Research On Relational Task Scheduling In Untrusted Environment Of Industrial Internet Of Things

The proposal of "Made in China 2025" action program marks that China has entered the stage of industrial artificial intelligence.Industrial Internet of things is the core technology to promote the traditional industrial manufacturing and production to achieve information and intelligence.Different from other Internet of things application technology,it has the characteristics of high real-time and high reliability.The traditional solution is to put the tasks of information processing and business decision-making into the cloud for computing.However,with the increasing number of industrial equipment,the traditional cloud computing model has the problem of high time delay,which can not meet the requirements of real-time task processing in industrial field.Therefore,edge computing is widely used as an effective method.It reasonably migrates the cloud computing tasks to each edge node of the Internet of things,which can effectively achieve load balancing and reduce processing delay.Due to the limited storage and computing power and different stability of edge nodes,not every edge node is trusted for tasks.Therefore,how to schedule tasks becomes an urgent problem in this kind of environment where edge nodes are not trusted.Based on the above background,in order to provide high-quality task scheduling service in the untrusted environment,this thesis constructs a quality of service trust model to evaluate the edge nodes and designs a high-quality edge nodes filter algorithm.Next,it schedules the associated tasks based on the high-quality edge nodes.This thesis designs the scheduling algorithm of the associated tasks based on the idea of genetic algorithm.The main work of this thesis is as follows:(1)Aiming at the untrusted environment in the industrial Internet of things application scenario,this thesis analyzes and determines the main factors that affect the quality of service of edge nodes scheduling,and constructs the quality of service trust model based on this,including time trust,behavior trust and resource trust.The edge node is evaluated from three dimensions.In order to filter out high-quality edge nodes faster,this thesis proposes an edge nodes filtering algorithm based on clustering coding skyline query.The algorithm first clusters and groups the edge nodes,and then performs dominating comparison operation according to the grouping.Compared with the traditional skyline query algorithm,it reduces the number of invalid dominating comparison and improves the efficiency of edge nodes filtering.(2)Tasks in the industrial Internet of things have the characteristics of relevance and need to be processed in real time.In order to schedule the associated tasks reasonably to meet the needs of industrial applications,this thesis proposes an associated task scheduling problem based on high-quality edge nodes,and proves that the scheduling problem is NP-Hard.In order to find an efficient task scheduling scheme,this thesis proposes a directed optimization genetic algorithm to schedule the associated tasks based on the idea of genetic algorithm.The method of randomly selecting the crossover mutation position in the genetic algorithm is changed to the selection method of adaptive probability,and the crossover mechanism is reconstructed to adapt to the scheduling problem in this thesis and improve the speed of the algorithm.Finally,this thesis sets up the corresponding comparative experiments for the two algorithms.The experimental results show that the proposed edge nodes filtering algorithm based on clustering coding skyline query can quickly and accurately filter out high-quality edge nodes,and the proposed associated task scheduling algorithm also has high performance in solving the scheduling scheme,which can ensure that the tasks are reasonably scheduled.