基于CT-RAG和学习量子粒子群的云计算任务-资源分配算法  

Task-Resource Allocation Algorithm Based on CT-RAG and Study Quantum-behaved Particle Swarm Algorithm

作　　者：宗苏[1] 

机构地区：[1]赣南师范学院科技学院,江西赣州341000

出　　处：《计算机测量与控制》2014年第5期1537-1539,1567,共4页Computer Measurement &Control

摘　　要：目前已有的云计算任务-资源分配算法仅针对独立任务进行同构资源分配,同时在分配时未考虑任务优先级;为了克服其缺点,提出了一种基于虚拟CT-RAG(Task-Resource Assignment Graph in Cloud Environment,CT-RAG)和学习量子粒子群的任务-资源分配模型;首先,定义了虚拟CT-RAG图和任务优先级,并描述了采用其获取任务-资源分配方案初始解的方法;然后采用具有学习能力的量子粒子群在可行解空间中寻优,通过为粒子安装学习机,粒子在每轮迭代的过程中根据适应度的变化情况自适应地调整动作选择概率,从而加快获取全局最优解和加快收敛速度;仿真实验表明:文中方法能有效地解决云计算环境下依赖型任务的异构资源调度,获取了全局最优解356.67,较其它方法具有较大的优越性。The given task-- resource allocation algorithm only considers the independent task and homogeneous resource, and also do not mention task priority. In order to conquer their defects, a task--resource allocation model based on CT--RAG （Task--Resource Assign- ment Graph in Cloud Environment） and studying Quantum--behaved Particle Swarm was proposed. Firstly, the virtual CT--RAG and task priority was defined, and the initial solution was obtained by using CT--RAG. Then using the studying Quantum--behaved Particle Swarm to search the global optimum solution, every particle was installed studying machine, so it can change the action selection probability accord- ing to the fitness and finally converge to the global optimum solution. The simulation experiment shows the result in the solution in this paper can realize task--resource allocation in cloud environment, the optimal solution is 356.67, and compared with other methods, it has larger priority.

关 键 词：任务-资源分配 云计算 量子粒子群 学习 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]