Research On Task Scheduling Model And Algorithm Of Cloud Computing Based On Game

As an important paradigm,cloud computing is a powerful computing capability that allows users to use it to access the computing resources and services they need quickly,flexibly and at low cost,and to digitize,intelligentize and network their business.However,owing to the diverse performance of virtual machine(VM)resources managed by cloud servers,unreasonable scheduling schemes will make the task cannot be completed on time as required,and will cause the waste of service providers’ resources,which makes it necessary for data centers to consider the needs of each user and the performance of servers to maximize the benefits for both parties.What’s more,different task requirements such as computation-intensive tasks and cost-sensitive tasks also affect the allocation scheme,so an excellent scheduling scheme is of great importance to the performance of cloud computing.In this paper,we propose two game scheduling models for the task scheduling problem,combining knowledge of game theory and optimization algorithms,and apply the ideas of game theory to design a new algorithm to solve the proposed models.The specific work of this paper is as follows:(1)In the scheduling process,a multi-objective task scheduling model based on a two-layer game(TLGM)is proposed in order to maintain the scheduling performance while balancing the interests between users and service providers.The two-layer game is divided according to the different competing relationships of the players.The inner layer is a game between task,where tasks improve their own utility by changing their respective strategies;the outer layer is a game between users and service providers.In addition,this paper also sets up a strategy update mechanism for players according to the characteristics of the model.The experimental part verifies the validity of the proposed model by setting different amounts of tasks,and solving this model using a common multi-objective algorithm yields a set of scheduling solutions that satisfy the interests of both users and service providers.(2)In TLGM,the payoff functions of the user and the service provider are defined in a weighted manner,which does not portray well the trade-offs between the two players in the actual game process with respect to their preferred multiple objectives.Therefore,in order to enable service providers and users to better balance their respective multiple objectives in the game process,this paper firstly formulates the task scheduling problem as a bi-objective game problem and then constructs a task scheduling model based on the bi-objective game(TSBOG).In this model,the energy consumption and resource utilization of the service provider and the cost and completion rate of the task of the user are calculated simultaneously.In addition,a crossover and variation operator is proposed in order to balance the multiple objectives of each player.The effectiveness of the model is demonstrated by solving it using several common many-objective optimization algorithms.(3)In order to solve TSBOG efficiently and to address the conflict between convergence and diversity and the lack of Pareto selection pressure due to the increase in the number of objectives in the many-objective algorithm,this paper proposes a many-objective optimization algorithm base on game(Ma OEA-Game)by mapping the elements of multi-objective optimization to game theory.At the same time,the paper designs a partition selection strategy and a partition deletion strategy as the player’s strategy set.The experimental part was conducted with four advanced high-dimensional multi-objective evolutionary algorithms on seven DTLZ test problems and this algorithm was solved for the TSBOG model.The numerical experimental results establish that Ma OEA-Game is more suitable for solving the scheduling model than other optimization algorithms and can effectively improve the benefits of each player and obtain an efficient cloud task allocation scheme.