Research On Task Aware Service Caching Algorithm In Edge Computing Networks

With the development of Internet of things and mobile communication technology,the number of various devices and data volume have increased exponentially in recent years.At the same time,there are many emerging applications that are computationally intensive and delay sensitive,such as automatic driving,virtual / augmented reality(VR /AR)and face recognition.However,due to the limited computing,storage resources and battery capacity,intelligent terminal devices are not suitable for long-time complex computing and can not meet the demand of computing delay.At the same time,although traditional cloud computing can provide powerful computing power and sufficient resources,it still can not meet the requirements of real-time due to the introduction of excessive communication delay.Mobile edge computing technology came into being in this context.By deploying computing and storage resources near the network edge to provide services,mobile edge computing can reduce communication delay and the pressure on the backhaul network.Therefore,it has become a new computing paradigm to solve the contradiction between the high computing delay of local computing and the high communication delay introduced by cloud computing.Although mobile edge computing has received extensive attention and achieved many research results,the existing relevant research generally ignores the impact of service caching on task offloading strategy in edge network,as well as the dependency and fine-grained division within complex tasks,making the resource utilization of edge server is not enough,so it is difficult to meet the delay and energy consumption requirements of emerging applications.In this paper,a general directed acyclic graph task dependency model is used,and the constraints of the service caching of edge servers with limited storage resources on offloading strategy and the uncertainty of task generation and its workload are considered.At the same time,we uses queuing theory to model the execution process of subtasks on the edge server,and fully tap the parallelism in the task dependency structure,so as to minimize the overall service delay of the entire task.The offloading strategy for dependent tasks when the service caching decision on edge servers has been determined and the joint optimization of service caching strategy and computing resource allocation strategy in the edge network are studied in this paper.We proved NP-hard properties of optimization problems by derivation,then proposed heuristic algorithms which are based on the task dependency structure and the attributes of each edge server to obtain the strategy aiming at minimizing the service delay.Finally,through sufficient simulation experiments,the proposed algorithms are compared with the latest related algorithms to prove the effectiveness and performance advantages of the proposed algorithms.