移动云计算复杂网络环境下任务粒度的应用划分和调度方法

项目名称： 移动云计算复杂网络环境下任务粒度的应用划分和调度方法

项目编号： No.71501096

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 管理科学

项目作者： 张毅

作者单位： 南京理工大学

项目金额： 17.4万元

中文摘要： 应用划分问题是移动云计算领域的一类重要研究问题，基于社交关系组建网络使得移动云计算具有复杂网络的环境。本项目拟研究移动云计算复杂网络环境下任务粒度的应用划分问题，以最小化网络中所有设备能耗为优化目标，建立数学模型。设计高效元启发式算法用于全局搜索，基于任务转移和交换邻域构造复合邻域自适应任务调度算法用于局部搜索，设计平衡全局和局部搜索的动态控制策略并提出混合应用划分算法。构造基于路径的可行任务调度生成算法，设计可跨网移动任务组合的挖掘算法，基于任务组合的转移和交换邻域构造基于网间和网内的双层任务调度算法，设计随机扰动策略并提出高效应用划分算法。本项目的研究成果可应用于实际移动云计算平台并可推广至其他具有复杂网络环境的移动计算系统（如物联网等平台），具有广泛的应用前景、重要的理论价值和实际意义。

中文关键词： 调度优化；优化算法；应用划分；移动云计算

英文摘要： The application partition problem (APP) in the mobile cloud computing (MCC) is an important research problem. Building the network based on social relationships enables the MCC to have complex network. We study task-grained APPs in the MCC with complex network, and construct their math models to minimize the power consumption of all the devices. A novel effective meta-heuristic is designed for the global search; a self-adaptive task scheduling algorithm is constructed for the local search, based on a composite neighborhood which integrates both move and swap neighborhoods; a dynamical control policy is introduced to balance the global and the local search and a hybrid application partition algorithm is proposed. A path-based algorithm is constructed to generate feasible task schedules; an algorithm is proposed to mine those task groups which are mobilizable among sub-networks, and a two-level task scheduling algorithm is designed for the optimization among and inside sub-networks; a random perturbation policy is investigated and an effective and efficient application partition algorithm is proposed. The study can be used in the real MCC and have the widespread use to other mobile computing systems with similar complex network, such as the Internet of Things.

英文关键词： scheduling optimization;optimization algorithm;application partition;mobile cloud computing