RT Conference Proceedings
T1 𝛼-OMC: cost-aware deep learning for mobile network resource orchestration
T2 Alfa-OMC: cost-aware deep learning for mobile network resource orchestration
A1 Bega, Dario
A1 Gramaglia, Marco
A1 Fiore, Marco
A1 Banchs Roca, Albert
A1 Costa-Pérez, Xavier
AB Orchestrating resources in 5G and beyond-5G systemswill be substantially more complex than it used to bein previous generations of mobile networks. In order to takefull advantage of the unprecedented possibilities for dynamicreconfiguration offered by network softwarization and virtualizationtechnologies, operators have to embed intelligence innetwork resource orchestrators. We advocate that the automated,data-driven decisions taken by orchestrators must be guided byconsiderations on the cost that such decisions involve for theoperator. We show that such a strategy can be implemented viaa deep learning architecture that forecasts capacity rather thanplain traffic, thanks to a novel loss function named alfa-OMC. Weinvestigate the convergence properties of alfa-OMC, and providepreliminary results on the performance of the learning processin case studies with real-world mobile network traffic.
YR 2019
FD 2019-09-23
LK https://hdl.handle.net/10016/28332
UL https://hdl.handle.net/10016/28332
LA eng
NO Proceeding of: The 2nd International Workshop on Network Intelligence (NI 2019): Machine Learning for Networking (part of 2019 IEEE International Conference on Computer Communications (IEEE INFOCOM 2019)), 29 April, 2019, Paris, France
NO The work of University Carlos III of Madrid was supported by the H2020 5G-MoNArch project (Grant Agreement No. 761445) and the work of NEC Europe Ltd. by the 5GTransformer project (Grant Agreement No. 761536).
DS e-Archivo
RD 27 jul. 2024