Senza Fili | Tech Brief

What’s the best way to get to Open RAN?

Optimizing pooling gains can reduce TCO by up to 42%

Monica Paolini, Senza Fili

©Senza Fili 2021 What’s the best way to get to Open RAN? |1|
 What’s the best way to get to Open RAN?
There are multiple migration paths to Open In an Open RAN deployment, should the DUs be at the cell site or co-located in a
RAN. Each operator has the flexibility to pick the remote data center with the DUs? It depends on transport costs, which unlock
path best suited to its needs and what is different sources of pooling gains.
available and cost effective in its network. Operators with low transport costs can reduce the TCO by 36% to 42% by locating
the DUs in a centralized location with the CU.
In this paper, we compare the TCO for two
operator cases. In the first case, the operator has Operators with high transport costs can reduce the TCO by 25% to 12% by
access to low-cost transport and may own a moving the DUs to the cell site, and leaving the CU in a centralized location.
fiber network that reaches most of its macro cell
sites. In the second case, the operator is in the
opposite situation: it faces high transport costs
and limited fiber availability at the macro cell
sites. This is likely to change over the next few
years, but the operator does not want to wait
until then to deploy Open RAN.

How should the adoption of Open RAN differ in

the two cases? And what does this tell us about
all the operators that find themselves in a
midway situation?

Which Open RAN is This is a follow-up to our earlier “Which Open RAN is best for you?“ paper based on the same TCO model. The model examines the
financial impact of Open RAN architecture choices under variable costs and resource availability. The first paper was on how
best for you? transport costs affect the topology of Open RAN deployments, and it provided a more detailed description of the underlying model.
The current paper looks at how operators can maximize pooling gains as they transition to Open RAN.
Get the companion
©Senza Fili 2021 What’s the
of best
Openway toarchitectures
get to Open over
RAN?traditional RAN architectures in three earlier papers,
|2| “Future
paper We demonstrated the TCO advantage RAN
proofing mobile network economics,” “How much can operators save with a Cloud RAN?“ and “In-building virtualization.”
Open RAN is a fundamentally new approach to planning for, deploying and Migration paths to Open RAN:
operating the RAN. In Open RAN, virtualization, disaggregation, open interfaces and
multi-vendor coexistence create a more flexible and dynamic wireless network,
What are the options?
coupled with a core that is undergoing a similar transformation toward openness.
These benefits of Open RAN are truly valuable only to the extent to which they also Mobile operators have increasingly committed to moving to Open RAN,
increase efficiency in the use of network resources – e.g., spectrum and equipment – but the timing and deployment options vary among them. Crucial to the
and reduce the cost of providing wireless connectivity. success of implementing Open RAN is the choice of a migration path
toward a fully virtualized and disaggregated architecture.
Our analysis shows how the network topology leverage different sources of Open
RAN pooling gains and how these, in combination with the transport costs, affect the Both greenfield and brownfield operators may initially choose to have a
TCO. In the first paper, we showed that an operator with high transport costs is more distributed Open RAN topology and later move gradually toward a
better off keeping the DU at the cell site, and one with low transport costs benefits more centralized one. In this case the DU is initially at the cell site, and
from locating the DU at a remote data center with the CUs. Here we go one step eventually it may get relocated to a remote data center with the CUs.
further and look at how the pooling gains differ for operators with high versus low
transport costs. A main reason to have DUs at the cell site is that this reduces the
transport costs. Remote DUs require a high-capacity fronthaul (FH) link
Pooling gains accrue because the hardware for the DU and CU can be shared by from each cell site. DUs located at the cell site need a lower-capacity
multiple RRUs. In a traditional, fully distributed RAN, RU, DU and CU processing is all midhaul (MH) link to connect to the CUs at a central location.
allocated to a single cell – or sector – thus eliminating the potential for pooling
gains. In an Open RAN environment, pooling gains come from different sources: An operator with high FH costs or limited fiber availability may initially
prefer to install the DUs at the cell site. When fiber becomes available or
▪ DU at the cell site – Hardware serving multiple RRUs is shared, and this leads to a cheaper, the operator can move DUs to a centralized location and enjoy
reduction in equipment-related capex and opex. One server at the cell site may the pooling gains from virtualized and shared DU servers. Many European
act as the DU for multiple RRUs. As the cell site’s capacity increases, the hardware operators – Telefonica is one example – are deeply committed to Open
costs may stay the same or increase at a slower rate than the capacity. This results RAN, and yet they are proceeding with caution in their migration to
in a lower per-bit cost as the cell site’s capacity grows. centralized DUs because a fiber connection may be unavailable or overly
▪ DU at a centralized data center – Hardware serving multiple cell sites is shared, expensive.
and the reduction in equipment-related capex and opex is higher than when the
DU is located at the cell site, because sharing is more efficient and larger servers At the opposite side, an operator that owns a fiber network that reaches
can be used – processing-related costs are lower for larger servers. the cell sites is likely to co-locate DUs and CUs in a remote data center
from the beginning, because this is the most cost-effective path. This is
▪ CU at a centralized data center – Hardware is shared, and the CU can serve the path chosen by Rakuten – the greenfield operator in Japan that
multiple cell sites. In our TCO model, CUs are always remote, so the pooling gains leverages a $30/month fiber connection cost and has access to its own
are the same across scenarios and cell types. fiber network.
To realize a net TCO benefit, however, pooling gains have to be balanced with the Many operators will find themselves in an in-between situation, and hence
transport costs. The pooling gains may not be sufficiently high to justify centralized they may at different paces to a virtualized and centralized topology.
DUs, which typically increase transport costs. If, on the other hand, transport costs Furthermore, the transition path and speed may change across different
are low, moving the DUs to a centralized location can increase the pooling gains areas within a network, depending on the transport assets the operator
enough to lower the overall TCO. owns or can lease in those areas.

©Senza Fili 2021 What’s the best way to get to Open RAN? |3|
 TCO model: Scenarios and assumptions

Scenarios Our model compares the TCO for two scenarios:

▪ Scenario 1 – Distributed topology: DUs are located at the cell sites with RUs, and
MH connects DUs to the CU.
▪ Scenario 2 – Centralized topology: CU and DUs are in the same location, and FH
connects RUs to the CU/DUs.

Scope Our results covers Open RAN scenarios that include CU, DU, MH and FH capex
and opex costs over six years, with all the capex incurred in the first year. Because the
RU-related costs are constant across scenarios, they do not affect the transport versus
location tradeoffs, and we do not include them in the results shown here.

Cell sites We compared four cell profiles:

1- 5G-NR, 20 MHz channels, frequency-division duplex (FDD) with 4T4R multiple
input, multiple output (MIMO).
2- 5G-NR, 100 MHz channels, time-division duplex (TDD) with 32T32R MIMO, 8
layers.
3- 5G-NR, 400 MHz channels, mmWave, 4 layers.
4- Combination of profile 1 and profile 3.

Each site has 3 cells (sectors) in the first three profiles, and 6 cells in the fourth.

Network 1,667 cell sites.

Transport Ethernet transport, with star packet links, using radio over Ethernet (RoE)
and supporting the Enhanced Common Public Radio Interface (eCPRI) 7.2x O-RAN
Open Fronthaul Interface over colored wavelength-division multiplexing (WDM).

Remote data centers DUs (scenario 2) and CU (scenarios 1 and 2) are in data centers
where hardware resources are shared across RUs, resulting in higher efficiency
because of pooling gains.

Inputs Cost, requirements, and traffic inputs are from trials and customers of
Mavenir, Intel and HFR Networks.

©Senza Fili 2021 What’s the best way to get to Open RAN? |4|
Low transport costs: Move the DUs with the CUs to remote data centers
For an operator that has easy and low-cost access to fiber for FH over the 7.2 split, co-locating the DUs with the CUs (scenario 1) is the most cost-effective solution,
delivering a cumulative TCO cost reduction ranging from 36% to 42% compared with the second scenario with DUs the cell site. The cost savings are highest for the
cell with the lowest capacity (cell profile 1, 20 MHz), because in this case the pooling-gains difference between the two scenarios is the largest: the DU server at the
cell site is underused and cannot be shared across cell sites. If the DU at the cell site supported more than the 3 cells that we assume here, the TCO for scenario 1
would look better, because DU resources would be shared more efficiently. However, cell profile 4 combines 6 cells, yet the TCO savings are similar to cell profile 1
with 3 cells. This is because this fourth profile includes 20 MHz and mmWave cells, and the TCO cost savings for mmWave are lower. As a result, the overall cost
saving for profile 4 is slightly lower than that for cell profile 1. As capacity increases, the costs savings for moving the DU to a centralized location decrease (i.e., from
42%, to 37% and 36%), because with the higher throughput, the pooling gains in scenario 2 remain largely unchanged, but the pooling gains in scenario 1 increase,
reducing the overall cost savings.

©Senza Fili 2021 What’s the best way to get to Open RAN? |5|
High transport costs: Keep the DUs at the cell site
Operators with high transport costs can get TCO savings ranging from 12% to 15% by deploying the DUs at the cell site. This lowers the transport costs and
operators benefit from pooling gains at the cell site. In the high-transport case, the pooling gains have less impact, because monthly transport lease costs
overshadow the CU and DU costs. The FH requirements in scenario 2 (DU and CU remote) add more in costs than the pooling gains save. As the cell capacity goes
up, the costs gap between scenario 1 and scenario 2 decreases. This is because transport costs as a percentage of the overall TCO decrease, as a result of the lower
per-Gbps cost of transport links of higher capacity.

Many of the initial Open RAN adopters are deploying DUs at the cell sites initially to benefit from these initial cost savings and other benefits, such as disaggregation,
avoidance of vendor lock-in and wider equipment choice. In a later phase of the transition path to Open RAN, we expect them to move the DUs to a centralized
location when transport prices go down or they have deployed its own fiber network.

©Senza Fili 2021 What’s the best way to get to Open RAN? |6|
 Per-bit comparison: Better TCO with more capacity

We have looked at the TCO for different scenarios and cases. But, as we noted, the
total TCO changes significantly for different cell types and, more importantly,
transport costs.

To compare the impact of these changes more directly, we looked at the per-bit
TCO, which shows the TCO as a function of throughput. To simplify the comparison,
we present the per-bit TCO for each case as a percentage of the highest per-bit TCO
case, which is for scenario 2 with high transport costs, for cell profile 1 (20 MHz).

Across the four groups (two scenarios with high and low transport costs), the per-bit
TCO is highest for cell profile 1, because the throughput is lowest. In these cases, the
deployment costs are lower, but the capacity difference is higher than the cost
difference. This high-level result would hold in traditional RAN deployments as well
and simply shows that deploying cells with higher capacity improves the cost-
effectiveness of the network.

The per-bit cost is consistently higher for operators with high transport costs. This is
to be expected, because the opex contribution is higher due to the FH and MH lease
costs.

The most interesting comparison is between scenario 1 and scenario 2, which shows
the impact of pooling gains that depend on DU locations – at the cell site or in a
remote data center. For the low-transport-cost operator, the per-bit TCO decreases
for scenario 2, as pooling gains from the remote DU come into play. For the high-
transport-costs operator, the per-bit TCO is lower for scenario 1, which is the most
cost effective.

Takeaways Operators with low transport costs can maximize Open RAN’s pooling gains by locating the DUs in remote data centers
with the CUs.

Operators with high transport costs are better off with the DU located at the cell site to lower the transport costs. In their
case, minimizing transport costs is more effective than maximizing pooling gains.

The cost efficiency of Open RAN deployments is higher as the capacity of the cells increases.

©Senza Fili 2021 What’s the best way to get to Open RAN? |7|
About Mavenir
Mavenir is building the future of networks and pioneering advanced technology, focusing on the vision of a single, software-based
automated network that runs on any cloud. As the industry’s only end-to-end, cloud-native network software provider, Mavenir is
transforming the way the world connects, accelerating software network transformation for 250.
Learn more at http://www.mavenir.com

About Intel
Intel is an industry leader, creating world-changing technology that enables global progress and enriches lives. Inspired by Moore’s
Law, we continuously work to advance the design and manufacturing of semiconductors to help address our customers’ greatest
challenges. By embedding intelligence in the cloud, network, edge and every kind of computing device, we unleash the potential of
data to transform business and society for the better. To learn more about Intel’s innovations, go to newsroom.intel.com and
intel.com

©Senza Fili 2021 What’s the best way to get to Open RAN? |8|
About Senza Fili
Senza Fili provides advisory support on wireless technologies and services. At Senza Fili we have in-depth expertise
in financial modeling, market forecasts and research, strategy, business plan support, and due diligence. Our client
base is international and spans the entire value chain: clients include wireline, fixed wireless, and mobile operators,
enterprises and other vertical players, vendors, system integrators, investors, regulators, and industry associations.
We provide a bridge between technologies and services, helping our clients assess established and emerging
technologies, use these technologies to support new or existing services, and build solid, profitable business models.
Independent advice, a strong quantitative orientation, and an international perspective are the hallmarks of our
work. For additional information, visit www.senzafili.com.

About Monica Paolini

Monica Paolini, PhD, founded Senza Fili in 2003. She is an expert in wireless technologies and has helped clients
worldwide to understand technology and customer requirements, evaluate business plan opportunities, market their
services and products, and estimate the market size and revenue opportunity of new and established wireless
technologies. She frequently gives presentations at conferences, and she has written many reports and articles on
wireless technologies and services. She has a PhD in cognitive science from the University of California, San Diego
(US), an MBA from the University of Oxford (UK), and a BA/MA in philosophy from the University of Bologna (Italy).

© 2021 Senza Fili. All rights reserved. We thank Mavenir and Intel for supporting the preparation of this paper. The views and statements expressed in this report
are those of Senza Fili, and they should not be inferred to reflect the position of Mavenir or other parties mentioned in this document. The document can be
distributed only in its integral form and acknowledging the source. No selection of this material may be copied, photocopied, or duplicated in any form or by any
means, or redistributed without express written permission from Senza Fili. While the document is based on information that we consider accurate and reliable,
Senza Fili makes no warranty, express or implied, as to the accuracy of the information in this document. Senza Fili assumes no liability for any damage or loss
arising from reliance on this information. Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Mavenir
and the M logo are trademarks owned by Mavenir Systems, Inc. Trademarks mentioned in this document are the property of their respective owners. Photos by
Shutterstock.

©Senza Fili 2021 What’s the best way to get to Open RAN? |9|