CE25 - Sciences et génie du logiciel - Réseaux de communication multi-usages, infrastructures de hautes performances

HiErarchIcal DIsaggregated Scheduling for beyond-5G networks – HEIDIS

Submission summary

The entire telecom industry is going through a profound transformation driving the move towards open architectures and software-based networks. This trend is being moved ahead faster and gained momentum thanks to open-source software and standards for communication infrastructure components. On the one hand, an open architecture approach can help operators emancipate themselves from vendors' lock-in and the derived high operational and capital expenditures. On the other hand, it can also help vendors escape complex and high-barrier hardware design and production lines, and focus instead on advanced functionalities, interfaces and software life-cycle maintenance and licensing models. As an initiative to drive openness and intelligence for the next-generation wireless networks, the Open Radio Access Network (O-RAN) alliance was recently founded by five major service providers, including Orange, to break the last barrier in the development of fully softwarized radio access networks. In legacy RAN systems, hardware components and proprietary software codes are tightly coupled, so that atomic closed implementations do not have interfaces to support the interoperability between different vendors. The resulting ossification of the RAN generates vendor-locked islands in the cellular backhauling provider networks. The virtualization of network functions and the open-interfaces enabled in O-RAN allows software-based network functions and naturally, enables interoperability between different components, software editors and vendors. Moreover, it unveils the disaggregation of RAN functions, permitting the independent development of each RAN function, leading to innovation and flexibility in scaling capacity. Compared to the traditional RAN, a novel three-layer scheduling logic appears at the forefront in O-RAN, composed of non-real-time, near-real-time, and real-time layers. These new decision-making layers have to interact, and their algorithmic behavior, depending on each other, has still to be defined, evaluated, and experimented. The non-real-time decision-making is left to the so-called orchestration layer, where a non-real-time RAN Intelligent Control (Non-RT RIC) function takes long-term decisions. The Non-RT RIC can be built on ONAP (Open Network Automation Platform), another open-source initiative that was chosen by the O-RAN alliance as reference orchestration/network management layer, and current proofs-of-concept (PoC) use both ONAP and ORAN. ONAP supports intelligent RAN optimization in non-real-time using data analytics and AI/ML training and inference. The Near-RT RIC is a new intermediary stack between orchestrators (non-RT layer) and RT schedulers: it is new in its notion (not existing before) and in its function, i.e., algorithmically linking real-time scheduling decisions to orchestration decisions. More precisely, it supports near real-time control and optimization of O-RAN resources and nodes with near real-time control loops. It can also pilot the behavior of RT schedulers. Driven by this evolution in the radio access segment, the full end-to-end resource scheduling, allocation and provisioning chain is going to change. To some extent, this end-to-end network evolution is expected to follow and link to the redesign of the radio access, with a 3-layer scheduling as well. HEIDIS will study and experiment the linkage among radio, link and computing resource management architectures and algorithms, and in particular coordination challenges among the resulting 3-layer vertical and horizontal scheduling holistic architecture.

Project coordination

Sahar HOTEIT (Laboratoire des Signaux et Systèmes)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

SMILE SMILE
CEDRIC CENTRE D'ETUDES ET DE RECHERCHE EN INFORMATIQUE ET COMMUNICATIONS
L2S Laboratoire des Signaux et Systèmes
Orange SA Orange Labs Networks / Service TGI/OLN/SAS/WBH
EDF ELECTRICITE DE France, Direction R&D / Département de Recherche SYSTEME

Help of the ANR 784,767 euros
Beginning and duration of the scientific project: January 2022 - 36 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter