original algorithm, and new algorithm is called RFPQ (the fair priority query of reservation formula algorithm).As seen from the figure, when total traffic was no more than overall system bandwidth, two kinds of resulting results of allocative decision were the same; When the rtPS traffic carrying capacity acquires a certain degree, adopt existing algorithm, the rtPS business will tie up the bandwidth of BE business, cause professional linear minimizing of BE, so that not reach qos requirement at last; And in the present invention, the bandwidth that the rtPS business is obtained is no longer linear to be increased, and its growth trend slows down; The bandwidth of nrtPS and BE business then slowly reduces.Compare with existing algorithm, the allocated bandwidth equalize trend of three kinds of business, even if the flow of rtPS business is very heavy like this, system also can hold more low priority traffice under the prerequisite of guaranteed qos.

Fig. 4 has represented the situation of the resultant bandwidth of each Business Stream under the situation that the nrtPS traffic carrying capacity increases.As seen from the figure, adopt existing algorithm, not increasing of nrtPS business can impact the rtPS business, and the BE business is declined to a great extent; Adopt this algorithm, when the nrtPS traffic carrying capacity increased, the bandwidth that rtPS obtained had small size decline, but within acceptable scope, can not reduce professional service quality and guarantee; The bandwidth downward trend of BE business obviously slows down with respect to existing algorithm, and this illustrates that this algorithm can make low priority traffice obtain certain treatment.

Fig. 5 has represented the variation of fairness index under the situation that the rtPS traffic carrying capacity increases.Fairness index reference literature is defined as the mean square deviation of each service bandwidth satisfaction:

$\mathrm{fairness}=\sqrt{{(S_{\mathrm{UGS}}-\stackrel{\‾}{S})}^2+{(S_{\mathrm{rtPS}}-\stackrel{\‾}{S})}^2+{(S_{\mathrm{nrtPS}}-\stackrel{\‾}{S})}^2+{(S_{\mathrm{BE}}-\stackrel{\‾}{S})}^2}$

Wherein, S=r _Allo/ r _Req, be the bandwidth satisfaction of business; S=(S _UGS+ S _RtPS+ S _NrtPS+ S _BEThe average of each service bandwidth satisfaction of)/4; As seen from the figure, adopt existing algorithm, can effectively reduce the mean square deviation of each service bandwidth satisfaction, make each professional bandwidth trend that obtains consistent, thereby improve resource scheduling algorithm each professional fairness.

The professional bandwidth reserved computational methods of rtPS are as follows:

At first we set up an approximate Mathematical Modeling.Consider from the most conservative situation, suppose that the every frame of system distributes to the bandwidth of certain rtPS business and be r _Res, establishing data input rate is r _In, r _Min≤ r _In≤ r _Max, the data output rating is r _Res, it is Poisson process that the rtPS business data packet arrives, arrival rate is λ, then at a period of time [t ₀, t ₀+ t] in the byte number that arrives be $X\left(t\right)=\underset{i=1}{\overset{N\left(t\right)}{\mathrm{\Σ}}}{l}_i,$ Wherein N (t) is [t ₀, t ₀+ t] in the number of the bag that arrives, l _iLength for bag.Because l _iIndependent same distribution, and independent with N (t) statistics, so X (t) is a compound poisson process, establishes l _iGenerating function be F (S), the generating function of Poisson process { N (t) } is G (S), G (S)=e ^{λ t[S-1]}, wherein t is a time parameter, and S is a Laplacian, and then: the generating function of X (t) is G (F (S))=e ^{λ t[F (S)-1]}, can obtain expectation and the variance of X (t) thus: E (X (t))=λ tE (l _i), D (X (t))=λ tE (l _i ²); When t was enough big, N (t) was also enough big, because l _iIndependent same distribution, by law of great number, X (t) Normal Distribution: X (t)～N (λ tE (l _i), λ tE (l _i ²)).Average data rate $r_{\mathrm{in}}\left(t\right)=\frac{X\left(t\right)}{t},$ Also Normal Distribution: r _In(t)～N (λ E (l _i), λ E (l _i ²)/t), r _MaxWith r _MinAs r _In(t) the upper limit and lower limit should have: (r _Max+ r _Min)/2=E (r _In(t))=λ E (l _i);

In fact, we can regard arrival, wait and the service process of bag as a batch processing system, and the arrival of bag is Poisson process, and the server is every a frame (T _S) handle once, a collection of packet of single treatment, the number of pack processing is made as Y (t), should satisfy following relation: $\underset{i=1}{\overset{Y\left(i\right)}{\mathrm{\&Sigma;}}}{l}_i<r_{\mathrm{res}}\&CenterDot;T_S;$ The time of handling is T _SWe consider the distribution of the stand-by period W (t) that wraps now: in a collection of bag of Chu Liing, we consider that bag that the stand-by period is the longest at the same time; Because formation is a First Come First Served FCFS system, so first bag arrives at first in this a collection of bag, thereby the stand-by period is also the longest.In order to make the stand-by period be no more than the qos parameter of regulation, we need obtain the distribution of the stand-by period W (t) of first bag.When this bag arrived, the bag of its front was a collection of processing, and the processing time all is T _S, in order to ask the distribution of stand-by period W (t), we can regard the bag of front as and handle one by one, and in the time of a frame, the byte number of processing is T _Sr _Res, therefore the number of bag is T _Sr _Res/ E (l _i), the average handling time of each bag is T _S/ (T _Sr _Res/ E (l _i))=E (l _i)/r _Res, so this just equivalence become the M/D/1 queuing system of classics, can obtain: P (W＜t)=1-ρ e ^{λ t}, wherein, ρ=λ h=λ E (l _i)/r _Res, again as can be known by (1), ρ=(r _Max+ r _Min)/2r _ResBy ρ＜1 as can be known, $r_{\mathrm{res}}>\frac{r_{\max }+r_{\min }}{2}.$ When time-delay surpasses the maximum delay t that stipulates when inserting _dThe time, packet will be dropped, and therefore can obtain the probability because of the time-delay packet loss: $P_d=P(W+T_S>t)={\mathrm{\&rho;e}}^{\mathrm{\&lambda;}(t-T_S)};$ Because the packet loss to real time business when inserting has regulation: P _d＜P ₀, P wherein ₀Packet loss thresholding for real time business.Thereby can obtain: $r_{\mathrm{res}}>\frac{r_{\max }+r_{\min }}{2}{P}_0{e}^{-\mathrm{\&lambda;}(t-T_S)},$ But have simultaneously $\frac{r_{\max }+r_{\min }}{2}<r_{\mathrm{res}}\&le;r_{\max },$ So the minimum bandwidth of reserving to rtPS is $\min (\frac{r_{\max }+r_{\min }}{2}{e}^{-\mathrm{\&lambda;}(t-T_S)},r_{\max }).$ Can get $r_{\mathrm{res}}=\frac{r_{\max }+r_{\min }}{2}{P}_0{e}^{-\mathrm{\&lambda;}(t-T_S)}.$

Wherein, r _MaxAnd r _MinBe respectively maximum bandwidth and minimum bandwidth, λ is the arrival rate of the bag of RtPS business, P ₀Be the packet loss thresholding, t is a maximum delay, T _SIt is the time of a frame.The bandwidth reserved of RtPS business just can be calculated by following formula each qos parameter according to business.

Claims (1)

1. In the BWA based on the fairness resource scheduling algorithm of resource reservation, its characteristic is, has following steps successively:

   Step (1) after new business arrives, based on the base station of the IEEE802.16 BWA qos parameter according to different business, is calculated as the bandwidth reserved that satisfies different qos parameters, is made as r _Res, comprising the bandwidth and the required bandwidth reserved that newly connects that connect and reserve, if ∑ r _Res≤ r _Total, then allow to insert, otherwise refusal inserts, r _TotalBe overall system bandwidth;

   Authorize business for non-application, r _Res=r _Max, r _MaxBe maximum input data transfer rate, set point;

   To the business of doing one's best, r _Res=0;

   For real-time polling service,

   

   Wherein, r _MinBe minimum input data transfer rate, set point;

   P ₀Be packet loss thresholding, set point;

   λ is the arrival rate of the bag of real-time polling service, given value;

   T is a maximum delay, T _SBe the time of a frame, be set point;

   For non-real-time polling service, r _Res=r _Min

   Step (2), the base station is according to IEEE802.16 medium specified MAC layer mechanism, periodically system bandwidth is distributed, and in the UL-MAP of descending sub frame message, notifies each subscriber station allocation result according to the existing bandwidth of each professional desired bandwidth and system:

   The first step: schedule level one

   Earlier system's bandwidth reserved is satisfied the bandwidth request of priority business from high to low, if professional desired bandwidth r _ReqLess than bandwidth reserved, then make the bandwidth r of schedule level one _{Allo_1}=r _Req, otherwise make r _{Allo_1}=r _Res, the highest business of described priority is meant that non-application authorizes business, all the other are followed successively by real time polling, non real-time poll and Best-Effort service;

   Second step: second-level dispatching

   Again the remaining bandwidth after the schedule level one is distributed to each business by weighting loop jump method, each professional weight factor is: k _i=λ _iD _i

   Wherein, λ _iFor each professional residue requires bandwidth to account for the ratio of total surplus requirement bandwidth: promptly

   

   If r _{Req, i}＜r _{Res, i}, make r _{Req, i}-r _{Res, i}=0;

   J=1,2 ..., n, for requiring to distribute the professional number of remaining bandwidth,

   r _{Req, i}And r _{Res, i}Be respectively i professional bandwidth and the bandwidth reserved of requiring,

   r _{Req, j}And r _{Res, j}Be respectively j professional bandwidth and the bandwidth reserved of requiring,

   d _iBe priority factors, set point, the priority factors of real time business should should be set different priority factors for each non-real-time service greater than the priority factors of non-real-time service;

   The resulting bandwidth of second-level dispatching is: r _{Allo_2}=(r _Total-∑ r _{Allo_1}) k _i/ ∑ k _i,

   Distribute through two-stage, each professional bandwidth that obtains is r _Allo=r _{Allo_1}+ r _{Allo_2}

Images