Research On Network Resource Scheduling Strategy Of Smart Factory Based On Fog Computing

In recent years,smart factory has become one of the most popular industrial concepts.In addition,the development of the ⅡoT(Industrial Internet of Things)has also injected new vitality into the smart factory.The deployment of smart factory requires its network architecture to have the ability of data real-time communication,task timely processing and flexible arrangement of network equipment.However,the existing network resources in the smart factory is under-utilized in the traditional cloud computing based smart factory network architecture,in which the long-distance transmission leads to a large delay.Therefore,the cloud-based IIoT network can no longer meet the delay requirements of delay-sensitive tasks in smart factory.The emergence of fog computing makes up for the shortcomings of cloud computing.By migrating the processing of tasks closer to the device side,it shortens the storage,transmission and computing between the end device data of the Internet of things and the cloud center.By using the characteristics of fog computing architecture,such as computing marginalization,flexible arrangement,distribution and heterogeneity,it can better meet all kinds of applications in smart factory,such as intelligent production,equipment early warning,production material optimization and flexible factories.Aiming at the key issues in the smart factory network architecture,this paper proposes a smart factory network architecture based on fog computing by analyzing the characteristics and requirements of smart factory applications,and introduces the structure of the architecture and its network management method in detail.In addition,based on the proposed architecture,in order to reduce task processing delay and improve fog computing network resource utilization,according to the state of network resources(including computing resources,storage resources,communication resources,etc.)and task parameters in the smart factory,this paper proposes two fog computing network resource scheduling scheme,that is,a real-time task scheduling scheme based on dynamic priority and a non-real-time task scheduling scheme based on queuing theory.Among them,the real-time task scheduling scheme is designed for factory operation scenarios with complex task environment and task arrival rate change frequent.This scheme proposes a dynamic priority model to be used as a mechanism for fog nodes to execute tasks,and a continuous dynamic task offloading strategy is proposed.This strategy executes the offloading policy regularly by monitoring the status of network resources in real time,and establishes a task offloading model,which aims at minimum task processing delay and total value of task completion.And the model is solved by NSGA-Ⅱ algorithm to get the best task offloading strategy.The non-real-time task scheduling scheme is designed for factory operation scenarios with stable task arrival rates.The scheme models the network resource status and task processing by queuing theory,establishes a static task offloading model with minimum delay as the goal,and solves it by the improved discrete monkey group algorithm.Experiments show that the two task scheduling schemes have greet applicability and effectiveness in their respective application environments.Finally,based on the two proposed task scheduling schemes,this paper proposes a resource reservation based task scheduling scheme for the coexistence of tasks with stable arrival rate and tasks with frequent arrival rate changes.In this scheme,virtualization technology is used to provide flexible computing and storage resources for the factory.The network resources are reasonably divided into two parts:one is used for processing and scheduling of tasks with stable arrival rate,and the other is used for scheduling tasks with frequent changes in arrival rate.Simulation results show that the scheme can make full use of the network resources of the intelligent factory,greatly reduce the task processing delay,and improve the carrying capacity of network architecture to task flow.