A New Partial Task Offloading Method in a Cooperation Mode under Multi-Constraints for Multi-UE  

作　　者：Shengyao Sun Ying Du Jiajun Chen Xuan Zhang Jiwei Zhang Yiyi Xu 

机构地区：[1]School of Information Science and Technology,Zhengzhou Normal University,Zhengzhou,450044,China [2]Henan Key Laboratory of Big Data Analysis and Processing,Henan University,Kaifeng,475004,China [3]School of Geography and Tourism,Zhengzhou Normal University,Zhengzhou,450044,China [4]Science and Engineering College,South China University of Technology,Guangzhou,510641,China [5]Department of Electrical and Electronic Engineering,Luohe Vocational Technology College,Luohe,462002,China [6]School of Computer Science,Beijing University of Posts and Telecommunications,Beijing,100876,China [7]Cardiff School of Engineering,Cardiff University,Cardiff,CF103XQ 15,UK

出　　处：《Computers, Materials & Continua》2023年第9期2879-2900,共22页计算机、材料和连续体（英文）

摘　　要：In Multi-access Edge Computing(MEC),to deal with multiple user equipment(UE)’s task offloading problem of parallel relationships under the multi-constraints,this paper proposes a cooperation partial task offloading method(named CPMM),aiming to reduce UE’s energy and computation consumption,while meeting the task completion delay as much as possible.CPMM first studies the task offloading of single-UE and then considers the task offloading ofmulti-UE based on single-UE task offloading.CPMMuses the critical path algorithmto divide the modules into key and non-key modules.According to some constraints of UE-self when offloading tasks,it gives priority to non-key modules for offloading and uses the evaluation decision method to select some appropriate key modules for offloading.Based on fully considering the competition between multiple UEs for communication resources and MEC service resources,CPMM uses the weighted queuing method to alleviate the competition for communication resources and uses the branch decision algorithm to determine the location of module offloading by BS according to the MEC servers’resources.It achieves its goal by selecting reasonable modules to offload and using the cooperation ofUE,MEC,andCloudCenter to determine the execution location of themodules.Extensive experiments demonstrate that CPMM obtains superior performances in task computation consumption reducing around 6%on average,task completion delay reducing around 5%on average,and better task execution success rate than other similar methods.

关 键 词：MEC partial task offloading parallel dependencies completion delay 

分 类 号：O15[理学—数学]