CN101242659A - Multi-service type call permission control method based on self-adapted control - Google Patents

Abstract

A permission control method for calling of various service type based on adaptive control, which pertains to wireless network technology. The invention provides differentiating method of various service type based on QoS requirements of different service type in wireless network system; establishes model of queue with priority, proposes resource combating method of different service type when system resource is scarce, and provides general solving procedure of call dropping probability when service switching and blocking probability of new call by analysis of Markov birth and death process; establishes adaptive fuzzy degrade control model, proposes degrade control method of the current call when system resource is scarce, realizes dynamic balance of user QoS and system resource utility; finally, establishes optimal allocation model of degrade amplitude objects by non-linear programming algorithm. Three models proposed by the invention constitutes a close-loop control system together, which is excellent in completeness and robustness, and easy to implement.

Description

Multi-service type call based on adaptive control allows control method

Technical field

The present invention relates to the control method in a kind of mobile communication technology field, specifically is that a kind of multi-service type call based on adaptive control allows control method.

Background technology

The 4th generation wireless network a main target be: the Multimedia Mobile business that provides ubiquitous, always has for each mobile subscriber.Thereby, how under wireless access technology (as the 3G honeycomb, IEEE802.11 WLAN, the bluetooth) environment of diversification, for the multimedia service with different QoS (service quality) requirement provides reliable service, just become one of problem the most key in the wireless network research of the 4th generation.For in the heterogeneous wireless network system, for the different kinds of business with different QoS requirement provides reliable service, be necessary to set up a kind of effective call Admission Control method simultaneously.

Call Admission Control has mainly solved such problem, that is: whether accept a new calling or switch call and enter network.A call request is accepted to satisfy following two conditions by call Admission Control mechanism: the QoS that can guarantee existing call in the network is not subjected to the new influence of calling out that inserts; Enough resources that has network satisfy the desired QoS of this call request.For a call Admission Control method, from user's angle, its resulting service of user expectation has good QoS and ensures; And on the other hand, from the angle of Virtual network operator, Virtual network operator expectation Internet resources have higher utilance.Therefore, the call Admission Control method need this seeks a Tradeoff point in to contradiction in user QoS and resource utilization ratio.For the Radio Network System of multi-service type, the QoS standard of the various types of traffic in the system has nothing in common with each other.Meanwhile, various types of traffic also has nothing in common with each other for network resources demand level and characteristics.Therefore, the call Admission Control method of multi-service type Radio Network System is a very complicated problems.

Find through literature search prior art, Fei Yu etc. on the IEEETransactions in January, 2007 Mobile Computing (international institute of electrical and electronic engineers mobile computing journal), publish an article " Optimal Joint Session Admission Control in Integrated WLANand CDMA Cellular Networks with Vertical Handoff (the optimal joint call Admission Control that has vertical switching in WLAN (wireless local area network) and the CDMA cellular network UNE) ", this article has proposed the overall earnings target function of multi-service type Radio Network System, by varying level QoS parameter (throughput, packet delay, SIR, calling call drop probability etc.) constraint, found the solution linear optimization problem, realized the maximization of network in general income based on the semi-Markovian decision process model.But the optimal model that this method proposed only with QoS as constraints to guarantee satisfying of minimum qos requirement, simultaneously user QoS and resource utilization ratio are not weighed.Also find in the retrieval; Yufeng Ma etc. on AmericanControl Conference in 2005 (U.S.'s control academic conference), publish an article " Intelligent CallAdmission Control Using Fuzzy Logic in Wireless Networks (intelligent call based on fuzzy logic in the wireless network allows control) "; this article has proposed the self adaptation call Admission Control method based on fuzzy logic, fuzzy control theory is applied to protect the control of the channel reservation number in the channel strategy.But this method is not considered that polytype is professional and is deposited traffic differentiation problem under the situation.Therefore, be necessary to seek a kind of optimum self adaptation call Admission Control method of balance user QoS and resource utilization ratio factor at the multi-service type system time.

Summary of the invention

The present invention is directed to above-mentioned the deficiencies in the prior art, proposed a kind of multi-service type call and allowed control method based on adaptive control, make it under the prerequisite of distinguishing the different service types qos requirement, dynamic balance user QoS and resource utilization ratio, and optimum distribution network system resource.

The present invention is achieved by the following technical solutions, the present invention includes following concrete steps:

Step 1 is set up the Mathematical Modeling of system business, with the arrival process of all types of business of system as Poisson process, service process obeys index distribution, and computing service takies the average arrival rate and the average service rate of resource;

Described arrival process with all types of business of system is as Poisson process, the service process obeys index distribution is specially: for a certain type service in the system, the arrival process of its new calling and switch call is all obeyed Poisson process, the equal obeys index distribution of average service time of new calling and switch call, as the total dissimilar business of J kind in the system, for k class business, k=1,2, ..., J, it is λ that the arrival process of its new calling is described as parameter _n ^(k)Poisson process, it is λ that the arrival process of its switch call is described as parameter _h ^(k)Poisson process, λ _n ^(k)With λ _h ^(k)Be respectively the average arrival rate of new calling and switch call; Simultaneously, the new average service time of calling out is 1/ μ _n ^(k), the average service time of switch call is 1/ μ _n ^(k), the two equal obeys index distribution, the average channel retention time that average service time promptly refers to, μ _n ^(k)With μ _h ^(k)Be respectively the average service rate of new calling and switch call, the investigation object of above-mentioned average arrival rate and average service rate is the number of calls of a certain type service in the system, and dimension is: call out number/unit interval;

Described computing service takies the average arrival rate and the average service rate of resource, be specially: for the professional resource that takies in the system, resource is supported all k=1,2, ..., the access of J type service, the definition base unit is bandwidth (BU), come the expression system can be and arrive the bandwidth resources amount that the least unit that is provided is provided, the heap(ed) capacity of system is C, and unit is BU, then before the maximum bandwidth resource C of system all assigns, for whole system, the average arrival rate that its all types business takies resource is ${\mathrm{\λ}}^T=\underset{k=1}{\overset{J}{\mathrm{\Σ}}}\left[({\mathrm{\λ}}_n^{\left(k\right)}+{\mathrm{\λ}}_h^{\left(k\right)}){\mathrm{BW}}^{\left(k\right)}\right],$ Wherein, BW ^(k)Represent the demand of the calling of each k class business for bandwidth resources, calling comprises new calling and switch call; For whole system, its all types business take resource average service rate be ${\mathrm{\μ}}^T=\underset{k=1}{\overset{J}{\mathrm{\Σ}}}\left[({\mathrm{\μ}}_n^{\left(k\right)}+{\mathrm{\μ}}_h^{\left(k\right)}){\mathrm{BW}}^{\left(k\right)}\right],$ Above-mentioned average arrival rate λ ^TWith average service rate μ ^TThe investigation object be the resource that all types business takies in the system, dimension is: the BU/ unit interval.

Step 2 is introduced the traffic differentiation vector to the Mathematical Modeling of system business, distinguishes the priority and the resource requirement of different kinds of business;

Described introducing traffic differentiation vector is distinguished the priority and the resource requirement of different kinds of business, and is specific as follows: belong to the calling of k class type of service for each, set up unique and the professional corresponding traffic differentiation vector (α of the type ^(k), β ^(k)) portray its service feature, wherein, k=1,2 ..., J, α ^(k)Be meant priority factor, β ^(k)Be meant the resource requirement coefficient.

Described priority factor α ^(k)Be used among priority query's model, its value depends on the transmission delay requirement of k class business, priority factor α ^(k)Be worth greatly more, the priority that is had is high more, and then its real-time is strong more, and is strict more to the requirement of transmission delay; Priority factor α ^(k)Be worth more for a short time, the priority that is had is low more, and then its real-time is weak more, and is loose more to the requirement of transmission delay, priority factor α ^(k)Span is: 0≤α ^(k)≤ 1.

Described resource requirement factor beta ^(k), being used for adaptive fuzzy and degrading among controlling models and the optimal allocation model of amplitude object of degrading, its value depends on the demand for network bandwidth resources of k class business, β ^(k)Be worth greatly more, its demand to network bandwidth resources is high more, β ^(k)Be worth more for a short time, its demand to network money bandwidth source is low more.

Step 3 is set up priority query's model, realizes the resource contention of different kinds of business when system resource is in short supply;

The described priority query's model of setting up, the basic assumption of its foundation is specific as follows: (1) before the maximum bandwidth resource C of system assigned, the resource of network was all obtained in the new calling of each type of service and switch call with identical priority; When the maximum bandwidth resource C of system that (2) and if only if has assigned, when promptly system resource is in short supply, set up waiting list based on priority factor; (3) when system resource is in short supply, for k class business arbitrarily, the control ratio of switch call call drop probability is newly called out and is stoped the control of probability to seem more important, this be because, be prevented from this two kinds of situations for ongoing calling generation call drop and the new calling of initiating, the former will produce bigger negative effect to user satisfaction.

The described priority query's model of setting up, specific as follows: according to the priority factor α of different service types ^(k), foundation has other waiting list of different priorities accordingly, when the maximum bandwidth resource C of system has assigned, sets up based on priority factor α ^(k)(k=1,2 ..., waiting list J), at this moment, total J parallel waiting list in the system, the priority of formation is respectively α ⁽¹⁾, α ⁽²⁾..., α ^(J), separate between each priority query, insert request for newly arrived switch call, according to its priority factor α ^(k)Value, it is added the corresponding priority level coefficient is α ^(k)The tail of the queue of waiting list, and newly arrived new calling inserts request, will directly be prevented from; When system resource is in short supply, be in the calling in the different priorities waiting list, obtain Internet resources with following common resource contention method.

The described resource contention of when system resource is in short supply, realizing different kinds of business, specific as follows: as to be in the calling in the same priority waiting list, will be according to the regular compete for system resources of " First Come First Served ", be in the calling in the different priorities waiting list, the calling of individual queue head of the queue will be under the control of resource contention controller compete for system resources, being in priority factor is α ^(k)The handover call request of waiting list head of the queue, will be according to its priority factor α ^(k), with probability $P^{\left(k\right)}=\frac{{\mathrm{\α}}^{\left(k\right)}}{\underset{i}{\mathrm{\Σ}}{\mathrm{\α}}^{\left(i\right)}}$ Success is competed and is obtained system resource, when a certain formation becomes sky, and its priority factor α ^(k)To be entered this formation up to next handover call request by temporary transient zero setting.

Step 4, priority query's model based on step 3, adopt counting process to describe the business of each type, and determine according to counting process whether switch call and new calling can obtain resource, and calculating switch call call drop probability stops probability with new the calling;

Described employing counting process is described the business of each type, specifically is meant: for k class business, and k=1,2 ..., J uses counting process N ^(k)(t) described, at t constantly, counting process N ^(k)(t) equal the bandwidth sum that system taken by all types business and add that priority factor is α ^(k)Waiting list in the queuing bandwidth number that takies of switch call, counting process N ^(k)(t) be made up of two parts stack: first is because before the maximum bandwidth resource C of system assigned, the bandwidth resources of system were open to all types of business with equal opportunity; Second portion is because after the maximum bandwidth resource C of system assigned, when belonging to the handover call request arrival system of k class business, will be added to priority factor was α ^(k)The waiting list tail of the queue, thereby take the bandwidth resources of waiting list buffer, counting process N ^(k)(t) represented actual occupied system bandwidth resource, wherein, the heap(ed) capacity of waiting list buffer is Q, and unit is BU.

Describedly determine that according to counting process whether switch call and new calling can obtain resource, are specially:

1. before the maximum bandwidth resource C of system all assigns, i.e. 0≤N ^(k)(t)＜and C, N ^(k)(t) be counting process, waiting list is empty, and all types of new callings and switch call will be obtained the resource of network with identical priority, and this moment, the birth rate of birth and death process was λ ^T, the death rate is N ^(k)(t) μ ^T, μ ^TBe meant all types business take resource average service rate;

2. after the maximum bandwidth resource C of system all assigns, i.e. C≤N ^(k)(t)＜and C+Q, the new call request of all arrival will directly be prevented from, and the handover call request that arrives, will be according to its priority factor α ^(k), enter corresponding with it waiting list, so the birth rate of birth and death process is λ _h ^(k), the death rate is P ^(k)C μ ^T, promptly priority is α ^(k)The buffer of formation only receive k class switch call business, and the call business in this queue buffer will be with probability P ^(k)Come the resource of contention system;

3. work as N ^(k)(t) 〉=and during C+Q, system channel and buffer resources all assign, and at this moment, all new calling or switch calls all will be prevented from, for N ^(k)(t) 〉=and C+Q, birth rate is 0, for N ^(k)(t)=and C+Q, the death rate is P ^(k)C μ ^T, for N ^(k)(t)＞and C+Q, the death rate is 0.

Described calculating switch call call drop probability stops probability with new the calling, is meant for k class business arbitrarily, lists about steady-state distribution probability π _j ^(k)(j=0,1 ...) and state equation, and then solve π _j ^(k)General expression formula, the switching call drop probability of k class business $P_{\mathrm{drop}}^{\left(k\right)}={\mathrm{\π}}_{C+Q}^{\left(k\right)},$ The new calling of k class business stops probability $P_{\mathrm{block}}^{\left(k\right)}=\underset{n=C}{\overset{C+Q}{\mathrm{\Σ}}}{\mathrm{\π}}_n^{\left(k\right)},$ Wherein: C is the maximum bandwidth resource, and Q is meant the maximum bandwidth capacity of waiting list buffer.

Step 5 is set up the adaptive fuzzy controlling models that degrades, and user QoS and resource utilization ratio are dynamically adjusted in the control that degrades of existing call in the realization system when system resource is in short supply;

The described adaptive fuzzy controlling models that degrades, its core is the fuzzy controller that degrades, the fuzzy controller that degrades has two input parameters: 1. the switch call call drop probability of each type of service departs from the amplitude, ao that its maximum is allowed switch call call drop probability _Drop ^(k), ${\mathrm{\&Delta;}}_{\mathrm{drop}}^{\left(k\right)}=P_{\mathrm{drop},\mathrm{tolerance}}^{\left(k\right)}-P_{\mathrm{drop}}^{\left(k\right)},$ Wherein, P _Drop ^(k)Be switch call call drop probability, P _{Drop, tolerance} ^(k)For maximum is allowed switch call call drop probability, Δ _Drop ^(k)Reflected the QoS level that the user is current; 2. the current amplitude exponent gamma that degrades of existing business in the system ^D, $0\&le;{\mathrm{\&gamma;}}^D\&le;{\mathrm{\&gamma;}}_{\max }^D<1,$ Wherein, γ _Max ^DBe the upper limit of the amplitude that degrades, γ ^DThen reflected the utilance level that system resource is current; The fuzzy controller that degrades carries out obfuscation, fuzzy reasoning, de-fuzzy processing with above-mentioned two input parameters, and next amplitude index that degrades constantly of existing business is exported as controlled quentity controlled variable.

Describedly carry out obfuscation, fuzzy reasoning, de-fuzzy and handle, specific as follows:

At first, defined its fuzzy set respectively for two input parameters that blur the controller that degrades, and corresponding membership function;

Then, on the cartesian product space of two input parameter fuzzy sets, define Fuzzy Rule Sets, set up 25 fuzzy inference rules; The input parameter will input to fuzzy reasoning mechanism after Fuzzy processing, behind fuzzy reasoning, obtain and export the corresponding fuzzy set vector of controlled quentity controlled variable;

At last, adopt gravity model appoach that this fuzzy set vector is carried out de-fuzzy, obtain exporting next amplitude index that degrades constantly of controlled quentity controlled variable existing business.

Described dynamic adjustment user QoS and resource utilization ratio, specific as follows: the existing business in the system took whole system resource C, when system resource is in short supply before being degraded, be responsible for that the existing business in the system is carried out QoS and degrade, according to the output controlled quentity controlled variable γ of the fuzzy controller that degrades ^DIt is carried out after QoS degrades, and the existing business in the system only takies (1-γ ^D) system resource of C, d/d γ ^DThe system resource of C will offer by the successful call request of priority query's model competition.

Step 6 is set up the optimal allocation model of the amplitude object that degrades, and when system resource is in short supply, in the existing business of system, optimally chooses degrade object and the corresponding amplitude that degrades thereof by making the target function value reach Minimal Realization.

The degrade optimal allocation model of amplitude object of described foundation, specific as follows:

At first, choose the target function of quadratic function as model, wherein, each quadratic term has reflected the degrade departure degree of relative its primary standard demand resource of back institute's resource that obtains of all types of business respectively, simultaneously, the existing call number by all types of business is weighted stack with above-mentioned quadratic term;

Then, according to the relation of identity that total demand satisfied that degrades, the upper limit of the amplitude that degrades and lower limit requirement have provided the constraints of optimal allocation model, are specially:

Represented the to degrade actual relation of identity that degrades and satisfied between the amplitude total amount of total demand and all types of business;

2. B ^(k)≤ β ^(k), expression waits to ask the upper limit of decision variable;

The amplitude upper limit that degrades of representing the k type of service promptly waits to ask the lower limit of decision variable.

Wherein, B ^(k)After the operation that degrades, the bandwidth resources that each existing call occupied of k class business; β ^(k)It is the resource requirement coefficient of k class business; N ^(k)Calling number for k class business existing in the system.

At last, find the solution the decision variable of optimal model, make the target function value reach minimum, obtain the degrade optimum allocation of amplitude of existing business in the system, its Mathematical Modeling is expressed as follows:

Find the solution decision variable: B _Degrade=[B ⁽¹⁾, B ⁽²⁾..., B ^(J)] ^T, to satisfy target function:

The optimal allocation model of controlling models, the amplitude object that degrades of among the present invention priority query's model, adaptive fuzzy being degraded is common to constitute a complete close loop negative feedback control system, has realized call Admission Control.

Compared with prior art, the present invention has following beneficial effect: (1) the present invention is directed to the characteristics that multi-service type call in the 4th generation wireless network allows control problem, the differentiating method of different service types has been proposed, three models have been set up simultaneously, solved the resource contention problem of multi-service type, the dynamic trade-off problem of user QoS and network resource utilization, and the optimal assignment problem of Internet resources; (2) the comprehensive utilization of the present invention various analysis and model, as markov queuing model based on priority, based on the adaptive fuzzy of the Intelligent Control Theory controlling models that degrades, and based on optimal allocation model of the amplitude object that degrades of nonlinear programming approach etc.; (3) above-mentioned model has been formed a close loop negative feedback control system jointly, has good completeness and robustness; (4) call Admission Control method proposed by the invention has generality, and be easy to realize, can provide important theory foundation and design reference for the radio resource management method of systems such as the third generation, super three generations, the 4th generation, WLAN (wireless local area network), wireless wide area network.

Description of drawings

Fig. 1 is the entire system structural representation;

Fig. 2 is the FB(flow block) of setting up of priority query's model;

Fig. 3 is priority query's structure of models schematic diagram;

Fig. 4 is the state transition rate figure of professional corresponding M/M/C/ (C+Q) type of k class markov formation;

Fig. 5 is a Δ _Drop ^(k)Membership function figure;

Fig. 6 is γ _i ^D(or γ _I+1 ^D) membership function figure;

Fig. 7 is the fuzzy controller system structure function schematic diagram that degrades;

Fig. 8 is a voice service switch call call drop probability P _Drop ⁽¹⁾With professional arrival rate λ _n ⁽¹⁾, λ _h ⁽¹⁾Changing trend diagram;

Fig. 9 is that voice service is newly called out the prevention probability P _Block ⁽¹⁾With professional arrival rate λ _n ⁽¹⁾, λ _h ⁽¹⁾Changing trend diagram;

Figure 10 is a video traffic switch call call drop probability P _Drop ⁽²⁾With professional arrival rate λ _n ⁽²⁾, λ _h ⁽²⁾Changing trend diagram;

Figure 11 is that video traffic is newly called out the prevention probability P _Block ⁽²⁾With professional arrival rate λ _n ⁽²⁾, λ _h ⁽²⁾Changing trend diagram;

Figure 12 is a data service switch call call drop probability P _Drop ⁽³⁾With professional arrival rate λ _n ⁽³⁾, λ _h ⁽³⁾Changing trend diagram;

Figure 13 is that the prevention probability P is newly called out in data service _Block ⁽³⁾With professional arrival rate λ _n ⁽³⁾, λ _h ⁽³⁾Changing trend diagram;

Figure 14 is the fuzzy controller output controlled quentity controlled variable changing trend diagram that degrades $({\mathrm{\&gamma;}}_i^D=0);$

Figure 15 is the fuzzy controller output controlled quentity controlled variable changing trend diagram that degrades $({\mathrm{\&gamma;}}_i^D=0.10);$

Figure 16 is the fuzzy controller output controlled quentity controlled variable changing trend diagram that degrades $({\mathrm{\&gamma;}}_i^D=0.20);$

Figure 17 is the fuzzy controller output controlled quentity controlled variable changing trend diagram that degrades $({\mathrm{\&gamma;}}_i^D\&Element;[\mathrm{0,0.25}],{\mathrm{\&Delta;}}_{\mathrm{drop}}^{\left(k\right)}\&Element;[-\mathrm{0.1,0.1}\left]\right).$

Embodiment

Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed execution mode and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

As shown in Figure 1, in the present embodiment by priority query's model, adaptive fuzzy degrades, and controlling models and the optimal allocation model of the amplitude object that degrades are common to constitute a complete close loop negative feedback control system, the resource contention method of all types of business when priority query's model provides system resource in short supply, reach the flow process of finding the solution of corresponding QoS parameter (switch call call drop probability stops probability with new the calling), the switch call call drop probability of each type of service of being tried to achieve is with negative feedback to the adaptive fuzzy controlling models that degrades, the core of this model is the fuzzy controller that degrades, blur and degrade controller based on the current amplitude that degrades of switch call call drop probability of being imported and system, through fuzzy reasoning, next amplitude demand that degrades constantly of system is exported to the optimal allocation model of the amplitude object that degrades as controlled quentity controlled variable.The optimal allocation model of the amplitude object that degrades is finished the operation that degrades to existing business in the system, has realized the optimum allocation of system resource.

Present embodiment is to the process of setting up of this system, and the call Admission Control method based on this system has been carried out analog simulation and interpretation of result, and concrete steps are as follows:

Step 1 has co-existed in J=3 kind business in the system, be respectively: during k=1, and voice service; During k=2, the stream video traffic; During k=3, data service, for the k class business in the system, it is λ that the arrival process of its new calling is described as parameter _n ^(k)Poisson process, it is λ that the arrival process of its switch call is described as parameter _h ^(k)Poisson process.The new calling of supposing three kinds of types of service all has identical arrival rate with switch call, and for the k class business in the system, the new average service time of calling out is 1/ μ _n ^(k), the average service time of switch call is 1/ μ _n ^(k), the two equal obeys index distribution, suppose that the service rate of three kinds of types of service satisfies following relation: ${\mathrm{\&mu;}}_n^{\left(1\right)}:{\mathrm{\&mu;}}_n^{\left(2\right)}:{\mathrm{\&mu;}}_n^{\left(3\right)}=3:2:1;$ ${\mathrm{\&mu;}}_h^{\left(1\right)}:{\mathrm{\&mu;}}_h^{\left(2\right)}:{\mathrm{\&mu;}}_h^{\left(3\right)}=3:2:1.$

Associated arguments (after standardization) regulation to system is as follows: the maximum bandwidth stock number C=500 of system, unit: BU, for each business, the maximum bandwidth capacity Q=60 of priority queue buffer, the traffic differentiation vector of three kinds of business is respectively: (α ⁽¹⁾, β ⁽¹⁾)=(0.5,1), (α ⁽²⁾, β ⁽²⁾)=(0.3,3), (α ⁽³⁾, β ⁽³⁾)=(0.2,2), wherein, α ⁽¹⁾+ α ⁽²⁾+ α ⁽³⁾=1; β ^(k)Unit be BU.

Step 2, set up priority query's model, as shown in Figure 3, be the flow process of setting up of priority query's model, before the maximum bandwidth resource C of system assigned, the priority waiting list did not still exist, the new calling or the switch call that belong to any type of arrival system, all will in its arrival, be received into system, obtain system resource with identical priority; When the maximum bandwidth resource C in system has assigned, foundation is based on the waiting list of call priority, and give switch call with higher priority, when a new request arrives system, judge at first whether it is handover call request, if switch call, then enter next step operation, otherwise this request is called out for new, directly stoped the queuing model end of operation; For the switch call of filing a request, read the priority factor α in its traffic differentiation vector ^(k), if this switch call belongs to k class business, then its priority factor is α ^(k), and be α with priority factor ^(k)Handover call request add the tail of the queue of priority waiting list to.

As shown in Figure 2, be priority query's structure of models schematic diagram, the waiting list with different priorities is independently of one another, be positioned at the call request of each waiting list head of the queue will be under the control of resource contention controller compete for system resources, being in priority level is α ^(k)The handover call request of waiting list head of the queue will be according to its priority factor α ^(k), to be proportional to α ^(k)Probability $P^{\left(k\right)}=\frac{{\mathrm{\&alpha;}}^{\left(k\right)}}{\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&alpha;}}^{\left(i\right)}}$ Success is competed and is obtained system resource.

Under the control of priority query's model, for k class business, its corresponding counting process N ^(k)(t) an available M/M/C/ (C+Q) type markov queuing model is described, as shown in Figure 4, (1) before the maximum bandwidth resource C of system all assigns, i.e. 0≤N ^(k)(t)＜during C, the birth rate of birth and death process is λ ^T, the death rate is N ^(k)(t) μ ^T(2) after the maximum bandwidth resource C of system all assigns, i.e. C≤N ^(k)(t)＜during C+Q, the birth rate of birth and death process is λ _h ^(k), the death rate is P ^(k)C μ ^T(3) all assign when system channel and buffer resources, i.e. N ^(k)(t) 〉=during C+Q, the birth rate of birth and death process is 0, works as N ^(k)(t)=during C+Q, the death rate is P ^(k)C μ ^T, work as N ^(k)(t)＞during C+Q, the death rate is 0.Present embodiment has been found the solution about steady-state distribution probability π by birth and death processes is carried out segmentation definition and analysis _j ^(k)(j=0,1 ...) and state equation, drawn π _j ^(k)General expression formula.

The state equation solving result is as follows: work as j=0, and 1,2 ..., during C-1,

When j=C,

Work as j=C+1, C+2 ..., during C+Q-1, Wherein, $x_1=\frac{A+\sqrt{A^2-4\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="S2008100342821E00051.png"\; state="\{\{state\}\}">B</figure-callout>}}}{2},$ $x_2=\frac{A-\sqrt{A^2-4\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="S2008100342821E00051.png"\; state="\{\{state\}\}">B</figure-callout>}}}{2},$

Because, $\underset{n=0}{\overset{C+Q}{\mathrm{\&Sigma;}}}{\mathrm{\&pi;}}_n\&equiv;0,$ So, π ₀Thereby value just can be determined because π ₀General expression formula too loaded down with trivial details, do not repeat them here.Based on above-mentioned π _j ^(k)General expression formula, and then the switch call call drop probability that can draw all types of business stops probability with new the calling.

Step 3 is set up the adaptive fuzzy controlling models that degrades, and the control that degrades of existing call obtains next amplitude index output controlled quentity controlled variable γ that degrades constantly of existing business in the realization system when system resource is in short supply _I+1 ^D, dynamically adjust user QoS and resource utilization ratio;

The degrade core of controlling models of adaptive fuzzy is the fuzzy controller that degrades, and the fuzzy controller that degrades is responsible for the Δ to input _Drop ^(k), γ _i ^DCarry out obfuscation, fuzzy reasoning, de-fuzzy processing, next amplitude index output controlled quentity controlled variable γ that degrades constantly of output existing business _I+1 ^D, as shown in Figure 7, specific as follows:

In the obfuscation stage, the fuzzy controller that degrades departs from the amplitude, ao of its maximum permissible value respectively to input parameter switch call call drop probability _Drop ^(k)The degrade amplitude exponent gamma current with existing business _i ^DCarry out the obfuscation operation, to Δ _Drop ^(k) Divide 5 fuzzy sets altogether: [NB_DeltaDrop, NS_DeltaDrop, ZO_DeltaDrop, PS_DeltaDrop, PB_DeltaDrop]; Its corresponding meaning is respectively: [negative sense departs from bigger, and negative sense departs from less, and nothing departs from, and forward bias is from less, and forward bias is from bigger], Δ _Drop ^(k)Membership function definition as shown in Figure 5.For input parameter γ _i ^D, also be divided into 5 fuzzy sets: [VS_Degrade, S_Degrade, M_Degrade, L_Degrade, VL_Degrade], its corresponding meaning is respectively: [amplitude that degrades index is very little, and the amplitude that degrades index is little, the amplitude that degrades index is moderate, and the amplitude that degrades index is big, and the amplitude that degrades index is very big], γ _i ^DThe definition of function as shown in Figure 6.Output parameter γ _I+1 ^DWith input parameter γ _i ^DBelong to a category together, therefore, output controlled quentity controlled variable γ _I+1 ^DFuzzy set and membership function and input parameter γ _i ^DDefinition identical.

In the fuzzy reasoning stage, the fuzzy controller that degrades carries out reasoning according to Fuzzy Rule Sets, obtain and export the corresponding fuzzy set vector of controlled quentity controlled variable, present embodiment is formulated Fuzzy Rule Sets according to following empirical principle: for k class business arbitrarily, if its current switching call drop probability P _Drop ^(k)(be presented as input parameter Δ greatly _Drop ^(k)Negative sense departs from bigger), and the amplitude exponent gamma that degrades of this moment _i ^DLess, so, should increase the amplitude index that degrades.For k class business arbitrarily, if its current switching call drop probability P _Drop ^(k)Lessly (be presented as input parameter Δ _Drop ^(k)Negative sense departs from bigger), and the amplitude exponent gamma that degrades of this moment _i ^DBigger, so, should reduce the amplitude index that degrades.Fuzzy Rule Sets is as shown in table 1, with a rule being marked for * number is example, and the implication of this rule can be expressed as: IF " switching the amplitude that the call drop probability departs from desired indicator is that negative sense departs from big (NB_DeltaDrop) " AND " the amplitude index that degrades very little (VS_Degrade) that system is current " THEN " the adjusted amplitude index that degrades of system should be moderate (M_Degrade) ".

Table 1 Fuzzy Rule Sets

Output: the adjusted amplitude exponent gamma that degrades of system i+1 D		Input 1: switch the amplitude, ao that the call drop probability departs from desired indicator drop (k)
							NB_DeltaDrop	NS_DeltaDrop	ZO_DeltaDrop	PS_DeltaDrop	PB_DeltaDrop
		Input 2: the amplitude exponent gamma that degrades that system is current i D	VS_Degrade	M_Degrade*	S_Degrade	VS_Degrade						VS_Degrade	VS_Degrade
S_Degrade	L_Degrade						M_Degrade	S_Degrade	VS_Degrade	VS_Degrade
			M_Degrade	VL_Degrade	L_Degrade	M_Degrade					S_Degrade	VS_Degrade
L_Degrade	VL_Degrade						VL_Degrade	L_Degrade	M_Degrade	S_Degrade
			VL_Degrade	VL_Degrade	VL_Degrade	VL_Degrade					L_Degrade	M_Degrade

Table is annotated: the fuzzy set abbreviation in the table is defined as follows.

1. Δ _Drop ^(k)Pairing 5 fuzzy sets: [NB_DeltaDrop, NS_DeltaDrop, ZO_DeltaDrop, PS_DeltaDrop, PB_DeltaDrop], its respective sense is: [negative sense departs from bigger, and negative sense departs from less, nothing departs from, and forward bias is from less, and forward bias is from bigger];

2. γ _i ^DOr γ _I+1 ^D) pairing 5 fuzzy sets: [VS_Degrade, S_Degrade, M_Degrade, L_Degrade, VL_Degrade], its respective sense is: [amplitude that degrades index is very little, and the amplitude that degrades index is little, the amplitude that degrades index is moderate, and the amplitude that degrades index is big, and the amplitude that degrades index is very big].

In the de-fuzzy stage, model adopts " gravity model appoach " that this fuzzy set vector is carried out de-fuzzy, and is last, obtains exporting the exact value of controlled quentity controlled variable.

Step 4, set up the optimal allocation model of the amplitude object that degrades: according to optimal model based on nonlinear programming approach, the optimum allocation method of having selected for use Lemke (Le Moke) algorithm to find the solution the amplitude object that degrades in the present embodiment.If in the current system, N ⁽¹⁾=150, N ⁽²⁾=60, N ⁽³⁾=85; At this moment, the total demand that degrades of system is 0.15 * C=75.So nonlinear programming problem can specifically be expressed as:

min y＝150(B ⁽¹⁾-1) ²+60(B ⁽²⁾-3) ²+85(B ⁽³⁾-2) ²

s.t.150(1-B ⁽¹⁾)+60(3-B ⁽²⁾)+85(2-B ⁽³⁾)＝75

0.75≤B ⁽¹⁾≤1

2.25≤B ⁽²⁾≤3

1.5≤B ⁽³⁾≤2

Target function can be of equal value be expressed as:

$\min y=\frac{1}{2}(B^{\left(1\right)},B^{\left(2\right)},B^{\left(3\right)})\left[\begin{array}{ccc}300& 0& 0\\ 0& 120& 0\\ 0& 0& 170\end{array}\right]\left(\begin{array}{c}{B}^{\left(1\right)}\\ B^{\left(2\right)}\\ B^{\left(3\right)}\end{array}\right)+{\left(\begin{array}{c}-300\\ -360\\ -340\end{array}\right)}^T\left(\begin{array}{c}{B}^{\left(1\right)}\\ B^{\left(2\right)}\\ B^{\left(3\right)}\end{array}\right)+1030$

Constraint equation can be of equal value be expressed as:

$\left(\begin{array}{ccc}150& 60& 85\end{array}\right)\left(\begin{array}{c}{B}^{\left(1\right)}\\ B^{\left(2\right)}\\ B^{\left(3\right)}\end{array}\right)=425$

Through finding the solution, can get optimal solution and be:

${\left(\begin{array}{c}{B}^{\left(1\right)}\\ B^{\left(2\right)}\\ B^{\left(3\right)}\end{array}\right)}_{\mathrm{optimal}}=\left(\begin{array}{c}0.7500\\ 2.7414\\ 1.7414\end{array}\right)$

Step 5, analog simulation and interpretation of result: introduce the notion of " traffic intensity ", describe the arrival rate of all types of business and the relative scale of service rate, traffic intensity is defined as:

To get τ=3 is example, and the switch call call drop probability that Fig. 8 to Figure 13 has provided three types of business respectively stops the variation tendency of probability with its professional arrival rate with new the calling, and Fig. 8 and Fig. 9 are respectively voice service P _Drop ⁽¹⁾With P _Block ⁽¹⁾With professional arrival rate λ _n ⁽¹⁾, λ _h ⁽¹⁾Changing trend diagram; Figure 10 and Figure 11 are respectively video traffic P _Drop ⁽²⁾With P _Block ⁽²⁾With professional arrival rate λ _n ⁽²⁾, λ _h ⁽²⁾Changing trend diagram; Figure 12 and Figure 13 are respectively data service P _Drop ⁽³⁾With P _Block ⁽³⁾With professional arrival rate λ _n ⁽³⁾, λ _h ⁽³⁾Changing trend diagram.

As follows for analysis of simulation result:

1. to the switching call drop probability P in the same type service _Drop ^(k)Stop probability P with new calling _Block ^(k)Variation tendency is done vertical contrast, and when professional arrival rate value was big, under the square one, comparison diagram 8 can get P with Fig. 9 _Block ⁽¹⁾Approximately than P _Drop ⁽¹⁾Exceed 0.7%; Contrast Figure 10 and Figure 11, P _Block ⁽²⁾Approximately than P _Drop ⁽²⁾Exceed 0.8%; Contrast Figure 12 and Figure 13, P _Block ⁽³⁾Approximately than P _Drop ⁽³⁾Exceed 0.9%.Therefore, can verify: when system resource was in short supply, for the business of any type, its switch call had higher priority than new the calling;

2. to the P between the different kinds of business _Drop ^(k)With P _Block ^(k)Variation tendency is done horizontal contrast, and under the situation of equal professional arrival level, comparison diagram 8, Figure 10 and Figure 12 can get, P _Drop ⁽²⁾With P _Drop ⁽³⁾Be in same level approximately, and P _Drop ⁽³⁾Whole value level is a little more than P _Drop ⁽²⁾But P _Drop ⁽²⁾With P _Drop ⁽³⁾The value level all be higher than P _Drop ⁽¹⁾When professional arrival rate value is big, P _Drop ⁽²⁾With P _Drop ⁽³⁾All than P _Drop ⁽¹⁾Exceed about 0.15%.Therefore, priority difference---the α between the different kinds of business ⁽¹⁾=0.5, α ⁽²⁾=0.3, α ⁽³⁾=0.2---verified;

If 3. to the curved surface in above-mentioned any one figure, respectively to plane (P _D/b ^(k), λ _n ^(k)) and plane (P _D/b ^(k), λ _h ^(k)) carry out projection (P _D/b ^(k)Represent axle P _Drop ^(k)Or axle P _Block ^(k)), can find that curved surface is at plane (P _D/b ^(k), λ _n ^(k)) in projection than curved surface at plane (P _D/b ^(k), λ _h ^(k)) in projection " gradient " (being similar to the notion of " slope " in the linear projection) bigger, the reason that produces this phenomenon is: because under the general status, switch call has higher priority than new the calling, therefore, P _Drop ^(k)Perhaps P _Block ^(k)Reach rate λ for new calling _n ^(k)Growth, " sensitivity " more.

For the fuzzy controller that degrades, present embodiment has carried out emulation to the variation tendency of its output controlled quentity controlled variable in all input values space, and the result especially, works as input as shown in figure 17 ${\mathrm{\&gamma;}}_i^D=0,$ Δ _Drop ^(k)When in [0.1,0.1], changing, corresponding output γ _I+1 ^DVariation tendency as shown in figure 14; In like manner, be input as when getting surely respectively ${\mathrm{\&gamma;}}_i^D=0.10$ With ${\mathrm{\&gamma;}}_i^D=0.20$ The time, corresponding output γ _I+1 ^DVariation tendency such as Figure 15 and shown in Figure 16.

As follows for analysis of simulation result:

1. output result shown in Figure 14, corresponding with the 1st row of the Fuzzy Rule Sets shown in the table 1; Output result shown in Figure 15, corresponding with the 2nd row of Fuzzy Rule Sets (simultaneously, also be subjected to the influence of other fuzzy rule, still, this moment the 2nd, row accounted for leading role); Output result shown in Figure 16, corresponding with the 4th row of Fuzzy Rule Sets (simultaneously, also be subjected to the influence of other fuzzy rule, still, this moment the 4th, row accounted for leading role).Above-mentioned corresponding relation is by investigating Figure 14, Figure 15, Figure 16 input parameter γ separately _i ^D, associative list 1 can obtain simultaneously;

2. just can find after as shown in figure 17 result being examined: the trellis raised grain that " rule " on the curved surface among the figure, arranged.This trellis texture is not difficult to find out the corresponding relation between the Fuzzy Rule Sets shown in Figure 17 and the table 1 owing to fuzzy set " rule " division is caused.Further observation can be found, among Figure 14, Figure 15 and Figure 16, all has " clue " of this trellis texture;

3. the discontinuous division of fuzzy set but can bring relatively continuously and level and smooth output (on the output curved surface, the slightly trellis texture of micro-protuberance only being arranged).Its main cause is to have used the operation of gravity model appoach de-fuzzy.This is also just meeting the expectation of the present invention for fuzzy degrade controller output accuracy and exquisiteness.

Claims (10)

1. the multi-service type call based on adaptive control allows control method, it is characterized in that, comprises the steps:

Step 1 is set up the Mathematical Modeling of system business, with the arrival process of all types of business of system as Poisson process, service process obeys index distribution, and computing service takies the average arrival rate and the average service rate of resource;

Step 2 is introduced traffic differentiation vector, the priority of differentiated service and resource requirement to the Mathematical Modeling of system business;

Step 3 is set up priority query's model, realizes professional resource contention when system resource is in short supply;

Step 4, priority query's model based on step 3, adopt counting process to describe the business of each type, and determine according to counting process whether switch call and new calling can obtain resource, and calculating switch call call drop probability stops probability with new the calling;

Step 5 is set up the adaptive fuzzy controlling models that degrades, and the control that degrades of existing call obtains next amplitude index that degrades constantly of existing business in the realization system when system resource is in short supply, dynamically adjusts user QoS and resource utilization ratio;

Step 6 is set up the optimal allocation model of the amplitude object that degrades, and when system resource is in short supply, in the existing business of system, optimally chooses degrade object and the corresponding amplitude that degrades thereof by making the target function value reach Minimal Realization.

2. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, described introducing traffic differentiation vector, specific as follows: as to belong to the calling of k class type of service for each, set up unique and the professional corresponding traffic differentiation vector (α of the type ^(k), β ^(k)) portray its service feature, wherein, k=1,2 ..., J, α ^(k)Be meant priority factor, β ^(k)Be meant the resource requirement coefficient,

Described priority factor α ^(k), being used among priority query's model, its value depends on the transmission delay requirement of k class business, priority factor α ^(k)Be worth greatly more, the priority that is had is high more, and then its real-time is strong more, and is strict more to the requirement of transmission delay; Priority factor α ^(k)Be worth more for a short time, the priority that is had is low more, and then its real-time is weak more, and is loose more to the requirement of transmission delay, priority factor α ^(k)Span is: 0≤α ^(k)≤ 1;

Described resource requirement factor beta ^(k), being used for adaptive fuzzy and degrading among controlling models and the optimal allocation model of amplitude object of degrading, its value depends on the demand for network bandwidth resources of k class business, β ^(k)Be worth greatly more, its demand to network bandwidth resources is high more, β ^(k)Be worth more for a short time, its demand to network money bandwidth source is low more.

3. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, and the described priority query's model of setting up, specific as follows: according to the priority factor α of type of service ^(k), set up the waiting list that has priority level accordingly, when the maximum bandwidth resource C of system has assigned, set up based on priority factor α ^(k)Waiting list, k=1,2 .., J, a total J parallel waiting list in this moment system, the priority of formation is respectively α ⁽¹⁾, α ⁽²⁾..., α ^(J), separate between each priority query, insert request for newly arrived switch call, according to its priority factor α ^(k)Value, it is added the corresponding priority level coefficient is α ^(k)The tail of the queue of waiting list, and newly arrived new calling inserts request, will directly be prevented from; When system resource is in short supply, be in the calling in the different priorities waiting list, obtain Internet resources with following common resource contention method.

4. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, the described resource contention of when system resource is in short supply, realizing business, specific as follows: as to be in the calling in the same priority waiting list, will be according to the regular compete for system resources of First Come First Served, be in the calling in the different priorities waiting list, the calling of individual queue head of the queue will be under the control of resource contention controller compete for system resources, being in priority factor is α ^(k)The handover call request of waiting list head of the queue, will be according to its priority factor α ^(k), with probability $P^{\left(k\right)}=\frac{{\mathrm{\&alpha;}}^{\left(k\right)}}{\underset{i}{\mathrm{\&Sigma;}}{\mathrm{\&alpha;}}^{\left(i\right)}}$ Success is competed and is obtained system resource, when a certain formation becomes sky, and its priority factor α ^(k)To be entered this formation up to next handover call request by temporary transient zero setting.

5. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, describedly determines that according to counting process whether switch call and new calling can obtain resource, are specially:

1. before the maximum bandwidth resource C of system all assigns, i.e. 0≤N ^(k)(t)＜and C, waiting list is empty, counting process N ^(k)(t) represented actual occupied system bandwidth resource, all types of new callings and switch call will be obtained the resource of network with identical priority, and this moment, the birth rate of birth and death process was λ ^T, the death rate is N ^(k)(t) μ ^T, μ ^TBe meant all types business take resource average service rate;

2. after the maximum bandwidth resource C of system all assigns, i.e. C≤N ^(k)(t)＜and C+Q, the new call request of all arrival will directly be prevented from, and the handover call request that arrives, will be according to its priority factor α ^(k), enter corresponding with it waiting list, so the birth rate of birth and death process is λ _h ^(k), the death rate is P ^(k)C μ ^T, promptly priority is α ^(k)The buffer of formation only receive k class switch call business, and the call business in this queue buffer will be with probability P ^(k)Come the resource of contention system;

3. work as N ^(k)(t) 〉=and during C+Q, system channel and buffer resources all assign, and at this moment, all new calling or switch calls all will be prevented from, for N ^(k)(t) 〉=and C+Q, birth rate is 0, for N ^(k)(t)=and C+Q, the death rate is P ^(k)C μ ^T, for N ^(k)(t)＞and C+Q, the death rate is 0.

6. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, described calculating switch call call drop probability stops probability with new the calling, is meant for k class business arbitrarily, lists about steady-state distribution probability π _j ^(k)State equation, j=0,1..., and then solve π _j ^(k)General expression formula, the switching call drop probability of k class business $P_{\mathrm{drop}}^{\left(k\right)}={\mathrm{\&pi;}}_{C+Q}^{\left(k\right)},$ The new calling of k class business stops probability $P_{\mathrm{block}}^{\left(k\right)}=\underset{n=C}{\overset{C+Q}{\mathrm{\&Sigma;}}}{\mathrm{\&pi;}}_n^{\left(k\right)},$ Wherein: C is the maximum bandwidth resource, and Q is meant the maximum bandwidth capacity of waiting list buffer.

7. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, the described adaptive fuzzy controlling models that degrades, its core is the fuzzy controller that degrades, and the fuzzy controller that degrades has two input parameters: 1. the switch call call drop probability of each type of service departs from the amplitude, ao that its maximum is allowed switch call call drop probability _Drop ^(k), ${\mathrm{\&Delta;}}_{\mathrm{drop}}^{\left(k\right)}=P_{\mathrm{drop},\mathrm{tolerance}}^{\left(k\right)}-P_{\mathrm{drop}}^{\left(k\right)},$ Wherein, P _Drop ^(k)Be switch call call drop probability, P _{Drop, tolerance} ^(k)For maximum is allowed switch call call drop probability, Δ _Drop ^(k)Reflected the QoS level that the user is current; 2. the current amplitude exponent gamma that degrades of existing business in the system ^D, $0\&le;{\mathrm{\&gamma;}}^D\&le;{\mathrm{\&gamma;}}_{\max }^D<1,$ Wherein, γ _Max ^DBe the upper limit of the amplitude that degrades, γ ^DThen reflected the utilance level that system resource is current; The fuzzy controller that degrades carries out obfuscation, fuzzy reasoning, de-fuzzy processing with above-mentioned two input parameters, and next amplitude index that degrades constantly of existing business is exported as controlled quentity controlled variable.

8. the multi-service type call based on adaptive control according to claim 7 allows control method, it is characterized in that, describedly carries out obfuscation, fuzzy reasoning, de-fuzzy and handles, and is specific as follows:

At first, defined its fuzzy set respectively for two input parameters that blur the controller that degrades, and corresponding membership function;

Then, on the cartesian product space of two input parameter fuzzy sets, define Fuzzy Rule Sets, set up 25 fuzzy inference rules; The input parameter will input to fuzzy reasoning mechanism after Fuzzy processing, behind fuzzy reasoning, obtain and export the corresponding fuzzy set vector of controlled quentity controlled variable;

At last, adopt gravity model appoach that this fuzzy set vector is carried out de-fuzzy, obtain exporting next amplitude index that degrades constantly of controlled quentity controlled variable existing business.

9. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, described dynamic adjustment user QoS and resource utilization ratio, specific as follows: the existing business in the system is before being degraded, take whole system resource C, when system resource is in short supply, is responsible for that the existing business in the system is carried out QoS and degrades, according to the output controlled quentity controlled variable γ of the fuzzy controller that degrades ^DIt is carried out after QoS degrades, and the existing business in the system only takies (1-γ ^D) system resource of C, d/d γ ^DThe system resource of C will offer by the successful call request of priority query's model competition.

10. the multi-service type call based on adaptive control according to claim 1 allows control method, it is characterized in that, and the degrade optimal allocation model of amplitude object of described foundation, specific as follows:

At first, choose the target function of quadratic function as model, wherein, each quadratic term has reflected the degrade departure degree of relative its primary standard demand resource of back institute's resource that obtains of all types of business respectively, simultaneously, the existing call number by all types of business is weighted stack with above-mentioned quadratic term;

Then, according to the relation of identity that total demand satisfied that degrades, the upper limit of the amplitude that degrades and lower limit requirement have provided the constraints of optimal allocation model, are specially:

The degrade actual relation of identity that degrades and satisfied between the amplitude total amount of total demand and all types of business of expression;

2. B ^(k)≤ β ^(k), expression waits to ask the upper limit of decision variable;

The amplitude upper limit that degrades of representing the k type of service promptly waits to ask the lower limit of decision variable;

Wherein, B ^(k)After the operation that degrades, the bandwidth resources that each existing call occupied of k class business; β ^(k)It is the resource requirement coefficient of k class business; N ^(k)Calling number for k class business existing in the system;

At last, find the solution the decision variable of optimal model, make the target function value reach minimum, obtain the degrade optimum allocation of amplitude of existing business in the system, its Mathematical Modeling is expressed as follows:

Find the solution decision variable: B _Degrade=[B ⁽¹⁾, B ⁽²⁾..., B ^(J)] ^T, to satisfy target function:

Images