Research On Efficient Task Offloading Strategy Forheterogeneous Dense Wireless Networks In Edge Computing

In the past 10 years,with the development of 5G and Internet of Things(IoT),the number of smart devices has exploded,and a large number of interactive applications have emerged,such as Virtual Reality(VR),Augmented Reality(AR),Autonomous Driving,and Image Recognition.The typical characteristic of these applications is that they require both intensive computing resources and very low latency,so they are not suitable for execution on smart devices with poor computing power and limited battery capacity.Traditional cloud computing technology cannot also guarantee the latency parameters because the device needs to communicate with the backbone network.To solve this dilemma,Mobile Edge Computing(MEC)technology came into being.Mobile Edge Computing means that smart devices offload their computing tasks to edge servers,which is deployed at the edge of the network(usually connected to edge gateways such as base stations),and utilizes the remaining computing power of the edge gateway for services.Communication between devices and edge servers requires a single-hop link in the best case,thus giving the opportunity to guarantee task execution latency.At present,Edge Computing technology has received extensive attention from academia and industry,but most of the existing researches ignore the dependencies between subtasks,competition in communication resources and the scheduling of computing resources on edge servers,making it difficult to adapt to the requirements of heterogeneous dense wireless networks under the next-generation basic communication architecture.In view of this,this article models the offloading tasks as Directed Acyclic Graph(DAG)in order to fully consider the data dependencies between subtasks and tap the potential parallelism between subtasks.At the same time,the offloading scenario of multi-heterogeneous devices covered by densely deployed heterogeneous edge servers is studied,and the wireless communication resource competition between devices and the task scheduling of edge servers are considered jointly.Through theoretical analysis,the offloading problem is modeled into the form of mathematical optimization,and the NP-hard property of the optimization problem is proved.In order to effectively solve this problem,this paper proposes a priority-based heuristic DAG task offloading scheduling algorithm.Based on this algorithm,a central offloading strategy is proposed,and the exponential computational complexity of this strategy is clarified.In order to improve the efficiency of decision-making,weigh the decision-making time and the benefit of offloading,this paper also utilizes game theory as a tool to propose a self-organized distributed offloading algorithm.After a finite number of iterations of this algorithm,it can achieve the state of Nash Equilibrium and export an efficient offloading strategy finally.Later,this paper studies the more complex offloading scenarios under edge-network collaboration,and proposes corresponding offloading strategies.In the end,through a series of simulations,the effectiveness of the proposed offloading algorithm is verified.