当前位置: X-MOL 学术J. Grid Comput. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Computing Offloading Strategy Using Improved Genetic Algorithm in Mobile Edge Computing System
Journal of Grid Computing ( IF 3.6 ) Pub Date : 2021-08-10 , DOI: 10.1007/s10723-021-09578-8
Anqing Zhu 1 , Youyun Wen 1
Affiliation  

For the current research on computing offloading, most of them only considers the multi-user task offloading decision problem or only considers the wireless resource and computing resource allocation. They have failed to comprehensively consider the impact of offloading decision and resource allocation on computing offloading performance, and it is difficult to achieve efficient computing offloading. For this reason, this paper proposes an edge computing task offloading strategy based on improved genetic algorithm (IGA). First, the weighted sum of task execution delay and energy consumption is defined as the optimization function of total overhead. Besides, the paper comprehensively considers the impact of users’ offloading decision, uplink power allocation related to task offloading and MEC computing resource allocation on system performance. Secondly, Genetic Algorithm (GA) is substituted to establish communication model, the offloading strategy is corresponding to the chromosome in algorithm and the gene is encoded by integer coding. Finally, IGA is used to solve the task to achieve efficient offloading. Among them, the use of integer coding, knowledge-based crossover and the mutation of population segmentation improves the optimization ability of this algorithm. Finally, experimental results show that the performance of IGA is the best, and the overall cost is about 52.7% of All-local algorithm and 28.8% of Full-edge algorithm.



中文翻译:

移动边缘计算系统中使用改进遗传算法的计算卸载策略

目前对计算卸载的研究,大多只考虑多用户任务卸载决策问题或者只考虑无线资源和计算资源分配。他们没有综合考虑分流决策和资源分配对计算分流性能的影响,难以实现高效的计算分流。为此,本文提出了一种基于改进遗传算法(IGA)的边缘计算任务卸载策略。首先,将任务执行延迟和能耗的加权和定义为总开销的优化函数。此外,本文综合考虑了用户的卸载决策、与任务卸载相关的上行功率分配和MEC计算资源分配对系统性能的影响。其次,代入遗传算法(GA)建立通信模型,算法中的卸载策略与染色体对应,基因采用整数编码。最后使用IGA解决任务,实现高效卸载。其中,整数编码的使用、基于知识的交叉和种群分割的变异提高了该算法的优化能力。最后,实验结果表明IGA的性能最好,总体成本约为All-local算法的52.7%和Full-edge算法的28.8%。IGA 用于解决任务以实现高效卸载。其中,整数编码的使用、基于知识的交叉和种群分割的变异提高了该算法的优化能力。最后,实验结果表明IGA的性能最好,总体成本约为All-local算法的52.7%和Full-edge算法的28.8%。IGA 用于解决任务以实现高效卸载。其中,整数编码的使用、基于知识的交叉和种群分割的变异提高了该算法的优化能力。最后,实验结果表明IGA的性能最好,总体成本约为All-local算法的52.7%和Full-edge算法的28.8%。

更新日期:2021-08-10
down
wechat
bug