移动Ad Hoc云环境中基于移动性预测的计算卸载算法研究
发布时间：2018-08-28 11:23
【摘要】：以智能手机和平板电脑为代表的移动设备的快速普及给人们的日常生活带来了极大的便利。但是,由于受到CPU性能、电池容量、存储容量等因素的限制,移动设备在处理计算密集型的任务时却表现欠佳,比如运算速度缓慢、掉电迅速等。移动云计算技术的出现为该问题的解决提供了一个很好的思路：通过把移动客户端上的计算密集型任务卸载到目标代理中去执行不仅可以大大缩减任务的处理时间而且还可以最大限度的降低移动设备的能耗。 在移动Ad Hoc云环境下,移动客户端节点和目标代理节点的位置都时刻处于动态变化的状态,受无线网络有效覆盖范围的限制和节点移动性的影响,客户端节点和代理节点之间的网络连接也是间歇性的。因此,当把计算任务从移动客户端节点卸载到代理节点中去执行时就有可能会出现计算卸载失败的问题。传统的计算卸载算法虽然在静态卸载环境下表现出色,但是在移动Ad Hoc云环境下却难以克服计算卸载失败所带来的问题。 为了解决移动Ad Hoc云环境下计算卸载失败的问题,本文在已有的静态卸载算法MET、MCT、MinMin、MaxMin、Sufferage的基础上提出了五个适用于动态卸载环境的新算法DynMETComm DynMCTComm、DynMinMinComm、DynMaxMinComm和DynSufferageComm。与传统的静态卸载算法相比,这些新算法不仅考虑了任务在卸载过程中的通信开销而且增加了对任务卸载失败后的处理,即当卸载到代理节点上的任务执行失败时,更新任务的到达时间为失败时间点,从而参与后续的调度。为了最大限度的避免任务卸载失败所产生的开销,本文又提出了基于移动性预测的新算法DynPredict。该算法在预测到任务执行失败时,会从有能力执行成功的代理节点中选择一个次优的代理重新进行卸载,从而保证任务顺利执行完成。仿真结果表明：带预测的算法DynPredict在大多数性能指标上都能表现出最优的性能,页DynMETComm则表现最差；在线算法DynMCTComm则通常接近甚至超过其他三个比较复杂的批调度算法(包括DynMinMinComm、DynMaxMinComm和DynSufferageComm)的性能。 本文的研究成果可以很好地应用于移动Ad Hoc云环境下的计算卸载,也为下一步任务迁移方向的研究奠定了基础。文中所采用的研究方法和思路对于移动云计算环境下的计算卸载的深入研究也具有一定的参考价值。
[Abstract]:The rapid popularization of mobile devices, represented by smart phones and tablets, brings great convenience to people's daily life. However, due to the limitations of CPU performance, battery capacity, storage capacity and other factors, mobile devices perform poorly in processing computationally intensive tasks, such as slow computing speed, rapid power loss, and so on. The emergence of mobile cloud computing technology provides a good way to solve this problem: by uninstalling computationally intensive tasks on mobile clients to target agents to perform tasks that can significantly reduce the processing time It can also minimize the energy consumption of mobile devices. In mobile Ad Hoc cloud environment, the location of mobile client node and target proxy node is always in a dynamic state, which is affected by the limit of effective coverage of wireless network and the mobility of nodes. The network connection between client node and proxy node is also intermittent. Therefore, when the computing task is unloaded from the mobile client node to the proxy node, the problem of computing uninstall failure may occur. Although the traditional computing unload algorithm performs well in static uninstall environment, it is difficult to overcome the problem of computing uninstall failure in mobile Ad Hoc cloud environment. In order to solve the problem of computing unload failure in mobile Ad Hoc cloud environment, this paper proposes five new algorithms, DynMETComm DynMCTComm,DynMinMinComm,DynMaxMinComm and DynSufferageComm., which are suitable for dynamic uninstall environment, based on the existing static uninstall algorithm MET,MCT,MinMin,MaxMin,Sufferage. Compared with the traditional static unload algorithm, these new algorithms not only consider the communication overhead of the task during the uninstall process, but also increase the processing of the task unload failure, that is, when the task unloaded to the proxy node fails to execute, The time of arrival of the update task is the point of failure, so as to participate in the subsequent scheduling. In order to avoid the overhead caused by unload failure, a new algorithm DynPredict. based on mobility prediction is proposed in this paper. When the task execution failure is predicted, the algorithm selects a sub-optimal agent from the agent node with the ability to execute successfully, so as to ensure the smooth completion of the task. The simulation results show that the DynPredict algorithm with prediction can show the best performance on most performance indexes, while the page DynMETComm shows the worst performance. DynMCTComm, an online algorithm, usually approaches or exceeds the performance of the other three more complex batch scheduling algorithms (including DynMinMinComm,DynMaxMinComm and DynSufferageComm). The research results in this paper can be applied to computing unload in mobile Ad Hoc cloud environment, and it also lays a foundation for further research on the direction of task migration. The research methods and ideas used in this paper also have a certain reference value for the further study of computing uninstall in mobile cloud computing environment.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN929.5
本文编号：2209251
[Abstract]:The rapid popularization of mobile devices, represented by smart phones and tablets, brings great convenience to people's daily life. However, due to the limitations of CPU performance, battery capacity, storage capacity and other factors, mobile devices perform poorly in processing computationally intensive tasks, such as slow computing speed, rapid power loss, and so on. The emergence of mobile cloud computing technology provides a good way to solve this problem: by uninstalling computationally intensive tasks on mobile clients to target agents to perform tasks that can significantly reduce the processing time It can also minimize the energy consumption of mobile devices. In mobile Ad Hoc cloud environment, the location of mobile client node and target proxy node is always in a dynamic state, which is affected by the limit of effective coverage of wireless network and the mobility of nodes. The network connection between client node and proxy node is also intermittent. Therefore, when the computing task is unloaded from the mobile client node to the proxy node, the problem of computing uninstall failure may occur. Although the traditional computing unload algorithm performs well in static uninstall environment, it is difficult to overcome the problem of computing uninstall failure in mobile Ad Hoc cloud environment. In order to solve the problem of computing unload failure in mobile Ad Hoc cloud environment, this paper proposes five new algorithms, DynMETComm DynMCTComm,DynMinMinComm,DynMaxMinComm and DynSufferageComm., which are suitable for dynamic uninstall environment, based on the existing static uninstall algorithm MET,MCT,MinMin,MaxMin,Sufferage. Compared with the traditional static unload algorithm, these new algorithms not only consider the communication overhead of the task during the uninstall process, but also increase the processing of the task unload failure, that is, when the task unloaded to the proxy node fails to execute, The time of arrival of the update task is the point of failure, so as to participate in the subsequent scheduling. In order to avoid the overhead caused by unload failure, a new algorithm DynPredict. based on mobility prediction is proposed in this paper. When the task execution failure is predicted, the algorithm selects a sub-optimal agent from the agent node with the ability to execute successfully, so as to ensure the smooth completion of the task. The simulation results show that the DynPredict algorithm with prediction can show the best performance on most performance indexes, while the page DynMETComm shows the worst performance. DynMCTComm, an online algorithm, usually approaches or exceeds the performance of the other three more complex batch scheduling algorithms (including DynMinMinComm,DynMaxMinComm and DynSufferageComm). The research results in this paper can be applied to computing unload in mobile Ad Hoc cloud environment, and it also lays a foundation for further research on the direction of task migration. The research methods and ideas used in this paper also have a certain reference value for the further study of computing uninstall in mobile cloud computing environment.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN929.5
【参考文献】
相关期刊论文 前2条
1 朱雨虹;徐昌庆;;基于无线局域网的室内覆盖分布系统模型[J];计算机工程;2008年07期
2 陈岭;许晓龙;杨清;陈根才;;基于三次样条插值的无线信号强度衰减模型[J];浙江大学学报(工学版);2011年09期
，本文编号：2209251
本文链接：https://www.wllwen.com/kejilunwen/wltx/2209251.html
