联盟模式下云资源提供最优化算法

为了提高资源利用率和云资源提供者的收益,提出一种云联盟模式下基于合作博弈的资源提供算法.算法将资源联盟收益最优化形式化为线性规划问题,通过合作博弈核求解了该收益最优化问题,并证明了处于合作核中的资源提供策略同样是线性规划最优解.进一步,为了获得稳定的资源联盟结构,设计了基于离散Markov链的动态联盟构造模型,通过该模型可以得到收益最大化的最优联盟结构.仿真实验结果表明,对于具有博弈理性的资源提供者而言,资源联盟可以有效提高个体资源提供者的收益,同时,由于结盟代价的存在,稳定的联盟结构更加倾向于形成规模更小的联盟.

Cloud Resource Provision Optimization Algorithm in Federation Mode

In order to improve the resource utilization and the resource providers' revenue,a resource provision algorithm based on cooperative game in cloud coalition mode is proposed. The revenue optimization for the resource coalition is formalized as a linear programming problem in our algorithm and the optimization problem is solved by the core in cooperative game. Then,we prove that the resource provision strategy in the core of cooperative game is also the optimal solution of linear programming. Meanwhile,for obtaining the stable resource coalition structure,a dynamic coalition formation model based on discrete Markov chain is designed,in which the optimal coalition structure maximizing resource provider's profit can be obtained. Simulation experimental results show that,for the resource providers with game rationality,the forming resource coalition can improve the resource providers' profit. And,due to the coalitional formation cost,the stable coalition structure prefers to selecting a smaller size of coalition.