Operating a satellite network involves multiple disciplines and layers that may congest at some moments. In particular, sudden traffic demands due to massive events may produce congestion at traffic level, causing outage. This can be solved by performing smart and optimized resource management. However, managing resources in a satellite payload is not fast as desired and may take some minutes or hours to finalize a particular resource allocation. Therefore, having a forecast of traffic congestion is a necessary tool to anticipate and react before the congestion occurs. In this paper we propose a novel approach of traffic congestion prediction by using Deep Learning techniques to produce a forecast. Aimed by real data provided by a European satellite operator, we show that it is possible perform a forecast of traffic congestion in the following two hours to manage the resources more efficiently to reduce the outage under heavily congested moments

This work has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 101004215 (ATRIA) and by the Spanish ministry of science and innovation under project IRENE (PID2020-115323RB-C31/AEI/10.13039/501100011033) and grant from the Spanish ministry of economic affairs and digital transformation and of the European union – NextGenerationEU [UNICO-5G I+D/AROMA3D-Space (TSI-063000-2021-70).