CN102710508A - Virtual network resource allocation method - Google Patents

Abstract

The invention relates to the technical field of computer networks, and provides a virtual network resource allocation method. The method comprises the steps of: building a resource allocation model; submitting competitive demands to an InP (Internal Network Processor) through SPs (Service Providers); calculating the quantity and consumption of the resource through the InP; acquiring the resource, assuming the consumption and calculating the income through each SP; adjusting the competitive strategies through each SP; and waiting for competing at the next time based on the adjusted competitive strategies. In the scheme of the invention, a novel virtual network resource allocation method is provided; the resource is periodically allocated to a plurality of service providers based on the trend of the demands on the virtual network resource, so that the resource allocation can be carried out as demands, and the performance of the whole network can be improved. Moreover, a scheme for effectively selecting the competitive strategies is provided at the same time; by using the scheme, the service providers can be guided to select the competitive strategies; the real network demands can be quickly obtained; the network performance can be improved; and the complexity of the algorithm can be effectively reduced.

Description

The virtual network resource allocation methods

Technical field

The present invention relates to technical field of the computer network, particularly a kind of virtual network resource allocation methods.

Background technology

In the network virtualization environment, traditional Internet Service Provider is divided into two parts: infrastructure provider (InP) and service provider (SP).InP is responsible for disposing and management bottom-layer network, i.e. bottom physical resource; SP is through renting the virtual network of resource construction oneself, for terminal use (EU) provides service to InP.In order fully effectively to utilize the resource of physical network, for the user of virtual network provides convenience and the service of quality assurance is arranged, the resource allocation of virtual network has become an inevitable major issue.

In the prior art, the virtual network resource allocation methods mainly is divided into two types, static allocation and dynamic assignment: static allocation is that virtual network has distributed resource before its life cycle finishes, not allow the change in resources of being distributed afterwards; Common static allocation mode can be shone upon when initialization, such as taking nearby principle to select the bottom physical network nodes.Dynamic assignment then can dynamically be adjusted institute's Resources allocation according to running status and changes in demand in the virtual network life cycle; Common dynamic assignment mode can be according to the dynamic resource demand, and further consider that dynamic network model, periodicity emphasis distribute or the fairness doctrine etc. carried out resource distribution.

But; The dumb performance and the efficient that has obviously had influence on network of static resource allocation; It does not consider the resource requirement possibility dynamic change of each virtual network usually; Cause initial virtual network mapping result can't satisfy the dynamic need of virtual network, possibly cause virtual network normally to move.And that the algorithm of dynamic assignment is compared static allocation is more flexible and efficient; But more complicatedly usually dispose to realize with being difficult to; Such as; The topological structure that needs virtual network in the frequent updating model of model Network Based, periodicity emphasis distribute does not consider the harm that the greedy behavior of selfish virtual network brings for whole network environment, and the fairness doctrine can't be applicable to the complex environment that service type is more.

Summary of the invention

The technical problem that (one) will solve

To the shortcoming of prior art, the present invention is in order to solve the problem that virtual network resource distribution mode in the prior art is dumb or be difficult to realize, a kind of novel virtual network resource allocation methods that dynamically carries out is provided.

(2) technical scheme

Solve the problems of the technologies described above, the present invention specifically adopts following scheme to carry out for this reason:

At first, the present invention provides a kind of virtual network resource allocation methods, and said method comprises step: 101) the Resource Allocation in Networks participant is carried out modeling, make up the Dynamic Resource Allocation for Multimedia model;

102) participate in the rival demand information of a plurality of service providers of resource contention to infrastructure provider submission one dimension, this service provider was to the demand of resource when this competition was represented in this rival demand;

103) infrastructure provider receives the demand information that all participate in the service provider of resource contentions, calculation services provider obtainable stock number and the required loss of bearing;

104) each service provider obtains the Internet resources that infrastructure provider distributes, and bears loss and calculates the income that self effectiveness is improved;

105) each service provider adjusts its competitive strategy respectively according to loss and income;

106) each service provider adopts pending next round resource contentions such as adjusted competitive strategy.

Preferably, in the step 103, the account form of said stock number is:

$\underset{x}{\max }{\mathrm{\Σ}}_{i=1}^m{F}_i\left(x_i\right)$

s.t.Ax≤C

Wherein, vector x is represented the network bandwidth that all service providers are assigned with, x _iExpression provides the service provider SP of service i _iThe network bandwidth that is assigned with, m are represented service provider's number, and vectorial C representes the capacity of all links, the elements A in the matrix A _LiExpression SP _iWhether the path at place comprises link l, F _i(x _i) be SP _iUtility function,

Wherein, b _iExpression SP _iRival demand information during this competition, K _iUser's set of expression subscribed services i, α _kBe the weight of user k, u _kThe satisfaction function of expression user k.

Preferably, in the step 103, SP _iThe required loss of bearing is:

${\mathrm{\τ}}_i={\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

Wherein, is the optimal solution of equation

s.t.Ax≤C; When

has neither part nor lot in allocated bandwidth for service i, the optimal solution of this equation.

Preferably, in the step 104, SP _iThe income that self effectiveness is improved is:

${\mathrm{\ψ}}_i=F_i\left(x_i^{*}\right)-{\mathrm{\τ}}_i={\mathrm{\θ}}_i{U}_i\left(x_i^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)+{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

Wherein, θ _iBe SP _iReal demand, function

Preferably, in the step 105, the adjustment algorithm of competitive strategy is:

201) init state-action schedule SA is provided with algorithm parameter;

202) perception current network environment is subscribed to the situation of this service and is represented the current state s of living in of service provider according to the resource provision amount that perceives, the situation that self takies resource and terminal use;

203) based on state of living in, the service provider adopts ε-greedy algorithm to concentrate from competitive strategy and chooses the action a that a strategy constitutes the service provider;

204) the immediately award of calculation services provider from behind state s execution a, obtaining;

205) current state-action in update mode-action schedule (s, a) to pairing value SA (s, a);

206) carry out action a and get into NextState s';

207) whether evaluation algorithm restrains, if algorithmic statement then obtains the optimal policy value, submits its optimum competitive strategy to infrastructure provider; If algorithm is not restrained, then proceed Action Selection next time.

Preferably, in the step 204, said award immediately is poor for service provider's profit under the state s and the following service provider's profit of state s '.

Preferably, in the step 205, said value SA (s a) is:

SA(s,a)＝SA(s,a)+β·(R(s,a)+γ·E(s')-SA(s,a))

Wherein, s' is the state of service provider after state s executes action a; E (s ') is the effectiveness of state s', E (s')=max _aSA (s', a); γ is the discount rate of time-delay repayment E (s'), γ ∈ [0,1]; β is a learning rate, has indicated the degree of belief of giving to the updated portion of improving, β ∈ [0,1].

Preferably, in the step 207, upgrade the front and back difference according to definite state-action schedule and whether come evaluation algorithm whether to restrain less than certain threshold value.

Preferably, in the step 207, said threshold value is 10 ^-4

Preferably, in the step 106,, competitive strategy participates in competition next time if having adjustment then to jump to step 102; Otherwise continue to wait for.

(3) beneficial effect

In scheme of the present invention, a kind of novel virtual network resource allocation methods is provided, to the dynamic of virtual network resource requirement, periodically Resources allocation is given a plurality of service providers, makes resource allocation carry out as required, improves overall performance of network.Owing to also proposed a kind of scheme of effective selection competitive strategy among the present invention simultaneously, through this scheme, competitive strategy is selected by guide service provider, obtains the live network demand fast, improves network performance, effectively reduces algorithm complex.

Description of drawings

Fig. 1 is the schematic flow sheet of virtual network resource allocation methods in the embodiments of the invention;

Fig. 2 is the schematic flow sheet of Competitive Tactics'Choice algorithm in the preferred embodiments of the present invention;

Fig. 3 is the structural representation of network model in the specific embodiment of the present invention.

Embodiment

To combine the accompanying drawing in the embodiment of the invention below, the technical scheme in the embodiment of the invention is carried out clear, intactly description, obviously, described embodiment is a part of embodiment of the present invention, rather than whole embodiment.Based on the embodiment among the present invention, the every other embodiment that those of ordinary skills are obtained under the prerequisite of not making creative work belongs to the scope that the present invention protects.

The present invention provides a kind of novel virtual network resource allocation methods in order to solve the defective of static state in the prior art or Dynamic Resource Allocation for Multimedia mode, and to the dynamic of virtual network resource requirement, periodically Resources allocation is given a plurality of service providers.Owing to also proposed a kind of scheme of effective selection competitive strategy among the present invention simultaneously, through this scheme, competitive strategy is selected by guide service provider, obtains the live network demand fast, effectively reduces algorithm complex.

At first, referring to Fig. 1, the general flow of virtual network resource allocation methods of the present invention is:

101) set up resource allocator model: the Resource Allocation in Networks participant is carried out modeling, make up the Dynamic Resource Allocation for Multimedia model.

Three types of roles are arranged in resource allocator model, be respectively the InP of infrastructure provider, service provider SP and terminal use EU.InP is unique in the model, be responsible for to dispose and the managing physical Internet resources, for a plurality of service providers (following represent a plurality of service providers with the SPs that abridges) provide resource; SPs is responsible for disposing and the managing virtual network, and each virtual network provides service by unique SP management to a plurality of terminal uses.Suppose to have m SP, each SP provides a kind of service, is expressed as SP={1 ..., m}, according to the service that the terminal use orders, the user is assigned to the m group, and every group corresponding to a kind of service (being provided by service provider SP).Subscribed services i is (by service provider SP _iProvide) user set by K _iExpression supposes that each user only orders a service the same time period, and then the total number of users of system does

102) SPs submits competition information to InP: participate in the demand information of the SPs of resource contention to InP submission one dimension, this demand information is represented the demand of SP to resource.

SP in this mechanism _iAll Policies constitute set of strategies B _i, consider that set of strategies is the situation of discrete set, suppose SP _iN is arranged _iIndividual different strategies, promptly

Definition SP _iReal demand be θ _i, expression SP _iFor the real demand amount of the network bandwidth, B _iIn comprise SP _iReal demand θ _i, i.e. θ _i∈ B _iSP _iDemand strategy is b during each the competition _i, i.e. b _i∈ B _i, to take turns in the competition process every, the demand information of all SP represented by m dimensional vector b, i.e. b={b ₁..., b _m.

103) InP calculating SPs gained resource and required loss: InP receive the demand information that all participate in the SPs of resource contentions, calculate the obtainable stock number of SPs.

Wherein, the stock number computational methods do

$\underset{x}{\max }{\mathrm{\Σ}}_{i=1}^m{F}_i\left(x_i\right)$

s.t.Ax≤C

Wherein, vector x is represented the network bandwidth that all SPs are assigned with, x _iExpression SP _iThe network bandwidth that is assigned with, vectorial C representes the capacity of all links, the elements A in the matrix A _LiExpression SP _iWhether the path at place comprises link l, F _i(x _i) be SP _iUtility function, by SP _iDemand information and user satisfaction function determine jointly, Wherein, α _kBe the weight of user k, u _kThe satisfaction function of expression user k is established F then _i(x _i)=b _iU _i(x _i).

InP calculates the loss that SPs brings to system according to the stock number of SPs gained, and its value is SP _iWhen not adding network in the network utility function of all SPs with, deduct and use SP _iThe utility function that adds other SPs behind the network with, computational methods do

${\mathrm{\τ}}_i={\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

Wherein,

is the optimal solution of equation

s.t.Ax≤C; Promptly

be when having neither part nor lot in allocated bandwidth for i, the optimal solution of this equation.

104) SP _iObtain the Internet resources that InP distributes, bear loss and calculate the income that self effectiveness is improved, computational methods do

${\mathrm{\ψ}}_i=F_i\left(x_i^{*}\right)-{\mathrm{\τ}}_i={\mathrm{\θ}}_i{U}_i\left(x_i^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)+{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

105) SPs adjusts its competitive strategy respectively according to loss and income.

106) SPs adopts pending next round resource contentions such as adjusted demand information.Wherein, if having adjustment then to jump to step 102, competitive strategy participates in competition next time; Otherwise continue to wait for.

Further referring to Fig. 2, the Competitive Tactics'Choice algorithm is preferably adopted in the adjustment of competitive strategy in the step 105, and the general flow of this algorithm is:

201) init state-action schedule, promptly the SA table is provided with each parameter.According to the policy selection algorithm, (s, a) corresponding to a SA value, promptly (s, a), expression state s executes and hopes the accumulation repayment (depending on the current time-delay of repaying immediately and expecting repayment) that obtains behind a SA for the every pair of state-action.The SA table is initialized as 0; Algorithm parameter β and γ are set, and γ (γ ∈ [0,1]) is the discount rate of time-delay repayment E (s'); β (β ∈ [0,1]) is a learning rate, has indicated to give how many degree of beliefs of updated portion of improving.

202) observe current environment state s.SP perception current network environment, the situation that situation that takies resource according to the InP resource provision amount that perceives, SP self and terminal use subscribe to this service is represented the current state s of living in of SP.Making up the state space that forms by all state s is the state space of Discrete Finite, meets the requirement of policy selection algorithm to state space basically, also can in native system, reflect current network conditions fully simultaneously.

203) select action a.Based on state s of living in, SP is from competitive strategy collection B _iIn choose a tactful b _jBehavior corresponding to one of SP action a, the set of a of then moving can be expressed as

(ε-greedy), under current state s, select to move a at random with probability ε, with the action that the selection of 1-ε probability has maximum SA value, system of selection is a=argmax to adopt ε-greedy algorithm _aSA (s, a).

204) R is repaid in calculating immediately.The foundation of rewards and punishments function R is very crucial, is determining the SA value that each state-action is right.In this algorithm, the rewards and punishments function adopts actively repayment, and promptly the income of action is big more, and the repayment R that obtains (s, a) big more, just reward big more.SP from the R of award immediately that behind state s execution a, obtains (s, a), i.e. SP under the state s _iIn timeslice t and next timeslice t+1 poor (be said poor for service provider's profit under the state s and the following service provider's profit of state s ' of rewarding immediately, the income calculation mode is seen step 104 explanation) of profit, computational methods do

$R(s,a){\mathrm{\ψ}}_i^{t+1}(b_i,x_i)-{\mathrm{\ψ}}_i^t(b_i,x_i)$

205) upgrade the SA table.Upgrade in the SA table, (s, a) to pairing SA value, update method does current state-action

SA(s,a)＝SA(s,a)+β·(R(s,a)+γ·E(s′)-SA(s,a))

Wherein, s' is the state of SP after state s executes action a; E (s ') is the effectiveness of state s', i.e. E (s')=max _aSA (s', a); γ (γ ∈ [0,1]) is the discount rate of time-delay repayment E (s'); β (β ∈ [0,1]) is a learning rate, has indicated to give how many degree of beliefs of updated portion of improving.

206) carry out action a and get into NextState s'.

207) whether evaluation algorithm restrains.SP implementation strategy selection algorithm if algorithmic statement promptly obtains the optimal policy value, is then submitted its optimum competitive strategy to InP; If algorithm is not restrained, then proceed Action Selection next time.Judge convergence method for difference before and after upgrading when the SA table less than certain threshold value (as 10 ^-4).

Part of links structure with a real network is that example is showed resource allocation methods of the present invention below.As shown in Figure 3, network environment is set to SP ₁With SP ₂The Internet resources of share I nP, SP ₁And SP ₂All pass through link N _AN _B, the bandwidth of this link is 20Mbps, node N _C, N _DBe used for service being provided to the user.The total number of users of system is 10, and the user that each user subscribes among a certain service and the same SP at random has equal weight.SP ₁The bidding strategies collection be B ₁=6,8,10,12, and 14}, wherein 10 is SP ₁The true bandwidth demand; SP ₂The bidding strategies collection be B ₂=7,9,11,13, and 15}, wherein 11 is SP ₂The true bandwidth demand.

Based on above-mentioned network example, the detailed implementation step of the virtual network resource allocation methods that the present invention proposes is following:

Set up resource allocator model.The InP of supplying party is unique, resource provision amount 20Mbps; The SP of party in request number is 2, and resources requirement is the demand information of its submission.

SPs submits demand information to InP respectively, and the InP Resources allocation is informed the required loss of SPs, and SPs obtains resource, bears loss, calculates effectiveness and improves income.Wherein, the system situation after the first competition is as shown in table 1:

SP	Demand	Bandwidth	Loss	Income
					1	12	9.6000	8.5010	14.1166
2	13	10.4000	8.8077	16.9522

The first competition of table 1 back system status

SPs adjusts demand.SPs selects demand through SA learning compete policy selection algorithm.With SP ₁Be example, demand selects concrete steps following:

1. according to SP ₁Strategy set B ₁=6,8,10,12, and 14}, the division state space is S={s ₁, s ₂, s ₃, s ₄, s ₅, the motion space is A ₁=6,8,10,12,14}.Initialization SA table is SA _{5 * 5}=0, it is 0.8 that parameter beta is set, and γ is 0.3.The random initializtion state is s2.

2. observe current environment state s.

3. select action a.According to ε-greedy algorithm from behavior aggregate A ₁The middle selection moves 8.

4. calculate and repay R immediately.R(s ₁,a)=0.0872。

5. upgrade the SA table.Upgrade the corresponding SA value of current state-action, i.e. SA (s ₁, a)=0.2758.

6. carry out action a and get into NextState s'.

7. whether evaluation algorithm restrains.The next stage competition is then waited in convergence, otherwise, forward to and 2. proceed to learn next time.The judgement convergence method is for thinking that when the SA table upgrades the front and back difference less than threshold value (10-4) SA shows to restrain.

SP ₁After 1300 study, its SA table convergence, as shown in table 2:

Table 2SP ₁SA table convergence situation after study

SA table through convergence can be known SP ₁Competitive strategy 10 is an optimal policy.Through the competition selection algorithm, the requirement vector of participating in the SPs of resource allocation is b={10,11}.Resource allocation conditions, its result is as shown in table 3:

SP	Demand	Bandwidth	Loss	Income
					1	10	9.5238	7.1129	15.4251
2	11	10.4762	7.4194	18.4208

Table 3 is competed back system status once more

The contrast demand is respectively that { 12,13} is with { 10, resource allocation result can be known during 11}, and when the SPs demand be its real demand, resource allocation result was optimum, the total revenue maximum of SPs.Above process has been showed a kind of resource allocation overall process of virtual network resource allocation methods.Under the prerequisite of fair and reasonable allocation of network resources, improved the total revenue of SPs.

In sum, the invention provides a kind of virtual network resource allocation methods, beneficial effect of the present invention is:

1) effectively suppressed the selfishness property of SP.Calculate SP through the loss function and obtain the required loss of bearing of resource, when SP lies about demand, cause loss to increase considerably, the honest reflection demand of excitation SP based on this function.The SP of all participations all is willing to report real demand, thereby makes resource allocation carry out as required, improves overall performance of network.

2) improvement of resource allocation participant overall utility is improved.Because the competition mechanism of using in this method itself satisfies the compatible characteristic of dominant strategy excitation, this method excitation SP reports real demand, in fair and reasonable Resources allocation, makes SPs gained total utility improve.And the rival demand information that adopts one dimension is represented the resource requirement of SPs, has further alleviated the flow load in the network.

3) the Competitive Tactics'Choice algorithmic statement is fast, and SPs selects dominant strategy fast, and it is consuming time effectively to reduce resource allocation.

Above execution mode only is used to explain the present invention; And be not limitation of the present invention; The those of ordinary skill in relevant technologies field under the situation that does not break away from the spirit and scope of the present invention, can also be made various variations and modification; Therefore all technical schemes that are equal to also belong to category of the present invention, and real protection scope of the present invention should be defined by the claims.

Claims (10)

1. a virtual network resource allocation methods is characterized in that, said method comprises step:

101) the Resource Allocation in Networks participant is carried out modeling, make up the Dynamic Resource Allocation for Multimedia model;

102) participate in the rival demand information of a plurality of service providers of resource contention to infrastructure provider submission one dimension, this service provider was to the demand of resource when this competition was represented in this rival demand;

103) infrastructure provider receives the demand information that all participate in the service provider of resource contentions, calculation services provider obtainable stock number and the required loss of bearing;

104) each service provider obtains the Internet resources that infrastructure provider distributes, and bears loss and calculates the income that self effectiveness is improved;

105) each service provider adjusts its competitive strategy respectively according to loss and income;

106) each service provider adopts pending next round resource contentions such as adjusted competitive strategy.

2. method according to claim 1 is characterized in that, in the step 103, the account form of said stock number is:

$\underset{x}{\max }{\mathrm{\Σ}}_{i=1}^m{F}_i\left(x_i\right)$

s.t.Ax≤C

Wherein, vector x is represented the network bandwidth that all service providers are assigned with, x _iExpression provides the service provider SP of service i _iThe network bandwidth that is assigned with, m are represented service provider's number, and vectorial C representes the capacity of all links, the elements A in the matrix A _LiExpression SP _iWhether the path at place comprises link l, F _i(x _i) be SP _iUtility function,

Wherein, b _iExpression SP _iRival demand information during this competition, K _iUser's set of expression subscribed services i, α _kBe the weight of user k, u _kThe satisfaction function of expression user k.

3. method according to claim 2 is characterized in that, in the step 103, and SP _iThe required loss of bearing is:

${\mathrm{\τ}}_i={\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

Wherein,

is the optimal solution of equation

s.t.Ax≤C; When has neither part nor lot in allocated bandwidth for service i, the optimal solution of this equation.

4. method according to claim 3 is characterized in that, in the step 104, and SP _iThe income that self effectiveness is improved is:

${\mathrm{\ψ}}_i=F_i\left(x_i^{*}\right)-{\mathrm{\τ}}_i={\mathrm{\θ}}_i{U}_i\left(x_i^{*}\right)-{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_{j,-i}^{*}\right)+{\mathrm{\Σ}}_{j=1,j\≠i}^m{F}_j\left(x_j^{*}\right)$

Wherein, θ _iBe SP _iReal demand, function

5. method according to claim 1 is characterized in that, in the step 105, the adjustment algorithm of competitive strategy is:

201) init state-action schedule SA is provided with algorithm parameter;

202) perception current network environment is subscribed to the situation of this service and is represented the current state s of living in of service provider according to the resource provision amount that perceives, the situation that self takies resource and terminal use;

203) based on state of living in, the service provider adopts ε-greedy algorithm to concentrate from competitive strategy and chooses the action a that a strategy constitutes the service provider;

204) the immediately award of calculation services provider from behind state s execution a, obtaining;

205) current state-action in update mode-action schedule (s, a) to pairing value SA (s, a);

206) carry out action a and get into NextState s';

207) whether evaluation algorithm restrains, if algorithmic statement then obtains the optimal policy value, submits its optimum competitive strategy to infrastructure provider; If algorithm is not restrained, then proceed Action Selection next time.

6. method according to claim 5 is characterized in that, in the step 204, said award immediately is poor for service provider's profit under the state s and the following service provider's profit of state s '.

7. method according to claim 5 is characterized in that, in the step 205, said value SA (s a) is:

SA(s,a)＝SA(s,a)+β·(R(s,a)+γ·E(s')-SA(s,a))

Wherein, s' is the state of service provider after state s executes action a; E (s') is the effectiveness of state s', E (s')=max _aSA (s', a); γ is the discount rate of time-delay repayment E (s'), γ ∈ [0,1]; β is a learning rate, has indicated the degree of belief of giving to the updated portion of improving, β ∈ [0,1].

8. whether whether method according to claim 5 is characterized in that, in the step 207, upgrade the front and back difference according to definite state-action schedule and come evaluation algorithm to restrain less than certain threshold value.

9. method according to claim 8 is characterized in that, in the step 207, said threshold value is 10 ^-4

10. method according to claim 1 is characterized in that, in the step 106, participates in competition next time if competitive strategy has adjustment then to jump to step 102; Otherwise continue to wait for.

Images