Virtualized base station for 5G cloud-RAN

Background and challenges

The 5G revolution is one of the most dynamically developed trends. It would change the way of using mobile services, cloud, fog and edge solutions. The IS-Wireless (ISW) use cases will show how the one of the most important elements of that revolution, 5g software defined base station will benefit using MORPHEMIC. ISW provides fully programmable Software Defined RAN functionality (SD-RAN) for mobile base stations supporting multiple radio access technologies (RATs) in parallel: 4G, 5G and WiFi. SD-RAN can be deployed as a set of network functions (NFs). Physical network functions (PNFs) run on typical base stations appliance or remote radio heads, whereas the virtual network functions (VNFs) can run on any cloud infrastructure and only require RRH for radio frontend.
Virtualization of RAN architecture allows the radio stack components to be flexible as regards its placement along cloud continuum as long as x-haul network requirements are met. The different options of locating base station (eNB) building blocks (i.e. protocols) are depicted in the figure below (CU – central unit, DU – distributed unit, RU – radio unit). The solution can be offered to 4G (LTE, LTE-A) and 5G cell vendors as the 3GPP-compliant future-proof protocol stack. It can be also provided as software as a service (SaaS) for operators or other service providers.

Challenges to tackle when thinking about fully cloud based mobile network deployments are (a)processing requirements of various protocols e.g. RLC/MAC/PHY, (b) optimal packaging (VM, Docker, Unikernel etc) of RAN protocols, (c) efficient composition of multiple atomic VNFs into bigger one, as well as (d) optimal placement of VNFs (cloud, edge, MEC) in order to maximize the benefits for network operator, service provider and customers. Here the MORPHEMIC framework is foreseen to become great enabler for all the stakeholders to ease the organizational burden related with RAN deployment (packaging/composition/placement of the SD-RAN functions) i.e. minimize OPEX and CAPEX as well). The level of flexibility is determined by (a) real capacities of transport networks (x-haul), (b) computing capabilities of the underlying NFVI/VIM solutions (e.g. do they support acceleration or not) (c) vertical applications which are planned to be deployed on top of such virtualized RAN infrastructure. Thus, the availability of the Morphemic optimization framework enables huge time savings due to automated deployment testing of SD-RAN with the help of sophisticated placement algorithms, given the heuristics which consider objectives and policies of network and infrastructure operators as well as vertical applications providers.

Scenario

The scenario foreseen consist of two essential cases: “case A” the use of MORPHEMIC tools for supporting network deployment options prior to the actual deployment and “case B” – validating a potential for utilizing the MORPHEMIC tools for suggesting optimization strategies for the already deployed mobile networks compliant with virtual RAN concept. For the “Case A” it is envisioned that a business customer requests the deployment of a network that fits his requirements e.g. “private network for public safety” or “ad-hoc network for a stadium”. The two configurations may certainly demand different performance, different security levels while offering also data centers (compute nodes) of various capabilities. In “case B” once location of a mobile network (virtual base station) was determined with the help of e.g. MORPHEMIC or without it, the optimal placement and configuration of the RAN radio stack may still be needed assuming the dynamically changing user traffic demands (i.e. the self-optimizing network), may in turn drive changes in VNF placement, scaling etc.

MORPHEMIC added-value

Case A&B

  • Automated identification of optimal placement strategies for virtual RAN building blocks, including multiple
    optimization criteria
  • Manual selection of cloud provider as well as optimal VNF placement and packaging gets self-adjusted
  • Mobile network deployment can easily be tailored for a vertical needs and multiple strategies evaluated
    automatically with the help of Morphemic
  • Automated assessment and evaluation of network deployment a priori
  • Predictive workload scaling and adaptation thanks to Morphemic toolboxes
  • Possibility of including operators’ policies in the network scaling and evolution process

KPIs

  • Network deployment cost (CAPEX) – target decrease by up to 30% (e.g. ca. 10mln EUR7 for a network of
    600 cells)
  • Network redesign time – savings up to 2 site visits per site/per year
  • Network deployment automation strategies – target 3-4 operator-oriented strategies will be evaluated (e.g.
    “performance / security tradeoff”, “low radio network capacity but high security”)