CAREER: Towards Self-Managing Wireless Networks Using M3

University Of Texas At Austin, Austin TX

Investigators

Abstract

The goal of this research is to automate two core management functionalities for wireless networks - performance management and fault management. These functionalities are critical to providing high-quality and reliable wireless communications. To achieve this goal, this research develops a systematic framework, called M3, which measures, models, and manages wireless networks. M3 uses measurement data to drive model construction. In this way, the model can more closely reflect a given network than a generic network model. In the meantime, the simplicity of the network model permits efficient analysis and control. To realize the framework, the research develops novel methodology for measuring and inferring wireless interference and traffic demands in a scalable and accurate fashion. Based on the measurement data, network embedding is used to construct simple and flexible models to facilitate network analysis and control. The research further develops robust optimization, stochastic programming, and outlier detection techniques for network optimization, troubleshooting, and recovery. To demonstrate the practical values of the approaches, this work will develop a complete toolkit for wireless network measurement, modeling, and management, and apply them to managing wireless networks at home, on campus, in enterprises, and at medical clinics. The techniques and toolkit developed can significantly improve current wireless network operations and facilitate new wireless network development. Finally, the research results of the project will be integrated into undergraduate and graduate curriculum.