Research On The Intelligent Decision Making For Industrial Edge Computing

With the continuous extension of the "Internet+" concept,the use of information technology and Internet platforms to achieve the deep integration of the Internet and industry has continued to develop,and Industry 4.0 has become a research hot word.Industrial production is faced with the problems of ever-increasing task volume,complex and ever-changing demand and unpredictable problems.In order to avoid problems such as task lag,network congestion,and low reliability during the industrial production process,edge computing mode can be used to reduce the delay to meet the real-time requirements of industrial production decision needs.How to coordinate the scheduling of multiple users and multiple servers with limited computing resources in the context of big data to meet the real-time and reliability requirements of the Industrial Internet of Things is a work of great research significance.This research fully analyzes the current domestic and foreign research on task scheduling in intelligent edge computing scenarios.Firstly,an intelligent decision-making model based on the industrial edge computing network architecture is proposed.It adopts the three-terminal cooperation model of the client,MEC(Multi-access Edge Computing),and the central processing unit(cloud)to meet the needs of big data processing.In order to make the system intelligent,a DG(Dynamic programming Greedy algorithm)algorithm-based intelligent task offloading scheme was designed to solve the problem of random offloading requests to the MEC server after collecting data from the client.For the problem that the MEC server receives multi-user and multi-tasking calculations at the same time,it is easy to cause a large delay.This paper proposes a repeat priority algorithm and a TMF algorithm(Topological sort and expected Minimum completion time First)and designs corresponding calculation offloading schemes.To optimize the execution time of a series of processes from sending offload requests to being processed.The main research work and innovations of this article are as follows:(1)Decision tree is a decision analysis method that utilizes the correlation of characteristic data streams,and is widely used in intelligent applications based on big data.Combining the current situation of large amounts of multitasking coexisting data in industrial production and the need for high real-time and reliability of the Industrial Internet of Things,This research makes reasonable use of the decision tree idea.The edge computing concept proposes an intelligent decision model for industrial edge computing from the perspective of characteristic data flow.In order to improve the success rate of task processing before the deadline,the user terminal sends a calculation offload request process to the MEC server,and the problem is modeled with the minimum lag time as the optimization goal.In order to improve the real-time reliability of the system,the MEC server receives multi-user and multi-tasking calculation and offloading process,and uses the average completion time of multi-tasking as the optimization target to model the problem.The model is evaluated from the two aspects of the amount of calculation data of the entire network and the processing time.The results show that the model can effectively reduce the system data processing amount and execution time,and reduce the calculation pressure of the MEC server.(2)Aiming at the problem of user offloading tasks to the MEC server,the original problem was first reduced to the classic 0-1 backpack problem,thus proving that the problem in this research is NP-hard,no optimal solution in polynomial time;Secondly,the DG heuristic algorithm is proposed,that is,the task quote is adjusted in real time according to the current remaining available bandwidth of the MEC server,so that more tasks can be processed on the MEC server before the deadline;Finally,simulation is performed.The experimental results show that the algorithm can effectively reduce task lag time and task lag rate,and meet the needs of large-scale data real-time processing in an industrial context.(3)Aiming at the problem that the MEC server receives multi-user and multi-tasking computing offloading,combined with the large amount of data in the context of big data,and the fact that the MEC server has limited computing capabilities,Firstly,considering the idea of machine learning decision tree,the feature data streams corresponding to all tasks are screened to avoid unrelated data from participating in the calculation of unloading.At the same time,the feature of repetitive multi-task feature data streams is used to propose a repeat priority algorithm.Simulation results show that the algorithm has obvious advantages in scenarios with high multitasking repetition rate,greatly improving system performance.Aiming at the situation where the repetitive priority algorithm is not effective under the scenario of low repetition rate of multiple tasks,the priority of the characteristic data stream of the decision tree is considered,and the topology sorting idea is introduced to propose the TMF algorithm,which can effectively prevent the MEC server from having too many caches and insufficient capacity,which will cause the task to fail.Issues addressed.Simulation results show that the algorithm has obvious advantages in multi-task low repetition rate scenarios,and can meet the real-time requirements of the Industrial Internet of Things while improving system throughput.