The TCO of virtualizing RAN in mobile operator infrastructures

Mobile 3D gaming apps, autonomous vehicles, industrial IoT, remote surgery, smart cars, and smart cities – mobile operators’ 5G network architectures are preparing to enable these and other advanced wireless use cases as they come available. 5G supports high-speed, low power and very low-latency connectivity to fulfill the potential of many scenarios, delivered with automation and the required performance on cloud-native distributed architectures.

Communications service providers (CSPs) have big decisions to make as the service capacity of their mobile networks grows and their plans for 5G architecture take shape. Virtualized system infrastructures will be increasingly critical - from centralized core operating sites to the enterprise edge and radio access network (RAN) application sites. 

Virtualized RAN (vRAN) offers baseboard functions running as virtualized network functions (VNFs) or cloud-native functions (CNFs) on commodity hardware. Using vRAN is a strategic decision point in the future of 5G networks, as it is expected to provide lower TCO, increased agility and flexibility, performance gains, and scalability. Operators are determining whether they should begin their vRAN deployments in existing 4G/LTE networks as a means of improving their economics now, and they are exploring how they may gain increased benefits by applying this approach to accelerate deployments of their 5G infrastructure. 

Economics at network core

Red Hat worked with ACG Research, in a sponsored examination, to determine which deployment model  offered the greatest advantages for CSPs. A TCO comparison was made between two scenarios for running a mobile operator’s 4G virtual evolved packet core (vEPC) and virtualized IP Multimedia Subsystem (vIMS) implementation, pitting deployment of multiple vertically integrated (silo-based) solutions against deployment of all services and applications in an integrated horizontal cloud. 

We used ACG Research’s Business Analytics Engine (BAE) to compare the infrastructures. The model was based on a representative national/tier 1 operator’s mobile network in three metropolitan regions with approximately 12 million subscribers and 5,000 cell sites between urban and rural areas. 

To arrive at the TCO we analyzed both capital expense (CapEx) and operational expense (OpEx) of each alternative over five years.

CapEx did not differ between the two but OpEx was significantly different. 

Four separately designed and run infrastructures made up the siloed case. Managing and securing four silos, obtaining a larger number of virtual infrastructure software licenses, and onboarding new hardware in the vertically integrated solution create significantly more OpEx than the other scenario. 

In contrast, the horizontal infrastructure is 41 percent more efficient than that of the siloed deployment, and the TCO over five years is 30 percent lower, according to the analysis.

Economics at the edge

To demonstrate the economic benefits of virtualizing components in the RAN, we modeled the lifecycle of a conventionally distributed RAN (DRAN) versus a vRAN 4G deployment in a tier 1 mobile operator environment. The model considered RAN components that were architecturally and topologically separate from the mobile network core. The representative network supports 10 million subscribers and has 12,000 existing radio sites. We posited that, over the course of a five-year lifecycle in the model, growth in traffic will increase the coverage of the network in urban, suburban, and rural areas,requiring  expansion to an additional 11,000 radio sites.

In this model, the DRAN had higher CapEx and higher OpEx than the vRAN, both contributing to a higher TCO. The DRAN option had a CapEx two times that of the vRAN, primarily due to the cost of BBU equipment at each cell site in the DRAN, compared with the cost of the vRAN’s infrastructure for BBU processing at fewer server sites. The DRAN implementation accrued greater OpEx from its higher site rental, BBU maintenance, fiber leases, and power and cooling costs compared with the smaller footprint and lower costs of the vRAN design.

In summary, centralized vRAN architectures were projected to deliver up to 44 percent lower TCO than conventional distributed RANs if all existing and future cell sites are virtualized.  Operators were projected to gain up to a healthy 27% improvement in TCO, even if they choose to virtualize only growth sites.

Smart steps to the 5G future

As 5G matures, we will be able to examine its operational and deployment costs in more detail and similarly analyze its economic benefits. We believe, though, that building 5G on virtualized cloud-native distributed infrastructures, using containerized networking apps and modular microservices, will only help increase the value of open horizontal cloud platforms.

Wherever operators are in their journey to 5G, they should be aware that building and supporting their own collection of open source community software elements may come with significant additional costs and risks compared to deploying with fully supported open source solutions. Economic analyses carried out in parallel to the studies outlined in this paper indicate that using Red Hat open source solutions for telco cloud deployments enables TCO advantages of up to 35 percent over DIY approaches.

Please investigate our March 10, 2020, webinar (now on demand), “Economic benefits of virtualizing the RAN in mobile operators’ Infrastructures,” to learn more details about our research and results.