Broadband cellular networks are emerging to be the most common means for mobile data access worldwide. Predictions from industry analysts indicate that the volume of data through cellular data networks will increase exponentially in near future. Understanding of the mobile data traffic via measurement and analysis is critical for the development of resource management techniques for these networks. While spectrum resources are of great concern, this project specifically focuses on the ?energy? required to operate the cellular network infrastructure, specifically base stations. The project undertakes a significant modeling exercise with two goals. One goal is ?intellectual,? driven towards understanding the spatio-temporal dynamics of mobile traffic and discovering possible structure or relationships. The project uses state-of-the-art machine learning tools to develop models using large-scale data collected directly from the operators? networks. Such modeling will bring new insights that in turn will help to deploy and manage future generation cellular data networks. The second goal is ?utilitarian.? Here, techniques are developed to predict base station loads for use in resource management, specifically energy. Algorithms are designed to exploit energy-optimization opportunities to turn off specific network resources based on the forecasted load.
The project has significant broader impact. It develops technologies to appreciably reduce energy consumption in cellular networks. Overall, this exercise will both reduce cost, and contribute to the environment. The project also contributes to several 'green? initiatives in both institutions and to the education and training of graduate students.