NeTS: Small: A Learning Approach to Managing Cellular Network Upgrades

点击翻译按钮获取中文摘要

In order to provide a high-quality user experience, cellular service providers periodically upgrade their network software to introduce new features, fix software bugs, enhance quality of experience to users, or patch security vulnerabilities. The rollout of these upgrades need to be carefully managed, as tens of millions of customers rely on these networks for their daily activities (including emergency, navigation, and alert notifications). This project will develop techniques to manage cellular upgrades to minimize network disruption and improve user experience across the nation. Through collaboration with industry, the project will follow a fast track to technology transfer. The researchers will mentor undergraduate and graduate students, with a focus on actively recruiting female and other under-represented minority students.Prior to deploying a software upgrade fully over cellular network, field evaluations are conducted on a limited scale over a selected subset of base-stations. These field evaluations are typically cumbersome and can be time consuming; however, if done correctly they can help alleviate a lot of the deployment issues in terms of service quality degradation. Carefully selecting the specific base-stations to test, as well as the number of base-stations that are tested, is important -- too few could lead to problems during large-scale rollouts, and too many would be prohibitive in terms of cost. This project aims to develop learning-based approaches to automatically determine how many to select and where to conduct the upgrade field tests, where the learning methods are optimized to detect those base-station settings that could lead to failure of the upgrade. The researchers will evaluate the effectiveness of their approach using both real traces from major cellular networks as well as synthetic traces.