No abstract

No abstract

No abstract

No abstract

Although deep learning has proven to be a successful and widely used technology across various industries, its drawbacks such as large models and difficulties in layout and maintenance in practical tasks have gradually become prominent. In view of the limitations and issues with the traditional method of measuring antenna parameters for Mobile Communication Base Stations (MCBS-APM), we are exploring the development of a new system that is designed to overcome the inefficiencies and potential risks associated with conventional labor-intensive methods. An effective measurement system which is composed of a Novel Instance Segmentation Network with Dual Attention and Focal Loss can accurately fathom out the antenna parameters in mobile communication base stations and completely subvert traditional measurement methods. To begin with, antenna video data is collected by unmanned aerial vehicle (UAV) which flies around the base station; then a designed instance segmentation network is employed to process and segment mobile communication base station antennas. At last, we implement real-time adjustments to control the actions of the UAV based on algorithmic measurements displayed on the accompanying mobile application. Our measurement system has been shown to greatly enhance measurement efficiency and accuracy, as evidenced by the results of our experiments. Quantitative results that are in line with industry standards show that our measurement system has strong robustness and reproducibility. Impact StatementThe angle of down-tilt for mobile communication base station antennas is critical in ensuring optimal network coverage and communication quality. Traditionally, measuring the antenna declination parameters has been inefficient and poses danger to engineers as they have to climb base stations. We propose a fully automated MCBS-APM system using UAV. Our effective measurement system, which is composed of a Novel Instance Segmentation Network with Dual