Sixth-generation (6G) systems are expected to inaugurate the era of massive and extremely heterogeneous network slicing, where tenancy would be extended to the final consumer with the proliferation of advanced and diverse digital services. This introduces significant challenges to network management and orchestration in terms of scalability and sustainability. This article starts by elaborating on the architectural and artificial-in-telligence-based algorithmic designs to achieve energy efficiency (EE) in 6G, as well as analyzing their trade-offs. The article then introduces a novel statistical federated learning (StFL)-based analytic engine for zero-touch 6G massive network slicing, which performs slice-level resource prediction by learning in an offline fashion while respecting some preset long-term service level agreement (SLA) constraints defined in terms of the empirical cumulative distribution function and the percentile statistics, and hence uses a new proxy-Lagrangian two-player strategy to solve the local non-convex federated learning task without settling for surro-gates only. This guarantees 20 × lower service-lev-el agreement (SLA) violation rate with respect to the federated averaging (FedAvg) scheme, while achieving more than 10 × EE gain compared to an SLA-constrained centralized deep learning algorithm, which paves the way to sustainable mas-sive network slicing. Finally, the key open research directions in this emerging area are identified.
Zero-touch AI-driven distributed management for energy-efficient 6G massive network slicing
IEEE Network, Vol.35, N°6, November/December 2021
© 2021 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
PERMALINK : https://www.eurecom.fr/publication/6791