Elastic Scheduling of Micro-Modules in Edge Computing Based on LSTM Prediction

Abstract

With the rapid development of the Internet of Things and intelligent devices, edge computing, as a new computing model, has gradually become a key technology to solve the problem of data processing and resource scheduling. In this paper, an elastic scheduling technique of micro-modules based on edge computing is proposed to improve resource utilization and service stability in an edge computing environment. By introducing the LSTM model, this paper predicts the time series data of the edge micro-module service so as to realize dynamic resource scheduling and elastic scaling. The experimental results show that the LSTM model has excellent performance in micro-module elastic scaling and service request error rate, which is better than the traditional XGBoost, random forest, Ridge regression and logistic regression algorithms, and can effectively cope with load fluctuations in edge computing environment, and improve the performance and stability of the system. At the same time, combining active and passive elastic scaling strategies, the scheduling mechanism proposed in this paper can dynamically adjust resource allocation to meet the needs of different scenarios. Despite the good results achieved in the experiment, with the diversification of edge computing application scenarios, future research needs to further optimize the model to adapt to more complex edge computing environments and large-scale data processing requirements. The research in this paper provides the theoretical basis and practical guidance for resource scheduling and service optimization in edge computing and has important application prospects.