Research On Edge Server Deployment In Smart Factories

Compared with traditional industries,intelligent manufacturing puts forward higher requirements for real-time production data processing and scalability of computing systems.As an extension and extension of cloud computing,edge computing has the characteristics of near-terminal and distributed deployment,which can solve the problems of network congestion and high latency that are prone to occur in cloud computing.Optimal deployment of edge servers is a key step in building an intelligent manufacturing edge computing system and improving edge server resource utilization.Therefore,the deployment methods of edge server and dynamic scaling to meet the requirements of low latency are proposed,and the main research contents are as follows:1.Edge Server deployments for low latency requirements.Firstly,aiming at the low service latency requirements of intelligent production lines,an edge server deployment problem model is constructed with time threshold and potential location as constraints,and low latency and load balancing as the optimization goals.Then,a Gapstatistic++ algorithm is designed to determine the number of edge servers,which is based on the number and distribution of intelligent terminal devices in the intelligent production line,and obtains the number of edge servers to be deployed by comparing the difference between the discreteness of the sample dataset to be classified and the reference data set.Finally,based on the K-means++ algorithm and the Hungarian algorithm,the initial heterogeneous edge server placement location and the association relationship between intelligent terminal devices and edge servers are obtained,and the final association is determined by minimizing the delay and the load balancing objective function constraints.The simulation results show that the algorithm meets the requirements of low-latency service of intelligent production line.2.Dynamically scaling deployment of Edge Servers.The increase in the number of terminal devices and computing tasks,the diversification of computing task types and the dynamic computing load brought about by the expansion of intelligent production lines may lead to problems such as excessive service delay and load imbalance of the original edge computing system.To solve this problem,a dynamic scaling deployment method for edge servers is proposed.Firstly,the dynamic scaling deployment model of edge servers is constructed with time thresholds and potential locations as constraints,and latency,load balancing,and scaling costs as optimization goals.Then,the ER-SSE algorithm and the Gap-statistic++ algorithm are combined to determine the optimal number of additional edge servers.Finally,a two-layer nested joint optimization algorithm(CM-DA)is designed,and the first layer first initializes the positioning of new edge servers and the association division of all intelligent terminal devices based on the multi-attribute similarity spectrum clustering algorithm,and then solves the dynamic scaling deployment model of edge servers and combines Metropolis decision-making to obtain the optimal deployment solution.In the second layer,the control node dynamically associates the overloaded edge server with the terminal device through real-time analysis and correlation matrix.Experimental results show that the algorithm meets the scalability requirements of intelligent production line for computing system.Figure [22] Table [6] Reference [60]...