Agile and Lightweight Learning for On-demand Networks (ALL-ON)

Mobile networks are becoming increasingly convoluted and unsustainable in their attempt to provide better services and enhance user experience. To avoid unnecessary resource over-provisioning, they could be dynamically scaled and optimized according to the user demands. The conventional network optimization methods and algorithms lack proactive and lightweight solutions for topology management and control. These deficits can be tackled with the help of carefully integrated ML solutions that can predict the demand change, learn system parameters interplay, and converge to the optimum faster. Our study will deliver novel analytical frameworks and practical ML solutions to optimize the energy consumption of the wireless networks on the fly. The outcomes of this research are expected to make future networks sustainable and intelligent.