CN107168797A - Resource regulating method based on dynamic game under cloud environment - Google Patents
Resource regulating method based on dynamic game under cloud environment Download PDFInfo
- Publication number
- CN107168797A CN107168797A CN201710335668.5A CN201710335668A CN107168797A CN 107168797 A CN107168797 A CN 107168797A CN 201710335668 A CN201710335668 A CN 201710335668A CN 107168797 A CN107168797 A CN 107168797A
- Authority
- CN
- China
- Prior art keywords
- resource
- user
- game
- cloud environment
- scheduling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5077—Logical partitioning of resources; Management or configuration of virtualized resources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/5021—Priority
Abstract
The invention belongs to the scheduling of resource technical field based on cloud computing, and in particular to the resource regulating method based on dynamic game under a kind of cloud environment, comprise the following steps:Build the resource dispatching model based on dynamic game under cloud environment;Resource dispatching model based on dynamic game, the priority for selecting resource is distributed according to the attribute of each user task height;Take task attribute identical user first to file first to select the strategy of allocation priority, and the income of each user is defined as to its QoS satisfaction;The Nash Equilibrium Solution of game is solved using reverse induction.The present invention is directed to multiple users resource competing problem caused by submission task simultaneously under cloud environment, set up cloud environment dynamic game resource dispatching model, the resource contention between each user is modeled and analyzed using dynamic game theory, the QoS demand of all users is met as much as possible;So as to meet the demand of each user to the full extent;The Nash Equilibrium Solution of Dynamic Game Model is solved using reverse induction, it is achieved thereby that resource most rationally effective configuration.
Description
Technical field
The invention belongs to the scheduling of resource technical field based on cloud computing, and in particular to rich based on dynamic under a kind of cloud environment
The resource regulating method played chess.
Background technology
Cloud computing is a kind of emerging calculating mould grown up in the technical foundation such as grid computing, parallel computation and P2P
Type.A large amount of computing resources, storage resource and the software resource that it will be stored on data center's cluster link together (including
The software resources such as CPU, internal memory, the hardware resource such as hard disk, and development environment, application program), carry out centralized and unified management, shape
Into large-scale shared virtualization pool, it is supplied to user to use on demand, provides " be ready to assemble at the first call and leave at the wave of a hand " for it and seem
The cloud service of " ability is unlimited ".With developing rapidly for distributed computing technology, cloud computing has become a crucial research field.
Cloud computing is built upon on architecture (Service Oriented Architecture, SOA), is carried in the way of service
For computing resource, user can obtain cloud service on demand.
With the fast development of cloud computing, vast resources are incorporated in cloud, and user's request is continuously increased, and cloud computing center is provided
Service be on the increase, be faced with huge scheduling of resource pressure.Because cloud computing system is in a dynamic environment, and
And user group is very huge, each application program has the resource requirement constantly changed, causes its performance needs to move
State is satisfied[8].On the other hand, cloud computing shields the complexity of bottom hardware using virtualization technology, improves flexibly
Property.Under multi-user environment, the application program of a large number of users is operated on a physical equipment, shares hardware and storage device.
The demand Resource supply of cloud computing and multi-user resource are shared, and cause cloud computing resources scheduling problem to turn into after safety in cloud computing
The second hang-up after problem.
During cloud computing service scheduling of resource, multiple users successively submit task requests, according to the when segmentum intercalaris of request
Point can be divided into multiple user's groups.They require to use same class resource, but ensure what is first submitted in scheduling process again
User's group can not monopolize such resource, and which forms the resource contention between different user groups.In order to meet each user's group
Demand, meet constraints such as task completion time, completion required by task expense etc. of user's group proposition, while reaching money
Source most rationally effectively uses and configured, it is necessary to set up the competitive model based on dynamic game, design it is a kind of it is fair effectively
Resource regulating method, the race problem for solving multi-user's non-concurrent application resource in cloud service scheduling of resource both improved
Scheduling of resource efficiency ensures resource allocation fairness again.
Prior art typically uses the scheduling of resource based on intelligent optimization algorithm, from control centre's overall interests,
Intelligent algorithm applies to cloud computing resources scheduling, improves scheduling of resource efficiency, it is ensured that the performance of system.But weak point is
Based on control centre's overall interests, it is impossible to meet the resource requirement of all users, it is impossible to ensure resource allocation fairness.
Another is the scheduling of resource based on game theory, meets individual rationality and excitation compatibility feature, weak point
It is undue pursuit cloud computing resources price, does not ensure the other demands of user, i.e., do not ensure the service quality QoS of user
(Quality of Service)。
Game theory is one of most basic theory of economics, is asked for solving the competition between Different Individual with mutual containing
Inscribe, be considered as the maximum achievement of 20th century domain of the social sciences acquirement in Western society science.And game theory is to be used to solve
Certainly the competition between Different Individual and mutual containing problem, are considered as 20th century domain of the social sciences in Western society science
The maximum achievement of acquirement.Game theory is modeled as the theory of research participant's competition to the competition between Different Individual
There is very high value with analysis aspect.Competition for multi-user's non-concurrent application resource in cloud service scheduling of resource is asked
Topic, the present invention proposes a kind of rationally effective resource dispatching model and method, for solving in cloud environment based on dynamic game opinion
Multi-user resource race problem, improves scheduling of resource efficiency.
The content of the invention
The present invention for scheduling of resource of the prior art based on intelligent optimization algorithm exist using control centre's overall interests as
It is main, it is impossible to meet the resource requirement of all users, it is impossible to ensure that resource allocation fairness problem and the resource based on game theory are adjusted
, there is undue pursuit cloud computing resources price in degree, do not ensure the other demands of user, i.e., do not ensure that the service quality of user is asked
Topic, proposes the resource regulating method based on dynamic game under a kind of cloud environment.
The technical scheme is that:Resource regulating method based on dynamic game under a kind of cloud environment, including following step
Suddenly:
Step 1:Build the resource dispatching model based on dynamic game under cloud environment;
Step 2:Resource dispatching model based on dynamic game, selection resource is distributed according to the attribute of each user task height
Priority;
Step 3:First to file is taken task attribute identical user first to select the strategy of allocation priority, and by each user
Income be defined as its QoS satisfaction;
Step 4:The Nash Equilibrium Solution of game is solved using reverse induction.
Scheduling of resource under cloud environment in resource regulating method based on dynamic game under described cloud environment, the step 1
Refer on the basis of resource virtualizing, the computing resource that each user performs required by task is dynamically distributed at scheduling of resource center, is
Execution task provides feasibility environment;Under cloud environment during scheduling of resource, physical resource is virtualized into virtual list by cloud system
Member, is used as the carrier of the task of execution.
Resource regulating method based on dynamic game under described cloud environment, the cloud environment dynamic game scheduling of resource mould
Type is a five-tuple, i.e. MCED-GRSM=(N, T, S, I, U), wherein:N=(N1,N2,…,Nn) be game participant collection
Close, participant is the users of sometime all transmission task requests of fragment;T=(T1,T2,…,Tn) be participant action it is suitable
Sequence, i.e., select the sequencing of resource in sometime fragment user;S=(S1,S2,…,Sn) be participant policy space,Represent participant NiPolicy space, each participant should have more than a kind of plan
Omit, i.e. h >=1;
I=(I1,I2,…,In) be participant information collection,Represent participant Ni
Information collection, each participant should have more than a kind of strategy, i.e. k >=1;
U=(U1,U2,…,Un) be participant revenue function set,
Represent user n in user 1,2 ..., selection strategy s after n-1 actionnIncome;Revenue function represents participant from game
Available income level, is together decided on by the strategy of all participants, the receipts obtained by the different strategy combination of participant
It is beneficial different.
Resource regulating method based on dynamic game under described cloud environment, the scheduling model meets three it is assumed that i.e.:
Rational conditions:Assuming that each user is rational, meaningless, malice will not be sent because of maximum return to be obtained and takes money
The task requests in source;Type is assumed:Assuming that the demand of different user is different, the QoS of proposition is different;Income is assumed:Assuming that all
The target of user is all on the premise of oneself QoS is met, to ask resource to complete task, the target at scheduling of resource center is to the greatest extent
The demand that all users propose may be met, feedback resources complete its task.
The income of user is defined as in resource regulating method based on dynamic game under described cloud environment, the step 3
Its QoS satisfaction is weighed using cloud environment scheduling of resource game theory, and cloud environment scheduling of resource game theory is with a triple
Represent that (N, S, U) is represented, wherein N represents all node sets, represents the set of all users in scheduling of resource;S is game theory
The set of middle directed edge, represents the strategy of user in scheduling of resource;U is then the set of user's income, is represented under Different Strategies
The income of acquirement.
Resource regulating method based on dynamic game under described cloud environment, the satisfaction of the QoS receipts of each user
Benefit is weighed, i.e. Ui(s1,…,si)=ε1Q1+ε2Q2+...+εnQn+ ..., wherein Q represents Service Quality Metrics, ε1,ε2,…,εn
Represent weight of the user to all kinds of indexs, ε1,ε2,…,εn∈(0,1)。εiQiUser is represented to i resource type service quality
Satisfaction;And the income at scheduling of resource center is the total revenue of all users, i.e.,:
Resource regulating method based on dynamic game under described cloud environment, the step 4 is specially:From game terminal section
Directly forward section start, the method then inversely concluded by game theory is referred to as the reverse induction in game;Solution is
Since the bottom of extension game theory, it is considered to user n subgame, if user's n-1 selection strategiesThen come for user n
Say selection strategyBetter than other any one strategies are selected, likewise, when user's n-1 selection strategiesWhen, user n is most
Dominant strategy isEach user knows these information, therefore user n-1 can selection strategyTo maximize the interests of oneself, this
Sample, solves sub- Equilibrium Game solutionBy that analogy, Equilibrium Game solution is finally solved
Resource regulating method based on dynamic game under described cloud environment, the cloud environment dynamic game scheduling of resource mould
In type, strategyIt is a Nash Equilibrium, if for all i ∈ N,It isOne it is optimal
Reaction, is for all si∈SiWith all i ∈ N, have:Wherein,
Represent that user i selects optimal policy after user i-1 actionIncome,Represent that user i takes action in user i-1
Non-optimal tactful s is selected afterwardsiIncome.
The beneficial effects of the invention are as follows:The present invention is for multiple users under cloud environment caused by submission task simultaneously
Resource competing problem, sets up cloud environment dynamic game resource dispatching model, using dynamic game theory between each user
Resource contention is modeled and analyzed, and the QoS demand of all users is met as much as possible;Based on weight of the user to all kinds of indexs
Revenue function is set up, it is comprehensive to consider each user to the type of resource and the difference of desirability;Devise optimal resource choosing
Strategy is selected, the high user of preferential support mission rank first selects task resource, and identical user takes first Shen for task rank
The principle please first selected, so as to meet the demand of each user to the full extent;Dynamic game is solved using reverse induction
The Nash Equilibrium Solution of model, it is achieved thereby that resource most rationally effective configuration.
Brief description of the drawings
Fig. 1 is dispatching method block diagram of the invention;
Fig. 2 is cloud environment scheduling of resource game theory structural representation;
Fig. 3 is cloud environment scheduling of resource structural representation;
Fig. 4 is the cloud environment scheduling of resource game theory structural representation of high-level task;
Fig. 5 is low level task cloud environment scheduling of resource game theory structural representation;
Embodiment
Embodiment 1:With reference to Fig. 1-Fig. 5, the resource regulating method based on dynamic game under a kind of cloud environment, including following step
Suddenly:
Step 1:Build the resource dispatching model based on dynamic game under cloud environment;Step 2:Resource based on dynamic game
Scheduling model, the priority for selecting resource is distributed according to the attribute of each user task height;Step 3:To task attribute identical
User takes first to file first to select the strategy of allocation priority, and the income of each user is defined as to its QoS satisfaction;Step
4:The Nash Equilibrium Solution of game is solved using reverse induction.
Specifically, scheduling of resource refers on the basis of resource virtualizing under cloud environment in step 1, scheduling of resource center dynamic
Ground distributes the computing resource that each user performs required by task, and feasibility environment is provided to perform task;Scheduling of resource under cloud environment
During, physical resource is virtualized into dummy unit by cloud system, is used as the carrier of the task of execution.
Specifically, cloud environment dynamic game resource dispatching model is a five-tuple, i.e. MCED-GRSM=(N, T, S, I,
U), wherein:N=(N1,N2,…,Nn) be game participant set, participant be sometime all transmission tasks of fragment please
The user asked;T=(T1,T2,…,Tn) be participant action order, i.e., sometime fragment user selection resource priority
Sequentially;S=(S1,S2,…,Sn) be participant policy space,Represent participant Ni
Policy space, each participant should have more than a kind of strategy, i.e. h >=1;
I=(I1,I2,…,In) be participant information collection,Represent participant Ni
Information collection, each participant should have more than a kind of strategy, i.e. k >=1;
U=(U1,U2,…,Un) be participant revenue function set,
Represent user n in user 1,2 ..., selection strategy s after n-1 actionnIncome;Revenue function represents participant from game
Available income level, is together decided on by the strategy of all participants, the receipts obtained by the different strategy combination of participant
It is beneficial different.
Resource regulating method based on dynamic game under described cloud environment, the scheduling model meets three it is assumed that i.e.:
Rational conditions:Assuming that each user is rational, meaningless, malice will not be sent because of maximum return to be obtained and takes money
The task requests in source;Type is assumed:Assuming that the demand of different user is different, the QoS of proposition is different;Income is assumed:Assuming that all
The target of user is all on the premise of oneself QoS is met, to ask resource to complete task, the target at scheduling of resource center is to the greatest extent
The demand that all users propose may be met, feedback resources complete its task.
Specifically, in step 3 income of user be defined as its QoS satisfaction using cloud environment scheduling of resource game theory come
Weigh, cloud environment scheduling of resource game theory represents that (N, S, U) is represented with a triple, and wherein N represents all node sets, generation
The set of all users in table scheduling of resource;S is the set of directed edge in game theory, represents the strategy of user in scheduling of resource;U
It is then the set of user's income, represents the income obtained under Different Strategies.
Specifically, QoS satisfaction is weighed with the income of each user, i.e. Ui(s1,…,si)=ε1Q1+ε2Q2+...+εnQn+ ..., wherein Q represents Service Quality Metrics, ε1,ε2,…,εnRepresent weight of the user to all kinds of indexs, ε1,ε2,…,εn∈
(0,1)。εiQiRepresent satisfaction of the user to i resource type service quality;And the income at scheduling of resource center is all users
Total revenue, i.e.,:
Specifically, step 4 is specially:Since the section that directly moves ahead of game terminal section, then inversely returned by game theory
The method received, is referred to as the reverse induction in game;Solution is since the bottom of extension game theory, it is considered to user n
Game, if user's n-1 selection strategiesThe then selection strategy for user nBetter than other any one plans of selection
Slightly, likewise, when user's n-1 selection strategiesWhen, user n optimal policy isEach user knows these information, therefore
User n-1 can selection strategyTo maximize the interests of oneself, so, sub- Equilibrium Game solution is solvedWith such
Push away, finally solve Equilibrium Game solution
Specifically, in cloud environment dynamic game resource dispatching model, strategyBe one receive it is assorted
Weighing apparatus, if for all i ∈ N,It isA peak optimization reaction, be for all si∈SiWith all i ∈ N,
Have:Wherein,Represent that user i selects optimal policy after user i-1 actionIncome,Represent that user i selects non-optimal tactful s after user i-1 actioniIncome.
Embodiment 2, with reference to Fig. 1-Fig. 5, the resource regulating method based on dynamic game under a kind of cloud environment is provided under cloud environment
Source scheduling refers to that on the basis of resource virtualizing the calculating money that each user performs required by task is dynamically distributed at scheduling of resource center
Source, provides feasibility environment to perform task, is the basis that each generic task is smoothly implemented.I.e. in cloud environment scheduling of resource process
In, physical resource is virtualized into dummy unit by cloud system, is used as the carrier of the task of execution.User performs task and often corresponds to one
Optimal dummy unit type is planted, and any sufficient physical resource (such as various kinds of sensors, computer, database) can
These corresponding dummy units are created, and then each user can be distributed to complete its task.But different establishment schemes is direct
The service quality QoS (Quality of Service) for influenceing user to obtain, that is, influence user to perform the efficiency of task, therefore
QoS is the main target of each user's contention.
Dynamic game opinion refers to that the action of participant has sequencing, and action is observed that the former choosing in the latter
Select, and make corresponding selection accordingly.And it is that one group of game sequence is described by the form of tree to extend game, Complete Information is won
Play chess and refer to during game, each game participant knows about the situation of Profit that other participants select Different Strategies,
All decision-makings occurred before knowing.And it is as follows the characteristics of multi-user services scheduling of resource:(1) user is according to respective demand
Task requests are submitted in different time node, it is believed that be that sequencing is carried out, be divided into not according to the request time of user
Same user's group;(2) although cloud computing resource pool convergence is infinitely great, resource performance has difference.If some high-performance resource
The resource performance will be substantially reduced by being used by multiple users, so the strategy of previous user can influence the plan of latter user
Slightly select, i.e., subscriber policy is mutually restricted;(3) cloud computing control centre grasp user application sequencing, request content and
Revenue function.Therefore, the process of multi-user services scheduling of resource can be regarded as extensive game with perfect information in dynamic game
Process, so the present invention is to set up the Mulitiple user resource scheduling based on extensive game with perfect information in cloud computing resources control centre
Model.
Model hypothesis, it is assumed that 1. rational conditions:Assuming that each user is rational, will not be because of maximum return to be obtained
Send the task requests that meaningless, malice takes resource.Assuming that 2. types are assumed:Assuming that the demand of different user is different, carry
The QoS gone out is different.Assuming that 3. incomes are assumed:Assuming that the target of all users is all on the premise of oneself QoS is met, request money
Source completes task.The target at scheduling of resource center is to meet the demand that all users propose as far as possible, and feedback resources complete it
Task.
Model definition
Define 1, cloud environment dynamic game resource dispatching model MCED-GRSM (Military Cloud Environment
Dynamic Game Resource Scheduling Model) it is a five-tuple MCED-GRSM=(N, T, S, I, U), its
In:
1) N=(N1,N2,…,Nn) be game participant set.Participant is to participate in the independent decision-making of game, independently hold
The individual or entity of result is carried on a shoulder pole, in different occasions, the definition of participant is different.Herein, participant is a certain
All users for sending task requests of time slice.
2) T=(T1,T2,…,Tn) be participant action order.The elder generation of resource is selected in sometime fragment user
Order, is first selected according to the high user of task attribute height progress resource selection sequence, i.e. task rank first in cloud environment afterwards
Available resources are selected to complete task;Then task rank identical user takes the principle of ordering that first to file is first selected.
3) S=(S1,S2,…,Sn) be participant policy space.SiRepresent participant NiPolicy space, it is each to participate in
Person should have more than a kind of strategy, i.e. h >=1.In cloud environment, the strategy of each user takes the optimal resources principle of selection.
4) I=(I1,I2,…,In) be participant information collection.Represent participant
NiInformation collection, each participant should have more than a kind of strategy, i.e. k >=1.Herein, before knowing during user's selection resource
The selection strategy of one user.
5) U=(U1,U2,…,Un) be participant revenue function set.
Represent user n in user 1,2 ..., selection strategy s after n-1 actionnIncome.Revenue function represents participant from game
Available income level, is together decided on by the strategy of all participants, the receipts obtained by the different strategy combination of participant
It is beneficial different.In cloud environment, participant's revenue function is user QoS satisfaction.
It is the strategy for being frequently utilized for representing to realize each income in scheduling process to define 2, cloud environment scheduling of resource game theory
The path form of expression.It has the structure typically set, and represents that (N, S, U) is represented with a triple, as shown in Figure 2.Wherein N tables
Show all node sets, represent the set of all users in scheduling of resource;S is the set of directed edge in game theory, represents resource
The strategy of user in scheduling;U is then the set of user's income, represents the income obtained under Different Strategies.
Income quantum chemical method based on dynamic game and computable general equilibrium under cloud environment, including, (1) income quantum chemical method:
The quantum chemical method of user's income is the basis of follow-up scheduling game theory analysis in cloud environment scheduling of resource, and directly affects
The result of scheduling of resource.Therefore, the tactful reasonably income that carries out to each user quantifies to be necessary.In actual money
In the scheduling process of source, scheduling of resource center resources convergence is infinitely more, and resource performance has excellent difference;Apply for the user of different task
To resource performance weighted.And scheduling of resource center is the QoS for meeting all users as far as possible, feedback resources perform it
Business, so as to obtain income.If the QoS of all users conversely can not be met as far as possible, benefit will be reduced, is lost.
It is recognized herein that the income of each user is its QoS satisfaction, i.e.,:
Ui(s1,…,si)=ε1Q1+ε2Q2+...+εnQn+…
Wherein Q represents Service Quality Metrics, such as response time, reliability, confidentiality etc..ε1,ε2,…,εnRepresent and use
Family is to the weight of all kinds of indexs, ε1,ε2,…,εn∈(0,1)。εiQiRepresent satisfaction of the user to i resource type service quality
Degree.
And the income at scheduling of resource center is the total revenue of all users, i.e.,:
(2) computable general equilibrium:Any one user application resource is intended to obtain the high-quality resource at scheduling of resource center, meets
The QoS of oneself, so as to complete task.If scheduling of resource center can not meet its QoS, user's benefit will be reduced.So face
To the different QoS of all application resource users, resource Selection Strategy how is distributed, the QoS that all users are met as far as possible is cloud
The key issue of environmental resource scheduling.
During scheduling of resource, each user sequentially selects high-quality resource, latter according to the attribute height of the task of submission
User can only select resource according to previous user behavior, to meet the QoS of oneself as far as possible, obtain maximum return U.This hair
Reverse induction (backward induction in games) in bright use game theory solves the Nash Equilibrium Solution of game.
Define 3, reverse induction:Since the section that directly moves ahead of game terminal section, then inversely concluded by game theory
Method, be referred to as the reverse induction in game.Solution is since the bottom of extension game theory.Consider that user n is won
Play chess, if user's n-1 selection strategiesThe then selection strategy for user nBetter than other are selected, any one is tactful.
Likewise, when user's n-1 selection strategiesWhen, user n optimal policy isEach user knows these information, therefore uses
Family n-1 can selection strategyTo maximize the interests of oneself, so, sub- Equilibrium Game solution is solvedBy that analogy,
Finally solve Equilibrium Game solution
Define 4, in MCED-GRSM models, strategyIt is a Nash Equilibrium, if for
All i ∈ N,It isA peak optimization reaction, be for all si∈SiWith all i ∈ N, have:
Wherein,Represent that user i selects optimal after user i-1 action
StrategyIncome,Represent that user i selects non-optimal tactful s after user i-1 actioniIncome.
Specifically, cloud environment scheduling of resource example is as shown in Figure 3.The example is described to be provided under a virtualization cloud environment
Source scheduling problem, scheduling of resource center resource pool convergence is infinitely great, and the QoS index of each user has response time, stability, secrecy
Property.Sometime fragment has four users to submit task to scheduling of resource center, and the QoS of each user is different, i.e., to difference
The weighted of resource type.Used first according to the distribution of the height of task attribute is each during scheduling of resource at scheduling of resource center
Family selects the sequencing of resource, then takes task rank identical user the first selection principle distribution of first to file successively suitable again
Sequence.And latter user can only select resource according to the behavior of previous user, to meet the QoS of oneself as far as possible.Scheduling
Model generates this betting data table 2 according to existing information.
The cloud environment resource schedule data table of table 2
Determine to calculate the income quantization that each user selects Different Strategies after this cloud environment scheduling of resource betting data,
Such as table 3,4.
Each user's income of the cloud environment scheduling of resource of table 3 (high-level task)
Each user's income of the cloud environment scheduling of resource of table 4 (low level task)
After income quantum chemical method, above-mentioned data input Gambit game softs are subjected to computable general equilibrium.It can thus be concluded that
To game theory as shown in Figure 4,5.
Experimental result represents that only one Nash Equilibrium Solution can be found from the game of high low level task respectively, constituted originally
The Nash Equilibrium Solution of secondary cloud environment scheduling of resource, as a result for:I.e. four use can select second per family
Strategy.
Above-mentioned Nash Equilibrium may be interpreted as:During cloud environment scheduling of resource, the resource of high-level task is ensured first
Distribute, then each user of same rank task takes action according to the strategy of previous user, high-quality resource is selected as far as possible to hold
Row task, makes the Income Maximum of oneself.I.e. in Nash Equilibrium, respectively task is performed with optimal resource can be selected per family.More than
As a result show, cloud environment resource dispatching model proposed by the invention and method more rationally, effectively can reflect that strategy is received
Benefit performs the influence of task to user, and can effectively carry out optimal scheduling of resource.
Claims (8)
1. the resource regulating method based on dynamic game under a kind of cloud environment, it is characterised in that comprise the following steps:
Step 1:Build the resource dispatching model based on dynamic game under cloud environment;
Step 2:Resource dispatching model based on dynamic game, distributes according to the attribute of each user task height and selects the excellent of resource
First weigh;
Step 3:First to file is taken task attribute identical user first to select the strategy of allocation priority, and by the receipts of each user
Benefit is defined as its QoS satisfaction;
Step 4:The Nash Equilibrium Solution of game is solved using reverse induction.
2. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that the step
Scheduling of resource refers on the basis of resource virtualizing under cloud environment in rapid 1, and scheduling of resource center is dynamically distributed each user and performed
The computing resource of required by task, feasibility environment is provided to perform task;Under cloud environment during scheduling of resource, physical resource quilt
Cloud system is virtualized into dummy unit, is used as the carrier of the task of execution.
3. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that:The cloud
Environment dynamic game resource dispatching model is a five-tuple, i.e. MCED-GRSM=(N, T, S, I, U), wherein:N=(N1,
N2,…,Nn) be game participant set, participant be sometime fragment it is all send task requests users;T=(T1,
T2,…,Tn) be participant action order, i.e., sometime fragment user selection resource sequencing;S=(S1,
S2,…,Sn) be participant policy space,Represent participant NiPolicy space,
Each participant should have more than a kind of strategy, i.e. h >=1;
I=(I1,I2,…,In) be participant information collection,Represent participant NiLetter
Breath collection, each participant should have more than a kind of strategy, i.e. k >=1;
U=(U1,U2,…,Un) be participant revenue function set,s2∈S2,…,sn∈Sn,Un(s1,s2,…,
sn) represent user n in user 1,2 ..., selection strategy s after n-1 actionnIncome;Revenue function represents participant from game
In available income level, together decided on by the strategy of all participants, obtained by the different strategy combination of participant
Income is different.
4. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that the tune
Spend model and meet three it is assumed that i.e.:Rational conditions:Assuming that each user is rational, will not be because of maximum return to be obtained
Send the task requests that meaningless, malice takes resource;Type is assumed:Assuming that the demand of different user is different, the QoS of proposition
It is different;Income is assumed:Assuming that the target of all users is all on the premise of oneself QoS is met, resource is asked to complete task,
The target at scheduling of resource center is to meet the demand that all users propose as far as possible, and feedback resources complete its task.
5. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that the step
The income of user is defined as its QoS satisfaction and weighed using cloud environment scheduling of resource game theory in rapid 3, and cloud environment resource is adjusted
Degree game theory represents that (N, S, U) is represented with a triple, and wherein N represents all node sets, represents in scheduling of resource and owns
The set of user;S is the set of directed edge in game theory, represents the strategy of user in scheduling of resource;U is then the collection of user's income
Close, represent the income obtained under Different Strategies.
6. the resource regulating method based on dynamic game under cloud environment according to claim 5, it is characterised in that:It is described
QoS satisfaction is weighed with the income of each user, i.e. Ui(s1,…,si)=ε1Q1+ε2Q2+...+εnQn+ ..., wherein Q is represented
Service Quality Metrics, ε1,ε2,…,εnRepresent weight of the user to all kinds of indexs, ε1,ε2,…,εn∈(0,1)。εiQiRepresent use
Satisfaction of the family to i resource type service quality;And the income at scheduling of resource center is the total revenue of all users, i.e.,:
7. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that:The step
Rapid 4 are specially:Since the section that directly moves ahead of game terminal section, the method then inversely concluded by game theory is referred to as winning
Reverse induction in playing chess;Solution is since the bottom of extension game theory, it is considered to user n subgame, if user n-1 is selected
Select strategyThe then selection strategy for user nBetter than other any one strategies are selected, likewise, as user n-1
Selection strategyWhen, user n optimal policy isEach user knows these information, therefore user n-1 can selection strategy
To maximize the interests of oneself, so, sub- Equilibrium Game solution is solvedBy that analogy, Equilibrium Game is finally solved
Solution
8. the resource regulating method based on dynamic game under cloud environment according to claim 1, it is characterised in that the cloud
In environment dynamic game resource dispatching model, strategyIt is a Nash Equilibrium, if for all i
∈ N,It isA peak optimization reaction, be for all si∈SiWith all i ∈ N, have:
Wherein,Represent that user i selects optimal policy after user i-1 actionIncome,Represent user i
Non-optimal tactful s is selected after user i-1 actioniIncome.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710335668.5A CN107168797A (en) | 2017-05-12 | 2017-05-12 | Resource regulating method based on dynamic game under cloud environment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710335668.5A CN107168797A (en) | 2017-05-12 | 2017-05-12 | Resource regulating method based on dynamic game under cloud environment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107168797A true CN107168797A (en) | 2017-09-15 |
Family
ID=59815573
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710335668.5A Pending CN107168797A (en) | 2017-05-12 | 2017-05-12 | Resource regulating method based on dynamic game under cloud environment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107168797A (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109218440A (en) * | 2018-10-12 | 2019-01-15 | 上海拟态数据技术有限公司 | A kind of mimicry web server isomery execution body dynamic dispatching method of displaying |
CN109460295A (en) * | 2018-10-19 | 2019-03-12 | 中南大学 | A kind of edge calculations performance optimization method based on multi-user's competitive behavior model |
CN109739628A (en) * | 2018-12-28 | 2019-05-10 | 北京工业大学 | A kind of cloud computing method for scheduling task based on expense |
CN109901050A (en) * | 2019-02-25 | 2019-06-18 | 哈尔滨师范大学 | A kind of three dimension system chip testing method for optimizing resources and system |
CN110266770A (en) * | 2019-05-30 | 2019-09-20 | 湖南大学 | Idle cloud resource dispatching method and device based on game theory |
CN110365515A (en) * | 2019-05-30 | 2019-10-22 | 东南大学 | Service internet multi-tenant satisfaction measure based on extensive entropy |
CN110505217A (en) * | 2019-08-05 | 2019-11-26 | 河北科技大学 | A kind of location privacy protection method merged based on game theory with block chain |
CN110599051A (en) * | 2019-09-19 | 2019-12-20 | 桂林电子科技大学 | Sub-game perfect Nash balanced fetching method of two agents |
CN110772798A (en) * | 2019-10-23 | 2020-02-11 | 桂林电子科技大学 | Method for searching Nash equilibrium sequence based on FIP structure |
CN110825517A (en) * | 2019-09-29 | 2020-02-21 | 清华大学 | Distributed resource dynamic allocation method based on evolutionary game theory |
CN110851268A (en) * | 2019-10-17 | 2020-02-28 | 中山大学 | Edge scheduling optimization method based on congestion game |
CN111597047A (en) * | 2020-05-15 | 2020-08-28 | 北京金山云网络技术有限公司 | Service deployment method, device, electronic equipment and storage medium |
CN112308279A (en) * | 2019-08-02 | 2021-02-02 | 南京理工大学 | Multi-server fund distribution method |
CN113283785A (en) * | 2021-06-09 | 2021-08-20 | 广东工业大学 | Cooperative scheduling optimization method for multi-task manufacturing resources |
CN113298392A (en) * | 2021-05-29 | 2021-08-24 | 西北工业大学 | Heterogeneous weapon target allocation method based on induction method |
CN113658422A (en) * | 2021-07-26 | 2021-11-16 | 江苏大学 | Optimal scheduling method for unmanned electric vehicle |
CN115208900A (en) * | 2022-07-15 | 2022-10-18 | 柏域信息科技(上海)有限公司 | Multi-cloud-architecture cloud service resource scheduling method based on block chain and game model |
CN115438931A (en) * | 2022-08-22 | 2022-12-06 | 成都飞机工业(集团)有限责任公司 | Production line assembly operation scheduling method, device, equipment and medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103095788A (en) * | 2011-11-02 | 2013-05-08 | 佳都新太科技股份有限公司 | Cloud resource scheduling policy based on network topology |
CN103677960A (en) * | 2013-12-19 | 2014-03-26 | 安徽师范大学 | Game resetting method for virtual machines capable of controlling energy consumption |
US9413814B2 (en) * | 2010-09-29 | 2016-08-09 | Citrix Systems, Inc. | Systems and methods for providing quality of service via a flow controlled tunnel |
CN106454849A (en) * | 2016-10-08 | 2017-02-22 | 重庆大学 | High-energy efficiency resource allocation method for cooperative cognitive radio network |
-
2017
- 2017-05-12 CN CN201710335668.5A patent/CN107168797A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9413814B2 (en) * | 2010-09-29 | 2016-08-09 | Citrix Systems, Inc. | Systems and methods for providing quality of service via a flow controlled tunnel |
CN103095788A (en) * | 2011-11-02 | 2013-05-08 | 佳都新太科技股份有限公司 | Cloud resource scheduling policy based on network topology |
CN103677960A (en) * | 2013-12-19 | 2014-03-26 | 安徽师范大学 | Game resetting method for virtual machines capable of controlling energy consumption |
CN106454849A (en) * | 2016-10-08 | 2017-02-22 | 重庆大学 | High-energy efficiency resource allocation method for cooperative cognitive radio network |
Non-Patent Citations (2)
Title |
---|
张健: "静态贝叶斯博弈在信息系统风险分析中的应用", 《计算机工程与应用》 * |
徐昕: "基于博弈论的云计算资源调度方法研究", 《中国优秀博士学位论文全文数据库 信息科技辑,第I139-1页》 * |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109218440A (en) * | 2018-10-12 | 2019-01-15 | 上海拟态数据技术有限公司 | A kind of mimicry web server isomery execution body dynamic dispatching method of displaying |
CN109460295A (en) * | 2018-10-19 | 2019-03-12 | 中南大学 | A kind of edge calculations performance optimization method based on multi-user's competitive behavior model |
CN109739628A (en) * | 2018-12-28 | 2019-05-10 | 北京工业大学 | A kind of cloud computing method for scheduling task based on expense |
CN109901050A (en) * | 2019-02-25 | 2019-06-18 | 哈尔滨师范大学 | A kind of three dimension system chip testing method for optimizing resources and system |
CN110266770A (en) * | 2019-05-30 | 2019-09-20 | 湖南大学 | Idle cloud resource dispatching method and device based on game theory |
CN110365515A (en) * | 2019-05-30 | 2019-10-22 | 东南大学 | Service internet multi-tenant satisfaction measure based on extensive entropy |
CN110266770B (en) * | 2019-05-30 | 2020-07-07 | 湖南大学 | Game theory-based idle cloud resource scheduling method and device |
CN110365515B (en) * | 2019-05-30 | 2022-04-08 | 东南大学 | Service internet multi-tenant satisfaction degree measuring method based on generalization entropy |
CN112308279A (en) * | 2019-08-02 | 2021-02-02 | 南京理工大学 | Multi-server fund distribution method |
CN110505217A (en) * | 2019-08-05 | 2019-11-26 | 河北科技大学 | A kind of location privacy protection method merged based on game theory with block chain |
CN110505217B (en) * | 2019-08-05 | 2021-11-02 | 河北科技大学 | Position privacy protection method based on game theory and block chain fusion |
CN110599051A (en) * | 2019-09-19 | 2019-12-20 | 桂林电子科技大学 | Sub-game perfect Nash balanced fetching method of two agents |
CN110825517A (en) * | 2019-09-29 | 2020-02-21 | 清华大学 | Distributed resource dynamic allocation method based on evolutionary game theory |
CN110825517B (en) * | 2019-09-29 | 2020-09-08 | 清华大学 | Distributed resource dynamic allocation method based on evolutionary game theory |
CN110851268A (en) * | 2019-10-17 | 2020-02-28 | 中山大学 | Edge scheduling optimization method based on congestion game |
CN110851268B (en) * | 2019-10-17 | 2023-06-13 | 中山大学 | Edge scheduling optimization method based on congestion game |
CN110772798A (en) * | 2019-10-23 | 2020-02-11 | 桂林电子科技大学 | Method for searching Nash equilibrium sequence based on FIP structure |
CN111597047A (en) * | 2020-05-15 | 2020-08-28 | 北京金山云网络技术有限公司 | Service deployment method, device, electronic equipment and storage medium |
CN113298392A (en) * | 2021-05-29 | 2021-08-24 | 西北工业大学 | Heterogeneous weapon target allocation method based on induction method |
CN113298392B (en) * | 2021-05-29 | 2023-06-06 | 西北工业大学 | Heterogeneous weapon target distribution method based on induction method |
CN113283785A (en) * | 2021-06-09 | 2021-08-20 | 广东工业大学 | Cooperative scheduling optimization method for multi-task manufacturing resources |
CN113658422A (en) * | 2021-07-26 | 2021-11-16 | 江苏大学 | Optimal scheduling method for unmanned electric vehicle |
CN115208900A (en) * | 2022-07-15 | 2022-10-18 | 柏域信息科技(上海)有限公司 | Multi-cloud-architecture cloud service resource scheduling method based on block chain and game model |
CN115208900B (en) * | 2022-07-15 | 2024-03-15 | 柏域信息科技(上海)有限公司 | Multi-cloud architecture cloud service resource scheduling method based on blockchain and game model |
CN115438931A (en) * | 2022-08-22 | 2022-12-06 | 成都飞机工业(集团)有限责任公司 | Production line assembly operation scheduling method, device, equipment and medium |
CN115438931B (en) * | 2022-08-22 | 2024-03-15 | 成都飞机工业(集团)有限责任公司 | Method, device, equipment and medium for scheduling assembly operation of production line |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107168797A (en) | Resource regulating method based on dynamic game under cloud environment | |
He et al. | Blockchain-based edge computing resource allocation in IoT: A deep reinforcement learning approach | |
Du et al. | Learning resource allocation and pricing for cloud profit maximization | |
Ge et al. | GA-based task scheduler for the cloud computing systems | |
CN110266745B (en) | Information flow recommendation method, device, equipment and storage medium based on deep network | |
CN108874538B (en) | Scheduling server, scheduling method and application method for scheduling quantum computer | |
CN107193658A (en) | Cloud computing resource scheduling method based on game theory | |
Xu et al. | Multi-objective data placement for workflow management in cloud infrastructure using NSGA-II | |
CN109558989A (en) | Queuing time prediction technique, device, equipment and computer readable storage medium | |
CN110009233B (en) | Game theory-based task allocation method in crowd sensing | |
Li et al. | A distributed QoS-constraint task scheduling scheme in cloud computing environment: model and algorithm | |
CN106372101B (en) | A kind of video recommendation method and device | |
CN114610474B (en) | Multi-strategy job scheduling method and system under heterogeneous supercomputing environment | |
CN102300011A (en) | Automated mechanism for populating and maintaining data structures in queueless contact center | |
CN109857535A (en) | The implementation method and device of task priority control towards Spark JDBC | |
Entezari-Maleki et al. | Modeling and evaluation of service composition in commercial multiclouds using timed colored Petri nets | |
Hassan et al. | Energy-aware resource and revenue management in federated cloud: A game-theoretic approach | |
Majumder et al. | Game theory-based allocation of critical resources during natural disasters | |
CN105005501B (en) | A kind of second order optimizing and scheduling task method towards cloud data center | |
Pal et al. | Adaptation of Johnson sequencing algorithm for job scheduling to minimise the average waiting time in cloud computing environment | |
CN110743164B (en) | Dynamic resource partitioning method for reducing response delay in cloud game | |
CN116362327A (en) | Model training method and system and electronic equipment | |
CN106603657B (en) | A kind of video conference method for optimizing resources based on IMS | |
CN109101329A (en) | The finegrained tasks distribution method and system of data are acquired by multiple mobile terminals | |
Mallikarjuna | The effective tasks management of workflows inspired by NIM-game strategy in smart grid environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170915 |