Research On Resource Allocation Method For Edge Computing First Networks In Industrial Scenarios

At the moment when industry 3.0 is approaching industry 4.0,the wide application of intelligent terminal equipment leads to the explosive growth of local production data,and the transformation of intelligent factory,intelligent production and intelligent logistics using unmanned machinery to replace traditional manual manufacturing and system platform to assist scientific decision-making is imminent.Represented by cloud computing,however,the traditional centralized network architecture can’t meet the industry under the Internet environment to the strict requirement of the task processing time delay.Edge computing sinks as the edge layer of cloud computing.Together with cloud computing,edge cloud collaborative network is constructed,which is widely used in the network architecture design of intelligent factories.At the same time,the concept of computing power network makes it possible for resources including computing,storage and network to be allocated and flexibly scheduled among cloud,edge and local areas on demand.Therefore,this paper proposes a hierarchical edge network collaboration architecture based on edge cloud collaboration and computing power network.In view of the industry under the background of the Internet the massive high heteropoly terminal intelligent factory environment,in this paper,the edge of the edge network based on hierarchical collaborative structure work force resources optimization scheduling,in order to ensure output and the cost of the factory.Specifically,first of all,this paper argues that even if the same local terminal equipment,produced in different time of the type and size of to-do tasks are not the same.Based on analytic hierarchy process(AHP),a quantitative evaluation system for the importance and urgency of real-time pending tasks of edge servers is designed.Then,in order to minimize the total delay and energy consumption of task processing,this paper subdivides the transmission,waiting,computing,migration delay and energy consumption into eight parts and takes them as constraints and takes the quantization urgency value of each task as coefficients to propose a multi-objective programming problem and establish a hierarchical edge network cooperative scheduling model.Then,in order to solve the NP hard problem,design a kind of immediate degree of perception is based on task force resource scheduling and allocation algorithm.Simulation results indicate that the hierarchical edge network collaboration architecture compared with the traditional edge cloud architecture,for edge work force utilization,and to handle task completion has significantly increased.At the same time,immediate degree of perception is based on task force resource scheduling and allocation algorithm is compared with the traditional scheduling algorithm and greedy scheduling algorithm,can effectively reduce the todo tasks discard rate and discarding the emergency degree and system overall delay processing tasks.The single working frequency of edge server provides fixed computing power,but the volume of real-time tasks to be processed of local terminal equipment is constantly changing.The computing power network of edge layer cannot realize dynamic intelligent matching and adjustment of computing power resource supply and demand by optimizing scheduling strategy,resulting in the waste of idle computing power resource.Therefore,this paper proposes a self-adaptation scheme of edge computing power network resources and tasks.First,in the original edge network collaborative scheduling based on the level of the multi-objective programming model,combined with the historical data and the task time flow,add tasks to each edge server to be processed real-time volume prediction model.Then,based on genetic algorithm,particle swarm optimization(PSO)algorithm and ridge regression algorithm based on machine learning and other commonly used prediction algorithm,this paper compares and analyzes the principle of the designed a real-time task size prediction algorithm based on ridge regression.The simulation results show that the use rate of edge computing force can be improved effectively by adding the prediction scheme.At the same time,the lack of sufficient historical data of the early stage of the reference and the local terminal equipment of a vast amount of this scenario,the ridge regression in convergence speed and accuracy,has more excellent performance.