CN107168797A - Resource regulating method based on dynamic game under cloud environment - Google Patents

Abstract

The invention belongs to the technical field of resource scheduling based on cloud computing, and specifically relates to a resource scheduling method based on a dynamic game in a cloud environment, comprising the following steps: constructing a resource scheduling model based on a dynamic game in a cloud environment; a resource scheduling model based on a dynamic game , allocate the priority of resource selection according to the attributes of each user task; for users with the same task attributes, adopt the strategy of applying first and choosing first to allocate priority, and define the income of each user as its QoS satisfaction; use the reverse induction method Find the Nash equilibrium solution of the game. Aiming at the problem of resource competition caused by multiple users submitting tasks at the same time in the cloud environment, the present invention establishes a dynamic game resource scheduling model in the cloud environment, uses dynamic game theory to model and analyze resource competition among users, and maximizes Meet the QoS requirements of all users; thus satisfy the needs of each user to the greatest extent; use the reverse induction method to solve the Nash equilibrium solution of the dynamic game model, thereby realizing the most reasonable and effective allocation of resources.

Description

Resource scheduling method based on dynamic game in cloud environment

technical field

The invention belongs to the technical field of resource scheduling based on cloud computing, and in particular relates to a dynamic game-based resource scheduling method in a cloud environment.

Background technique

Cloud computing is a new computing model developed on the basis of grid computing, parallel computing and P2P technologies. It links together a large number of computing resources, storage resources and software resources stored on the data center cluster (including hardware resources such as CPU, memory, hard disk, and software resources such as development environment and application programs), and performs centralized and unified management to form a large Large-scale shared virtual resource pools are provided to users on demand, providing them with cloud services that are "on call, go away" and seem to have "unlimited capabilities". With the rapid development of distributed technology, cloud computing has become a key research field. Cloud computing is based on Service Oriented Architecture (SOA), which provides computing resources in the form of services, and users can obtain cloud services on demand.

With the rapid development of cloud computing, massive resources are integrated into the cloud, user demands continue to increase, and the services provided by cloud computing centers continue to increase, facing huge pressure on resource scheduling. Because the cloud computing system is in a dynamic environment, and the user group is very large, each application program has continuously changing resource requirements, resulting in its performance needs to be dynamically satisfied ^[8] . On the other hand, cloud computing uses virtualization technology to shield the complexity of the underlying hardware and improve flexibility. In a multi-user environment, a large number of user applications run on one physical device, sharing hardware and storage devices. The on-demand resource supply and multi-user resource sharing of cloud computing make the resource scheduling problem of cloud computing become the second biggest problem after the security problem in cloud computing.

In the process of cloud computing service resource scheduling, multiple users submit task requests successively, and can be divided into multiple user groups according to the time node of the request. They require the same type of resources, but during the scheduling process, it must be ensured that the user group submitted first cannot monopolize this type of resource, which forms resource competition among different user groups. In order to meet the needs of each user group, meet the constraints proposed by the user group, such as task completion time, the cost of completing the task, etc., and at the same time achieve the most reasonable and effective use and allocation of resources, it is necessary to establish a competition model based on dynamic games , to design a fair and effective resource scheduling method, which is used to solve the competition problem of multi-user non-simultaneous application resources in cloud service resource scheduling, which not only improves the efficiency of resource scheduling but also ensures the fairness of resource allocation.

The existing technology generally adopts resource scheduling based on intelligent optimization algorithms. Starting from the overall interests of the scheduling center, intelligent algorithms are applied to cloud computing resource scheduling, which improves resource scheduling efficiency and ensures system performance. However, the shortcoming is that the overall interests of the dispatching center are the main priority, and it cannot meet the resource needs of all users, and cannot guarantee the fairness of resource allocation.

The other is resource scheduling based on game theory, which satisfies the characteristics of individual rationality and incentive compatibility. The shortcoming is that the price of cloud computing resources is excessively pursued, and other needs of users are not guaranteed, that is, the quality of service QoS (Quality of Service) of users is not guaranteed. ).

Game theory is one of the most basic theories in economics. It is used to solve the competition and mutual checks and balances between different individuals. It is considered in the western social sciences to be the greatest achievement in the field of social sciences in the 20th century. Game theory is used to solve the competition and mutual checks and balances between different individuals, and is considered to be the greatest achievement in the field of social science in the 20th century in Western social science. Game theory, as a theory of competition among participants, has very high value in modeling and analyzing the competition among different individuals. Aiming at the competition problem of multi-user non-simultaneous application for resources in cloud service resource scheduling, the present invention proposes a reasonable and effective resource scheduling model and method based on dynamic game theory, which is used to solve the problem of multi-user resource competition in cloud environment and improve resource scheduling. efficiency.

Contents of the invention

The present invention aims at resource scheduling based on intelligent optimization algorithm in the prior art, which mainly focuses on the overall interests of the scheduling center, cannot meet the resource requirements of all users, cannot guarantee the fairness of resource allocation, and resource scheduling based on game theory, there is an excessive pursuit of cloud computing The price of resources does not guarantee the other needs of users, that is, the quality of service of users is not guaranteed. A resource scheduling method based on dynamic game in cloud environment is proposed.

The technical solution of the present invention is: a resource scheduling method based on dynamic game in a cloud environment, comprising the following steps:

Step 1: Build a resource scheduling model based on dynamic games in the cloud environment;

Step 2: Based on the resource scheduling model of dynamic game, assign the priority of selecting resources according to the attributes of each user task;

Step 3: For users with the same task attributes, adopt the strategy of applying first and choosing first to allocate priority, and define the income of each user as its QoS satisfaction;

Step 4: Solve the Nash equilibrium solution of the game by using the reverse induction method.

In the resource scheduling method based on dynamic game in the cloud environment, the resource scheduling in the cloud environment in the step 1 means that on the basis of resource virtualization, the resource scheduling center dynamically allocates the computing resources required by each user to perform tasks, for The execution task provides a feasible environment; in the process of resource scheduling in the cloud environment, the physical resources are virtualized into virtual units by the cloud system as the carrier of the execution task.

In the resource scheduling method based on dynamic game in the cloud environment, the dynamic game resource scheduling model in the cloud environment is a quintuple, that is, MCED-GRSM=(N, T, S, I, U), wherein: N= (N ₁ ,N ₂ ,…,N _n ) is the set of participants in the game, and the participants are all users who send task requests in a certain time segment; T=(T ₁ ,T ₂ ,…,T _n ) is the participant The sequence of actions of the user, that is, the order in which the user selects resources in a certain time segment; S = ₍ S ₁ , S ₂ ,…,S _n ) is the strategy space of the participants, Represents the strategy space of the participant N _i , each participant should have more than one strategy, that is h≥1;

I=(I ₁ ,I ₂ ,…, _{In ) is} the participant’s information set, Represents the information set of participant N _i , each participant should have more than one strategy, that is, k≥1;

U＝(U ₁ ,U ₂ ,…,U _n ) is the set of profit functions of the participants, Indicates the income of user n choosing strategy s _n after user 1, 2, ..., n-1 actions; the income function indicates the level of income that participants can get from the game, which is jointly determined by the strategies of all participants, and different participants The benefits obtained by the strategy combination are different.

In the resource scheduling method based on dynamic game in the cloud environment, the scheduling model satisfies three assumptions, namely: rational assumption: it is assumed that each user is completely rational and will not send meaningless, malicious Resource-occupied task requests; type assumption: assuming that different users have different requirements, and propose different QoS; revenue assumption: assuming that all users’ goals are to request resources to complete tasks under the premise of satisfying their own QoS, the goal of the resource scheduling center It is to meet the needs of all users as much as possible, and feedback resources to complete their tasks.

In the resource scheduling method based on dynamic game in the cloud environment, the user's income in step 3 is defined as its QoS satisfaction using the cloud environment resource scheduling game tree to measure, and the cloud environment resource scheduling game tree uses a triplet Represents (N, S, U) representation, where N represents the set of all nodes, representing the set of all users in resource scheduling; S is the set of directed edges in the game tree, representing the user's strategy in resource scheduling; U is the user's income A collection of , representing the gains obtained under different strategies.

In the resource scheduling method based on dynamic game in the cloud environment, the QoS satisfaction is measured by the income of each user, that is, U _i (s ₁ ,...,s _i )=ε ₁ Q ₁ +ε ₂ Q ₂ +...+ε _n Q _n +…, where Q represents the service quality index, ε ₁ , ε ₂ ,…,ε _n represent the user’s weight on various indexes, ε ₁ ,ε ₂ ,…,ε _n ∈ (0,1). ε _i Q _i represents the user's satisfaction with the service quality of i-type resources; and the revenue of the resource scheduling center is the total revenue of all users, namely:

In the resource scheduling method based on a dynamic game in the cloud environment, the step 4 is specifically: start from the node directly preceding the end node of the game, and then use the game tree reverse induction method, which is called the reverse induction method in the game ; The solution is to start from the bottom of the extended game tree, considering the subgame of user n, if user n-1 chooses a strategy Then for user n the choice strategy Better than choosing any other strategy. Similarly, when user n-1 chooses strategy When , the optimal strategy for user n is Each user knows this information, so user n-1 will choose a strategy To maximize their own interests, in this way, solve the sub-equilibrium game solution By analogy, the equilibrium game solution is finally solved

In the resource scheduling method based on dynamic game in the cloud environment, in the dynamic game resource scheduling model in the cloud environment, the strategy is a Nash equilibrium if for all i∈N, yes An optimal response for , that is, for all s _i ∈ S _i and all i ∈ N, is: in, Indicates that user i chooses the optimal strategy after the action of user i-1 income, Denotes the benefit of user _i choosing non-optimal strategy si after user i-1 takes action.

The beneficial effects of the present invention are: the present invention aims at the problem of resource competition caused by multiple users submitting tasks at the same time in the cloud environment, establishes a dynamic game resource scheduling model in the cloud environment, and utilizes dynamic game theory to construct resource competition among users Model and analysis, to meet the QoS requirements of all users as much as possible; establish a revenue function based on the weight of users on various indicators, comprehensively consider the differences in resource types and demand levels of each user; design the optimal resource selection strategy, Prioritize the selection of task resources by users with high task levels, and adopt the principle of first application and first selection for users with the same task level, so as to meet the needs of each user to the greatest extent; use the reverse induction method to solve the Nash equilibrium solution of the dynamic game model, So as to achieve the most reasonable and effective allocation of resources.

Description of drawings

Fig. 1 is a block diagram of the scheduling method of the present invention;

Fig. 2 is a schematic diagram of the cloud environment resource scheduling game tree structure;

FIG. 3 is a schematic diagram of a cloud environment resource scheduling structure;

Fig. 4 is a schematic diagram of the game tree structure of cloud environment resource scheduling for high-level tasks;

Figure 5 is a schematic diagram of the low-level task cloud environment resource scheduling game tree structure;

detailed description

Embodiment 1: In combination with Fig. 1-Fig. 5, a resource scheduling method based on dynamic game in a cloud environment includes the following steps:

Step 1: Construct a resource scheduling model based on dynamic game in the cloud environment; Step 2: Based on the resource scheduling model of dynamic game, assign the priority of selecting resources according to the attribute level of each user task; Step 3: Take the same task attribute for users Apply first and choose the strategy of assigning priority first, and define the income of each user as its QoS satisfaction; Step 4: Use the reverse induction method to find the Nash equilibrium solution of the game.

Specifically, resource scheduling in the cloud environment in step 1 refers to the resource scheduling center dynamically allocating the computing resources required by each user to perform tasks on the basis of resource virtualization, and providing a feasible environment for task execution; the resource scheduling process in the cloud environment In the cloud system, physical resources are virtualized into virtual units, which serve as carriers for executing tasks.

Specifically, the cloud environment dynamic game resource scheduling model is a quintuple, that is, MCED-GRSM=(N,T,S,I,U), where: N=(N ₁ ,N ₂ ,…,N _n ) is The set of participants in the game, the participants are all users who send task requests in a certain time period; T=(T ₁ ,T ₂ ,…,T _n ) is the action sequence of the participants, that is, the user chooses resources in a certain time period sequence; S= ₍ S ₁ ,S ₂ ,…,S _n ) is the strategy space of the participants, Represents the strategy space of the participant N _i , each participant should have more than one strategy, that is h≥1;

I=(I ₁ ,I ₂ ,…, _{In ) is} the participant’s information set, Represents the information set of participant N _i , each participant should have more than one strategy, that is, k≥1;

U＝(U ₁ ,U ₂ ,…,U _n ) is the set of profit functions of the participants, Indicates the income of user n choosing strategy s _n after user 1, 2, ..., n-1 actions; the income function indicates the level of income that participants can get from the game, which is jointly determined by the strategies of all participants, and different participants The benefits obtained by the strategy combination are different.

In the resource scheduling method based on dynamic game in the cloud environment, the scheduling model satisfies three assumptions, namely: rational assumption: it is assumed that each user is completely rational and will not send meaningless, malicious Resource-occupied task requests; type assumption: assuming that different users have different requirements, and propose different QoS; revenue assumption: assuming that all users’ goals are to request resources to complete tasks under the premise of satisfying their own QoS, the goal of the resource scheduling center It is to meet the needs of all users as much as possible, and feedback resources to complete their tasks.

Specifically, in step 3, the user's benefit is defined as its QoS satisfaction, which is measured by the cloud environment resource scheduling game tree, which is represented by a triplet representation (N, S, U), where N represents The set of all nodes represents the set of all users in resource scheduling; S is the set of directed edges in the game tree, representing the user's strategy in resource scheduling; U is the set of user benefits, representing the benefits obtained under different strategies.

Specifically, the QoS satisfaction is measured by the income of each user, that is, U _i (s ₁ ,...,s _i )=ε ₁ Q ₁ +ε ₂ Q ₂ +...+ε _n Q _n +..., Among them, Q represents the service quality index, ε ₁ , ε ₂ ,…,ε _n represent the user’s weight on various indexes, ε ₁ , ε ₂ ,…,ε _n ∈(0,1). ε _i Q _i represents the user's satisfaction with the service quality of i-type resources; and the revenue of the resource scheduling center is the total revenue of all users, namely:

Specifically, step 4 is specifically: start from the node directly preceding the end node of the game, and then use the game tree reverse induction method, which is called the reverse induction method in the game; the solution starts from the bottom of the extended game tree, considering A subgame for user n, if user n-1 chooses a strategy Then for user n the choice strategy Better than choosing any other strategy. Similarly, when user n-1 chooses strategy When , the optimal strategy for user n is Each user knows this information, so user n-1 will choose a strategy To maximize their own interests, in this way, solve the sub-equilibrium game solution By analogy, the equilibrium game solution is finally solved

Specifically, in the cloud environment dynamic game resource scheduling model, the strategy is a Nash equilibrium if for all i∈N, yes An optimal response for , that is, for all s _i ∈ S _i and all i ∈ N, is: in, Indicates that user i chooses the optimal strategy after the action of user i-1 income, Denotes the benefit of user _i choosing non-optimal strategy si after user i-1 takes action.

Embodiment 2, in combination with Fig. 1-Fig. 5, a resource scheduling method based on dynamic game in cloud environment. Resource scheduling in cloud environment means that resource scheduling center dynamically allocates resources required by each user to perform tasks on the basis of resource virtualization. Computing resources provide a feasible environment for the execution of tasks, which is the basis for the smooth implementation of various tasks. That is, in the resource scheduling process of the cloud environment, physical resources are virtualized into virtual units by the cloud system, which are used as carriers for executing tasks. The user's execution of tasks often corresponds to an optimal type of virtual unit, and any sufficient physical resources (such as various sensors, computers, databases, etc.) can create these corresponding virtual units, and then can be assigned to each user to complete their tasks . However, different creation schemes directly affect the QoS (Quality of Service) obtained by users, that is, affect the efficiency of users in performing tasks, so QoS is the main goal for users to compete for.

Dynamic game theory means that the actions of the participants have a sequence, and the actions of the latter can observe the choices of the former, and make corresponding choices accordingly. The extended game is to describe a set of game sequences in the form of a tree. The complete information game means that in the process of the game, each game participant knows the benefits of other participants choosing different strategies, and also knows all the previous events. decision making. The characteristics of multi-user service resource scheduling are as follows: (1) Users submit task requests at different time nodes according to their own needs, which can be regarded as sequential, and are divided into different user groups according to the user's request time; (2) Although the cloud The computing resource pool approaches infinity, but resource performance varies. If a high-performance resource is used by multiple users, the performance of the resource will be greatly reduced, so the policy of the former user will affect the policy selection of the latter user, that is, the user policies restrict each other; (3) the cloud computing scheduling center grasps the user application sequence, request content and revenue function. Therefore, the process of multi-user service resource scheduling can be regarded as the process of complete information extended game in the dynamic game, so the present invention establishes a multi-user resource scheduling model based on complete information extended game in the cloud computing resource scheduling center.

Model assumptions, assumption 1. Rational assumption: Assume that each user is completely rational and will not send meaningless task requests that occupy resources maliciously in order to obtain the maximum benefit. Assumption 2. Type assumption: It is assumed that different users have different requirements and proposed different QoS. Assumption 3. Profit assumption: Assume that the goal of all users is to request resources to complete tasks under the premise of satisfying their own QoS. The goal of the resource scheduling center is to meet the needs of all users as much as possible, and feedback resources to complete their tasks.

model definition

Definition 1. MCED-GRSM (Military Cloud EnvironmentDynamic Game Resource Scheduling Model) is a five-tuple MCED-GRSM=(N,T,S,I,U), where:

1) N=(N ₁ ,N ₂ ,...,N _n ) is the set of game participants. A participant is an individual or organization that participates in the independent decision-making of the game and bears the results independently. In different occasions, the definition of a participant is different. In this paper, participants are all users who send task requests in a certain period of time.

2) T=(T ₁ ,T ₂ ,...,T _n ) is the action sequence of the participants. That is, the order in which users choose resources in a certain time segment, in the cloud environment, resource selection is first sorted according to the task attributes, that is, users with higher task levels first select available resources to complete the task; then users with the same task level apply first The ordering principle of first choice.

3) S=(S ₁ , S ₂ ,...,S _n ) is the strategy space of the participants. S _i represents the strategy space of participant N _i , and each participant should have more than one strategy, that is, h≥1. In the cloud environment, each user's strategy adopts the principle of selecting the optimal resource.

4) I=(I ₁ , I ₂ ,..., _{In ) is} the information set of the participants. Represents the information set of participants N _i , each participant should have more than one strategy, that is, k≥1. In this paper, when a user selects a resource, he knows the previous user's selection strategy.

5) U=(U ₁ , U ₂ ,...,U _n ) is the set of profit functions of the participants. Indicates the benefit of user n choosing strategy s _n after user 1, 2, ..., n-1 actions. The profit function represents the profit level that the participants can obtain from the game, which is determined by the strategies of all the participants, and the profit obtained by different strategy combinations of the participants is different. In the cloud environment, the participant benefit function is the user's QoS satisfaction.

Definition 2. Cloud environment resource scheduling Game tree is often used to express the expression form of the strategy path to realize each benefit in the scheduling process. It has a general tree structure, represented by a triple (N, S, U), as shown in Figure 2. Among them, N represents the set of all nodes, representing the set of all users in resource scheduling; S is the set of directed edges in the game tree, representing the user's strategy in resource scheduling; U is the set of user benefits, representing the benefits obtained under different strategies income.

Income quantitative calculation and equilibrium analysis based on dynamic game in cloud environment, including (1) income quantitative calculation:

The quantitative calculation of user benefits in cloud environment resource scheduling is the basis of subsequent scheduling game analysis and directly affects the results of resource scheduling. Therefore, it is very necessary to reasonably quantify the benefits of each user's strategy. In the actual resource scheduling process, the resources of the resource scheduling center approach to infinite, and the resource performance is different; users who apply for different tasks have different weights on resource performance. The resource scheduling center is to satisfy the QoS of all users as far as possible, and feedback resources to perform their tasks, so as to obtain benefits. On the contrary, if the QoS of all users cannot be satisfied as much as possible, the benefits will be reduced and losses will be incurred.

This paper considers that the benefit of each user is its QoS satisfaction, namely:

U _i (s ₁ ,…,s _i )＝ε ₁ Q ₁ +ε ₂ Q ₂ +...+ε _n Q _n +…

Among them, Q represents the service quality index, such as response time, reliability, confidentiality and so on. ε ₁ , ε ₂ ,…,ε _n represent the user’s weight on various indicators, ε ₁ ,ε ₂ ,…,ε _n ∈(0,1). ε _i Q _i represents the user's satisfaction with the service quality of i type resources.

The revenue of the resource scheduling center is the total revenue of all users, namely:

(2) Equilibrium analysis: Any user who applies for resources hopes to obtain high-quality resources from the resource scheduling center to satisfy his own QoS, so as to complete the task. If the resource scheduling center cannot satisfy its QoS, user benefits will be reduced. Therefore, in the face of the different QoS of all users who apply for resources, how to allocate resource selection strategies to satisfy the QoS of all users as much as possible is the key issue of resource scheduling in the cloud environment.

In the process of resource scheduling, each user selects high-quality resources in order according to the attributes of the submitted tasks, and the latter user can only select resources according to the behavior of the previous user, so as to satisfy its own QoS as much as possible and obtain the maximum benefit U. The present invention adopts the backward induction method in game theory (backward induction in games) to solve the Nash equilibrium solution of the game.

Definition 3. Reverse induction method: The method of starting from the directly preceding node of the end node of the game, and then reverse induction through the game tree is called the reverse induction method in the game. The solution is to start from the bottom of the extended game tree. Consider a subgame for user n, if user n-1 chooses a strategy Then for user n the choice strategy Better than choosing any other strategy. Similarly, when user n-1 chooses strategy When , the optimal strategy for user n is Each user knows this information, so user n-1 will choose a strategy To maximize their own interests, in this way, solve the sub-equilibrium game solution By analogy, the equilibrium game solution is finally solved

Definition 4. In the MCED-GRSM model, the strategy is a Nash equilibrium if for all i∈N, yes An optimal response for , that is, for all s _i ∈ S _i and all i ∈ N, is:

in, Indicates that user i chooses the optimal strategy after the action of user i-1 income, Denotes the benefit of user _i choosing non-optimal strategy si after user i-1 takes action.

Specifically, an example of resource scheduling in a cloud environment is shown in Figure 3. This example describes a resource scheduling problem in a virtualized cloud environment. The resource pool of the resource scheduling center approaches infinity, and the QoS indicators of each user include response time, stability, and confidentiality. In a certain period of time, four users submit tasks to the resource scheduling center, and the QoS of each user is different, that is, the weights for different types of resources are different. In the process of resource scheduling, the resource scheduling center first assigns the sequence of resource selection for each user according to the level of the task attribute, and then adopts the principle of first application and first selection for users with the same task level to allocate the sequence of priority. The latter user can only select resources according to the behavior of the former user, so as to satisfy his own QoS as much as possible. The scheduling model generates the game data table 2 based on the existing information.

Table 2 Cloud environment resource scheduling data table

After determining the data of this cloud environment resource scheduling game, calculate the quantification of the benefits of each user choosing different strategies, as shown in Tables 3 and 4.

Table 3 Benefits of each user in cloud environment resource scheduling (high-level tasks)

Table 4 Benefits of each user in cloud environment resource scheduling (low-level tasks)

After the quantitative calculation of the income, the above data is input into the Gambit game software for equilibrium analysis. From this, the game tree can be obtained as shown in Figure 4 and Figure 5.

The experimental results show that the only Nash equilibrium solution can be found from the game of high-level and low-level tasks respectively, which constitutes the Nash equilibrium solution of this cloud environment resource scheduling. The result is: That is, all four users will choose the second strategy.

The above Nash equilibrium can be interpreted as: in the process of resource scheduling in the cloud environment, the resource allocation of high-level tasks is guaranteed first, and then each user of the same level of tasks acts according to the strategy of the previous user, and chooses high-quality resources to perform tasks as much as possible, so that maximum benefit. That is, in Nash equilibrium, each user will choose the optimal resource to perform the task. The above results show that the cloud environment resource scheduling model and method proposed by the present invention can more reasonably and effectively reflect the impact of policy benefits on user execution tasks, and can effectively perform optimal resource scheduling.

Claims (8)

1. A resource scheduling method based on dynamic game under a cloud environment, characterized in that, comprising the following steps: Step 1: Build a resource scheduling model based on dynamic games in the cloud environment; Step 2: Based on the resource scheduling model of dynamic game, assign the priority of selecting resources according to the attributes of each user task; Step 3: For users with the same task attributes, adopt the strategy of applying first and choosing first to allocate priority, and define the income of each user as its QoS satisfaction; Step 4: Solve the Nash equilibrium solution of the game by using the reverse induction method. 2. The resource scheduling method based on dynamic game under the cloud environment according to claim 1, characterized in that, the resource scheduling under the cloud environment in the step 1 refers to that on the basis of resource virtualization, the resource scheduling center dynamically allocates each The computing resources required by users to perform tasks provide a feasible environment for task execution; during the resource scheduling process in the cloud environment, physical resources are virtualized into virtual units by the cloud system as the carrier for task execution. 3. the resource scheduling method based on dynamic game under the cloud environment according to claim 1, is characterized in that: described cloud environment dynamic game resource scheduling model is a quintuple, i.e. MCED-GRSM=(N, T, S ,I,U), where: N=(N ₁ ,N ₂ ,…,N _n ) is the set of participants in the game, and the participants are all users who send task requests in a certain time segment; T=(T ₁ ,T ₂ ,…,T _n ) is the action sequence of the participants, that is, the order in which users choose resources in a certain time segment; S= ₍ S ₁ ,S ₂ ,…,S _n ) is the strategy space of the participants, Indicates the strategy space of the participant N _i , each participant should have more than one strategy, that is h≥1; I=(I ₁ ,I ₂ ,…, _{In ) is} the participant’s information set, Represents the information set of participant N _i , each participant should have more than one strategy, that is, k≥1; U＝(U ₁ ,U ₂ ,…,U _n ) is the set of profit functions of the participants, s ₂ ∈S ₂ ,…,s _n ∈S _n ,U _n (s ₁ ,s ₂ ,…,s _n ) represents the benefit of user n choosing strategy s _n after user 1, 2,…, n-1 actions ; The profit function represents the profit level that the participants can obtain from the game, which is jointly determined by the strategies of all the participants, and the profit obtained by different strategy combinations of the participants is different. 4. The resource scheduling method based on dynamic game in the cloud environment according to claim 1, wherein the scheduling model satisfies three assumptions, namely: rational assumption: assuming that each user is completely rational and will not Send meaningless and malicious resource-occupied task requests to obtain the maximum benefit; Type assumption: Assuming that different users have different needs, the proposed QoS is different; Revenue assumption: Assuming that all users’ goals are to meet their own QoS, request Resources are used to complete tasks, and the goal of the resource scheduling center is to meet the needs of all users as much as possible, and feedback resources to complete their tasks. 5. The resource scheduling method based on dynamic game in the cloud environment according to claim 1, characterized in that, in the step 3, the user's revenue is defined as its QoS satisfaction using the cloud environment resource scheduling game tree to measure, the cloud The environmental resource scheduling game tree is represented by a triple (N, S, U), where N represents the set of all nodes, representing the set of all users in resource scheduling; S is the set of directed edges in the game tree, representing resource scheduling U is the user's strategy; U is the collection of user income, representing the income obtained under different strategies. 6. The resource scheduling method based on dynamic game in cloud environment according to claim 5, characterized in that: said QoS satisfaction is measured by the income of each user, that is, U _i (s ₁ ,..., s _i )=ε ₁ Q ₁ +ε ₂ Q ₂ +...+ε _n Q _n +..., where Q represents the service quality index, ε ₁ ,ε ₂ ,...,ε _n represent the user's weight on various indexes, ε ₁ ,ε ₂ ,…,ε _n ∈(0,1). ε _i Q _i represents the user's satisfaction with the service quality of i-type resources; and the revenue of the resource scheduling center is the total revenue of all users, namely: 7. The resource scheduling method based on dynamic game in the cloud environment according to claim 1, characterized in that: said step 4 is specifically: starting from the node directly preceding the end node of the game, and then reverse induction through the game tree , known as the reverse induction method in the game; the solution is to start from the bottom of the extended game tree, considering the subgame of user n, if user n-1 chooses the strategy Then for user n the choice strategy Better than choosing any other strategy. Similarly, when user n-1 chooses strategy When , the optimal strategy for user n is Each user knows this information, so user n-1 will choose a strategy To maximize their own interests, in this way, solve the sub-equilibrium game solution By analogy, the equilibrium game solution is finally solved 8. The resource scheduling method based on dynamic game in cloud environment according to claim 1, characterized in that, in the cloud environment dynamic game resource scheduling model, strategy is a Nash equilibrium if for all i∈N, yes An optimal response for , that is, for all s _i ∈ S _i and all i ∈ N, is: in, Indicates that user i chooses the optimal strategy after the action of user i-1 income, Denotes the benefit of user _i choosing non-optimal strategy si after user i-1 takes action.