Service Based Virtual RAN Architecture for Next Generation Cellular Systems

Zeydan, Engin; Mangues‐Bafalluy, Josep; Baranda, Jorge; Requena, Manuel; Türk, Yekta

doi:10.1109/access.2022.3144534

Cited by 16 publications

(10 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus, the NG-RAN architecture must support seamless interoperability between the hardware and software components from multiple vendors while participating in the creation of a RAN slice [68]. Such an openness and interoperability in the NG-RAN architecture improves several features, including network malleability, programmability, intelligence, and security [69], [70]. It also reduces total network expenditure, extends virtualization to the extreme edge, strengthens the management and orchestration framework, brings versatility to the market, and improves underlying resource allocation.…”

Section: B the Network Function Layermentioning

confidence: 99%

“…Those that have not yet been virtualized require additional research in terms of their softwarization and cloudification. However, there is no doubt that they will be fully virtualized in the near future or long term [70]. Based on this, the functionalities that are accommodated by the vCU (also known as the higher layer functionalities) and vDU (also known as the lower layer functionalities) are assumed to be fully virtualized and could be instantiated as VNFs in an aggregation data center and an edge data center, respectively.…”

Section: B the Network Function Layermentioning

confidence: 99%

“…The interested readers are also suggested to refer to the relevant publications, such as [43], [72], [73], as well as the references therein, for a more in-depth understanding. These eight virtual functionalities of the vCU and vDU -as well as the VLs that connect them -running in aggregation and edge data centers require virtual resources such as virtual compute, networking, and storage [70], [74].…”

Section: B the Network Function Layermentioning

confidence: 99%

“…These two functionalities are assumed to be the PNFs of a gNB. Both of these physical functionalities and the PLs connecting them are implemented on top of customized equipment and physical resources such as antennas, radio frequency spectrum, Ethernet cables, optical fiber, and other dedicated hardware installed on Macro, Micro, Pico, and Nano cellular network infrastructure sites [70].…”

Section: B the Network Function Layermentioning

confidence: 99%

“…To accomplish this, the ENI System employs ML-assisted algorithms to (a) predict the virtual resource requirements of the VMs and VNs of the requested RAN slice based on the historical data or learning from an environment in steps 5 and 15; (b) configure the virtual compute and storage resources of VMs, as well as the virtual networking resources of VNs, in steps 8 and 18; and (c) dynamically allocate them to the VNFCs and VLs in steps 11 and 21. In comparison to the state-of-the-art VNF mapping algorithms, the use of ML-assisted algorithms in the preceding steps of the virtual resource allocation phase is critical for obtaining the performance objectives such as optimizing virtual resource allocation performance, minimizing energy consumption, and intelligently scaling the requested RAN slice [69], [70], [120].…”

Section: The Virtual Resource Allocation Phasementioning

confidence: 99%

See 4 more Smart Citations

Mapping the VNFs and VLs of a RAN Slice Onto Intelligent PoPs in Beyond 5G Mobile Networks

Habibi

Yousaf²,

Schotten

2022

IEEE Open J. Commun. Soc.

View full text Add to dashboard Cite

The mapping of a virtual network service onto a physical network infrastructure is a challenging task due to the joint allocation of virtual resources across nodes and links, the diverse technical requirements of end-users, the coordination between multiple host domains, and others. This issue is exacerbated further by the extension of virtualization to the next-generation radio access network (NG-RAN) architecture and the provisioning of radio access network (RAN) slicing. To that end, this article focuses on the mapping problem of the virtual network functions (VNFs), as well as their internal and external virtual links (VLs), of a RAN slice subnet onto intelligent points of presence (I-PoPs) and transport networks in the NG-RAN architecture. In this context, in contrast to the majority of the state-of-the-art proposals, which frequently fail to achieve performance objectives and neglect resource allocation constraints, this article introduces automation and intelligence at an architectural level to map VNFs and VLs onto their corresponding physical nodes and links, with the goal of achieving superior efficiency in virtual resource utilization while granting the performance of a RAN slice subnet. Benefiting from a top-down approach, the key contributions of this article are: (i) to extend the architectural framework of network slicing towards the NG-RAN architecture and provide a comprehensive overview and critical analysis of the components and functionalities of a RAN slice subnet; (ii) to integrate the Experiential Network Intelligence (ENI) framework into a joint architecture of the network functions virtualization-management and orchestration (NFV-MANO), Third Generation Partnership Project-network slicing management system (3GPP-NSMS), and I-PoPs in order to render automation and intelligence to the management and orchestration aspects of a RAN slice subnet in the NG-RAN architecture; and (iii) to propose a learning-assisted architectural solution for mapping the VNFs, as well as their internal and external VLs, of a RAN slice subnet onto the underlying I-PoPs and transport networks.

show abstract

Section: B the Network Function Layermentioning

confidence: 99%

Section: B the Network Function Layermentioning

confidence: 99%