The Covid pandemic has shown that fifth-generation (5G) and beyond networks must be considered part of our critical infrastructure, and that reliable and sustainable ICT service must be provided both in urban centers and in rural areas, including the most remote areas. For this to happen in a sustainable manner, renewable energy sources are being used to power network infrastructure, with power grids utilizing solar and wind energy. The management of network resources in these scenarios requires the adoption of sophisticated artificial intelligence solutions that can take into account the variability of both network demand and energy availability, both of which are subject to occasional anomalous behavior. We propose machine learning techniques that are robust to sudden shifts in traffic demand and energy supply for automated resource management decisions in future networks powered by hybrid energy sources, and will demonstrate their benefits using real network and energy datasets.

2023

2025