A Multi-objective Task Offloading Strategy For Workflow Application In Mobile Edge-cloud Computing

Prompted by the remarkable progress in both edge computing and cloud computing,mobile edge computing has become a promising computing paradigm,where mobile users may take advantages of the low-latency property of edge computing and the rich-resource capacity of cloud computing to provide a high quality of service for mobile applications.In mobile edge computing,applications are offloaded to the server for execution in the form of workflow.Compared with traditional tasks,workflow considers the tasks dependency.This paper focuses on the workflow task offloading.The purpose of task offloading is to reduce energy consumption,workflow completion time and improve reliability to select the optimal resource to get the optimal migration strategy.In this paper,workflow offloading is defined as a constrained multi-objective optimization problem,and a hybrid algorithm involving differential evolution algorithm,artificial bee colony algorithm and decoding heuristic is developed to solve the NP hard problem of task allocation.Bee colony algorithm is an intelligent optimization algorithm inspired by bee foraging behavior.Through the division and cooperation of various types of bees,the optimal honey source can be found.Because this paper is a multi-objective optimization,we apply non-dominated sorting algorithm in the bee colony algorithm,and we calculate the crowding distance and get the set of non-dominant solutions for multi-objective optimization.Bee colony algorithms have the disadvantage of falling into local optimality,so the differential algorithm is embedded in the bee observation stage,and the differential evolution operator is used to update the selected honey source.This allows individuals in the population to quickly jump out of local optimality.The bee colony hybrid algorithm embedded with difference operator improves the global exploration ability and local development ability,and it helps bees to search for higher quality solutions,so as to obtain the optimal workflow offloading strategy.Its main objective is to minimize energy consumption,minimize workflow completion time and improve reliability.Workflow task offloading model is establishedFirstly,in the framework of mobile edge computing,the order and dependence of workflow are described.According to the analysis of the characteristics of workflow tasks,a multi-objective optimization model is established.Secondly,We proposed a multi-objective hybrid algorithm,called HABC,for finding the best offloading strategy.A differential evolution operator is embedded to artificial bee colony algorithm in order to improve the diversity of food sources.In addition,decoding heuristic algorithm is introduced into bee colony algorithm to generate a feasible task unloading strategy for workflow application.Finally,the effectiveness of the proposed method is evaluated.The results show,compared with other alternatives,our method can achieve a good balance between the goals,achieve lower energy consumption,shorter completion time and higher reliability.