基于全局性能优化的移动负载转移决策策略的研究
发布时间：2018-09-19 09:10
【摘要】：随着智能手机的出现，移动设备的计算能力有了显著提高，而人们对于移动应用的复杂性也提出了相应的高要求，并且后者的提高速度远远高于前者的提高速度，解决好二者的不匹配性成为了移动计算发展必须要克服的问题。解决移动设备计算能力和人们对于移动应用复杂性要求的不匹配性问题，主要有两个解决方向，一是克服移动设备的资源有限性问题，这方面主要是通过硬件的提高来解决，二是更为高效的利用已有的有限的资源，来更好的满足日益复杂的移动应用的要求，移动负载转移是其中较为典型的方法。 移动负载转移决策策略是移动负载转移模型中的核心部分，但由于当前的负载转移决策大多数是静态决策，对于资源环境的变化应对能力较弱，不能实时保证决策的有效性。 为解决以上问题，本文提出了基于全局性能优化的动态决策策略，本文的动态决策策略主要包含两部分内容： （1）对于决策结果的有效性分析，在应用的执行过程中，周期性地调用资源监控模块得到当前的资源信息，分析当前决策的有效性，一旦发现错误决策，及时纠正。 （2）对于决策相关的计算模型的有效性分析，，通过分析以往的实际执行结果，分析当前决策相关的计算模型（应用的执行时间估算模型和应用的执行能耗估算模型）的有效性，一旦发现当前的计算模型的计算结果误差较大时，及时更改计算模型，以保证后续决策的有效性。 （3）对于一个应用的完整执行路径，分析历史执行路径的总执行时间和总能耗，根据需要分析基于单个目标函数的最佳决策，影响下一次的决策。
[Abstract]:With the emergence of smart phones, the computing power of mobile devices has been significantly improved, and the complexity of mobile applications has been put forward the corresponding high requirements, and the speed of the latter is far higher than that of the former. Solving the mismatch between them has become a problem that must be overcome in the development of mobile computing. There are two main solutions to solve the mismatch between the computing power of mobile devices and the complexity requirements of mobile applications. One is to overcome the problem of limited resources of mobile devices, which is mainly solved by the improvement of hardware. Second, more efficient use of existing limited resources to better meet the requirements of increasingly complex mobile applications, mobile load transfer is one of the more typical methods. The mobile load transfer decision strategy is the core part of the mobile load transfer model, but because most of the current load transfer decisions are static decisions, the ability to cope with the changes in the resource environment is weak, and the effectiveness of the decision cannot be guaranteed in real time. In order to solve the above problems, this paper proposes a dynamic decision strategy based on global performance optimization. The dynamic decision strategy in this paper mainly includes two parts: (1) for the effectiveness analysis of the decision results, in the implementation process of the application, Periodically call the resource monitoring module to obtain the current resource information, analyze the effectiveness of the current decision, once the wrong decision is found, correct it in time. (2) Analysis of the validity of the calculation model related to the decision. Based on the analysis of the previous actual execution results, the validity of the current decision related computing models (the applied execution time estimation model and the applied execution energy estimation model) is analyzed. Once the error of the current calculation model is found to be large, the calculation model can be changed in time to ensure the effectiveness of the subsequent decision. (3) for an application of the complete execution path, The total execution time and total energy consumption of the historical execution path are analyzed and the optimal decision based on a single objective function is analyzed according to the need to influence the next decision.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.53;TP333
本文编号：2249688
[Abstract]:With the emergence of smart phones, the computing power of mobile devices has been significantly improved, and the complexity of mobile applications has been put forward the corresponding high requirements, and the speed of the latter is far higher than that of the former. Solving the mismatch between them has become a problem that must be overcome in the development of mobile computing. There are two main solutions to solve the mismatch between the computing power of mobile devices and the complexity requirements of mobile applications. One is to overcome the problem of limited resources of mobile devices, which is mainly solved by the improvement of hardware. Second, more efficient use of existing limited resources to better meet the requirements of increasingly complex mobile applications, mobile load transfer is one of the more typical methods. The mobile load transfer decision strategy is the core part of the mobile load transfer model, but because most of the current load transfer decisions are static decisions, the ability to cope with the changes in the resource environment is weak, and the effectiveness of the decision cannot be guaranteed in real time. In order to solve the above problems, this paper proposes a dynamic decision strategy based on global performance optimization. The dynamic decision strategy in this paper mainly includes two parts: (1) for the effectiveness analysis of the decision results, in the implementation process of the application, Periodically call the resource monitoring module to obtain the current resource information, analyze the effectiveness of the current decision, once the wrong decision is found, correct it in time. (2) Analysis of the validity of the calculation model related to the decision. Based on the analysis of the previous actual execution results, the validity of the current decision related computing models (the applied execution time estimation model and the applied execution energy estimation model) is analyzed. Once the error of the current calculation model is found to be large, the calculation model can be changed in time to ensure the effectiveness of the subsequent decision. (3) for an application of the complete execution path, The total execution time and total energy consumption of the historical execution path are analyzed and the optimal decision based on a single objective function is analyzed according to the need to influence the next decision.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.53;TP333
【参考文献】
相关博士学位论文 前1条
1 伍宇;移动计算中自适应负载转移决策模型研究[D];复旦大学;2012年
本文编号：2249688
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2249688.html
