Automatic Analysis Framework for Campus Wireless Network Performance Based on Web Services and Supervised Learning Methods

This study proposes an automatic analysis framework for wireless network performance based on Web service and supervised learning methods, aiming to increase the efficiency of wireless network performance management and optimization through user-involved, real-time data collection and analysis. It features a Web interface for speed tests, simultaneously collects network speed and wireless access point performance metrics in real-time, and employs supervised learning to pinpoint the critical factors influencing network performance. This framework innovatively establishes a bridge between users’ real experiences and massive amounts of network measurement data, enhancing the efficiency and scope of data collection. By analyzing the speed test results of users and a wide range of feature variables, this framework accurately predicts the performance of wireless networks and reveals the impact of different characteristics on performance. Our research indicates that the linear regression model demonstrates high accuracy in predicting network performance and provides valuable insights, suggesting that there is primarily a linear relationship between features and network performance. By utilizing deep learning models, it is possible to capture the complex nonlinear relationships between features and network performance, thereby further enhancing the accuracy of the model. The framework demonstrates its application potential in actual deployments and offers a novel perspective for the optimization and management of wireless networks.