移动云计算基于随机数据模型的最优控制策略  

Optimal control stragey based on stochastic data model for mobile cloud computing

作　　者：陈超[1] 王飞 盛玉萍 康葵 何海涛[1] 华才健[1] CHEN Chao;WANG Fei;SHENG Yu-ping;KANG Kui;HE Hai-tao;HUA Cai-jian(School of Computing Science,Sichuan University of Science and Engineering,Zigong 643000,China;High Performance Computing Center,Sichuan University of Science and Engineering,Zigong 643000,China;Analytical and Testing Center,Sichuan University of Science and Engineering,Zigong 643000,China)

机构地区：[1]四川理工学院计算机学院,四川自贡643000 [2]四川理工学院高性能计算中心,四川自贡643000 [3]四川理工学院分析测试中心,四川自贡643000

出　　处：《计算机工程与设计》2019年第6期1585-1589,1600,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(21644013、11647029);四川理工学院教学改革研究基金项目(JG-1636);自贡市科技局重点科技计划基金项目(2016CXM05)

摘　　要：移动云计算中,移动设备电池充电间隔时间的不确定性会对移动设备服务请求的卸载决策和本地执行服务请求时的CPU运行频率产生重要影响。为解决这一问题,研究基于随机数据模型的移动设备控制策略问题,考虑充电间隔长度不确定的情况下,定义移动设备的执行性能和功耗的均衡目标函数,证明移动设备的最优控制决策在充电间隔变化时也会发生改变。以最大化目标函数为基础,提出一种动态规划算法对移动设备的最优控制策略(计算卸载决策和CPU运行频率)进行求解。数值仿真结果表明,在3种不同的充电时间间隔的概率密度函数下,所提最优控制策略比基准算法能够得到更好的性能。In mobile cloud computing, the uncertain of the battery inter-charging intervals in the mobile device has great effects on the offloading decision of the service requests from the mobile device and the CPU operating frequency for processing local requests. For solving this problem, the control strategy problem for the mobile device based on the stochastic data model was researched. Considering the uncertain of the inter-charging intervals, a trade-off objective function that captured a desirable trade-off between performance and power consumption of the mobile device was defined. And it is proved that best-suited control decisions should change as time elaspses to take into account the effect of inter-charging intervals length variations. Based on the maximization of the objective function, a dynamic programming algorithm was presented to solve the optimal control strategy for the mobile device (the computation offloading decision and the CPU operating frequency ). The numerical simulation results show that, under three kinds of probability density functions of inter-charging intervals, the proposed optimal control strategy can get better performance compared with the baseline algorithms.

关 键 词：移动云计算 随机数据模型 控制策略 动态规划 计算卸载 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]