1st IEEE Energy Efficiency Tutorial:

5G Densification and Network Power Efficiency

Presented By –
Apurv Mathur, Principal Architect
Nokia Networks
[email protected]

Wednesday, September 19, 2018

Overview
• 5G Network Densification
• Small Cell Deployments and deployment challenges
• TCO & Energy Efficiency (Small Cell Focused)
• Techniques to improve energy efficiency

TCO – Total Cost of Ownership

2
ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
What is a Small Cell?
Lower Powered Radio Access Nodes that have the following characteristics:
➢ Small…
❖ Form Factor / Size: Typically ~3L for Indoor, ~10L for Outdoor Femto
❖ Coverage Area: Typical Range is between 10 meters & 2 Kilometers
➢ Adapted for different environments/deployments.. Pico
❖ Indoor: Residential, Enterprise,.. (Existing Infrastructure,..) Micro
❖ Outdoor: Street Level, Metro,.. (Wireless Backhaul Option,..)
❖ Public / Private Networks
➢ Power Optimized
Metro
❖ Size/Power: Design target to reduce power consumption for small form factor
DAS
❖ Indoor: Support for flexible deployments (PoE, etc.)
➢ Support of different RATs, Spectrum options..
❖ Licensed / Unlicensed Spectrum: CBRS, LAA, ..
LP-RRH
❖ RAT Capabilities: WiFi, 3G, 4G, 5G

Provide coverage & capacity solution for densely populated Urban Areas that cannot be sustained by Macro Cells
CBRS – Citizens Band Radio System; DAS – Distributed Antenna System; LAA – Licensed Assisted Access; LP-RRH – Low Power Remote Radio Head

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Small Cells – Key enablers for tomorrow’s applications

“Small cells are essential for the

Energy Efficiency
98%
operators future of our networks”* • Small Cells provide better energy efficiency
as compared to Macro layer

Outdoor Indoor Other Scenarios

• From Coverage to Capacity • Public indoor locations • SC open up new use cases
• From Hot Spot to Hot Zone • Enterprise deployments • Rural, Public Safety fast
• Help where Macros are difficult • Reliable IoT coverage layer response, Petroleum, Mining,
to deploy Stadiums, Protected buildings,
etc.

* - Informa Telecoms & Media Survey

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Small Cells Growth – Capacity & Coverage
New Non-Residential Small Cell deployments
predicted to grow at annual rate of 36%

New non-residential Small Cell Deployments forecast (2015-2025)

Source: Small Cells Forum [1]

Higher Cell/Site Density leading to higher energy demands

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Small Cells 5G Architecture Variants
Indoor Variants Outdoor Variants
To CU / NGC To CU / NGC

To CU / NGC To CU / NGC
Baseband Baseband
Unit Unit

5G Small Cell 5G Small Cell

CPRI/eCPRI
CPRI/eCPRI
Cell Cell
Aggregation
LP-RRH

• Distributed Baseband with LP-Radio … LP-Radio • Distributed Baseband with

Cell
Integrated RF Integrated RF
SFN or DAS
• Coverage or Capacity Uplift • Coverage or Capacity Uplift • Centralized baseband with
• Distributed Radios with distributed Radios
• Delay tolerant ethernet Centralized Baseband • Delay tolerant ethernet
fronthaul. Public/private fronthaul or Wireless backhaul • Coverage or Capacity Uplift
backhaul • In-Building coverage solution
• Street level/pole deployment - • Low latency fiber between
• Small Form Factor and PoE • Low-latency fiber between design optimized for low Baseband Unit and LP-RRH
capable – design optimized Baseband Unit and power/size
for low power aggregation point • Power distributed between
Baseband, fiber distribution
• End nodes power efficient, but and Radios
system not optimized for
power

CU – Central Unit; DU – Distributed Unit; DAS – Distributed Antenna System; LP-RRH – Low Power Remote Radio Head; NGC – Next Generation Core; SFN – Single Frequency Network
Small Cell Deployment Consideration->Magnified with 5G
Higher Bandwidth, Spectrum increases small cell density

Backhaul Backhaul/EPC
Macro Layer Connectivity Impact, Routing 34
78

Perf., Cost ,Complexity Security 65

Co-existence & Synchronization & Local Policy No fences nor cabinet..

Spectrum Engineering
Challenges
10k cells.... 10k cells....
EPC Impact Optimization Site Value Index
Signalling load
(Power, Backhaul, Access)
10k cells....

Performance Ease of
Macro Collaboration Deployment/Operation
Installation
• Form Factor / Stealth designs
Dense Urban Urban/Suburban Rural/
Hamlet

Quality of Services 10k cells.... • Site Value Index

Electrical Power Management
Transparent to User
Mobility • Advanced SON for No-Touch

Site Acquisition costs including access to power (24x7) play big role in Outdoor Small Cell Deployments

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Small Cells Deployment Examples

Harder to deploy and access

Small Cells in some deployments.
Varying power sources based on
region.

Small Cells on Street poles

Some deployments allow
often times co-located with
for easier access to
other auxiliary equipment (e.g.
backhaul/power, and
wireless backhaul, surveillance
better coverage.
camera, etc.) that also needs
power.

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Compounding factors will lead to Small Cells playing an early role in 5G
Difficulty in getting macro site acceptance
drives small cell needs
5G deployment challenges
Coverage Hot zones
densification
High frequency Location of available macro sites
bands leads to 5G coverage holes
Compact underlay Compact 5G ultra
of 5G macro broadband
Outdoor and Indoor use cases are 5G
dependent on continuous 5G small
coverage cells
Outside-in Indoor
Existing dedicated indoor solutions
Ultra broadband boost
(esp. DAS) are not extendable to
is mainly indoors 5G
Street-level 5G to Dedicated 5G
indoor
penetrate indoors densification

Energy Efficiency needs to be addressed across the diverse Small Cell 5G Deployments (Outdoor/Indoor,
cmWave/mmWave)
ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Other 5G Deployment considerations
• 5G Distributed gNB (CU-DU) Deployments L3

▪ Functional split between CU & DU L2 HI

5G CU
▪ Drive towards open interfaces (ORAN/xRAN)
▪ CU hosted on Cloud / Virtualized HW F1

▪ Energy Efficiency needs to be considered from E2E

perspective based on deployment 5G DU

L2 LO
L1
• Massive MIMO support RF

▪ Higher antenna count for mmWave with analog beamforming

▪ Higher TRXs for cmWave with digital beamforming
▪ Massive MIMO systems increase the BTS energy demands
due to higher RF power consumption

CU – Central Unit
DU – Distributed Unit
MIMO – Multiple-Input Multiple-Output
ORAN – Open RAN
TRX - Transceiver

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Networks OpEx and Energy Consumption
RAN Network data / non-Small Cells / Pre-5G

Energy cost can represent up to

60+% of 5y TCO*
• Energy cost typically higher than
Product cost

Small Cells TCO Factors:

• Site acquisition costs
• OpEx (transport/power
consumption)
• HW/SW costs
• Cell Density

TCO – Total Cost of Ownership * Macro Single RAN Deployments (2G/3G/4G)[2]

OPEX – Operational Expenses

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Small Cells TCO and 5G considerations

5G influence on Small Cells TCO (as compared to LTE)

• Site acquisition costs
Addressing energy costs for 5G Small Cells is key to
• OpEx (transport/power consumption) reducing TCO and increasing energy efficiency.

• HW/SW costs Newer devices, active antenna arrays. Lower costs over time
• Cell Density
e.g. >>10x number of mmWave Small Cells as compared to LTE Small Cells

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Power Consumption & Network loading
• Daily average traffic over 24h modelled
through 3 traffic loads (per ETSI ES 202 706-1)

• Each load is defined to emulate an “RF

load” as below (% of commissioned RF
max power)
ETSI Load models for LTE 20MHz
100% load E-TM1.1 TS 136 141
ETSI Busy Load 52,0%
ETSI Medium Load 33,0%
ETSI Low Load 5,2%
ETSI Avg Load 32,4%

Radio network loading varies across sites and also during different times of the day

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Power Savings Opportunities and breakdown
Load based Power Scaling - Example 5G Small Cells Power Breakdown*
Traffic Load 100 50 30 0 24h Average
Power saved 0 100 140 200 141.7
[W] (ideal*)
Hours/day 0 8 10 6 -

OPEX saved = (141.7/1000) kw x 5 years x (6.63c per kwh) x 50k nodes

= ~20 million US Dollars

• Example assumes 200W power consumption per node,

50k nodes in deployment (outdoor). It suggests that if
power usage can be scaled on load basis, it could result Observations:
in significant savings.
• High Digital power contribution due to lack of integrated SoCs
• In practice, not all system components scale with load ➢ Opportunity to improve static power consumption
and not all components have dynamic power scaling
capabilities. Idea is to maximize energy efficiency across • PA Efficiency much lower for initial mmWave Active Antenna systems.
the system. Much higher power consumption due to high Active antenna array
counts
• Reduction in both static and dynamic power ➢ Opportunity to improve dynamic power consumption
consumption is key!

EE Goal: Lower energy consumption with increased user traffic (i.e. maximize bits/Joule)
* Based on trends in current chipsets / silicon / RF Technologies (100% loading)

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
5G Small Cells EE Areas of Focus

Hardware Software Energy

Efficiency Functionality Management
Technology Evolution Adapt Power consumption RAN Level Intelligent Energy
to user traffic Management (e.g. SON)
• RF Design with optimized • Load based power
Power Amplifier savings (HetNet) • KPI based dynamic switching
Technology of network resources (units,
• Micro DTX, Tuning Sync cells, etc.)
• Power efficient and burst periodicity for 5G
integrated System-on-Chip Small Cells, Muting, ML/AI • Monitoring / Reporting of
and Memory devices leverage,.. Energy Status & Data

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Hardware Energy Efficiency
• Digital baseband hardware (L1-L3)
▪ 5G L1 in ASIC / SOC enabler for low footprint
Digital design
▪ Upper Layer (L2+) processing needs vary based
on Classical versus DU mode of operation
▪ Significant improvement in SoC technology
(14nm->10nm->7nm) and dynamic power scaling
allowing for energy efficient digital designs

Technology and Manufacturing day Intel[2]

• RF hardware (integrated RF in Small Cells)
▪ Integrated DFE/TRX Processing efficiencies Beamforming gains
▪ Analog RF Beamforming helps address RF power efficiency as it provides higher
EIRP (beamforming gain) than higher power PAs
o 5G NR systems benefit from advances in PA
o Example
efficiency for wider bandwidths. Next
generation devices with Envelope Tracking and o Traditional:
advanced GaN expected to push PA efficiency o 4x10W with 7dBi Directional Antenna = 53dBm EIRP
beyond 50%. o 40W/40% PA Eff. = 100W
o mmWave RFICs/PAs expected to improve in o cmWave Solution:
terms of RF efficiency over time. o 8x3W with 64 Element Array = 63 dBm EIRP
o 24W/40% PA Eff = 60W

ASIC – Application Specific Integrated Circuit; DFE – Digital Front-End; PA – Power Amplifier; SOC – System-On-Chip

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Software Energy Savings Techniques
Load Based Power Savings (HetNet)

• Applicable to HetNet scenario where Small Cells are deployed for capacity uplift with Macro
Overlay.
• Small Cell Shutdown at operator adjustable condition (time of day, traffic threshold, macro
coverage).
• Assuming 6 hours of no activity on small cells, up to 20% power savings (less for 5G).

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Software Energy Savings Techniques
Discontinuous Transmission (DTX) for Energy Efficiency for 5G

• Power saving by switching off the cell's power • In LTE reference symbols need to be transmitted
amplifier(s) during idle period 4 times every 1 ms, limiting sleep mode duration.
• DTX controls the PA supply voltage simply as • 5G bring significant improvement in power
consumption with flexible reference signal
• No data: no supply voltage to PA
design.
• Data: supply voltage to PA
• Network level power savings of ~50% for 10-
20% network utilization!

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Software Energy Savings Techniques
Other 5G EE Techniques
Sync burst periodicity
• 3GPP standards support different Synchronization (SS) Burst periodicity to decrease
power consumption. SS Block periodicity can be varied between 5 and 160ms.
• Increased SS Burst periodicity reduces power consumption in low load conditions.
• Higher periodicity may result in UE synchronization loss, especially for mmWave and high
speed UEs.

Muting / RF Resources
• mMIMO muting at lower loads. Turning off/on TRXs on load.
• Disabling mmWave Antenna sub-panels/elements at low loads (reduced spectral
efficiency/throughput).

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Intelligent RAN Energy Management
KPI based dynamic switching of network resources
Managing energy consumption at network level
• Capability to temporarily shut down gNBs during low traffic periods.
• Controlled network resource shutdown and startup based on preset schedule

Energy Saving Metering

• BTS Energy Metering and reporting the consumption via dedicated PM counters (Voltage and Power) to
Network Management system

SON for Multi-RAT/Multi-layer EE Management

• Improving energy efficiency of the multi-RAT system based on loading of the cells at network level
• Support heterogeneous network comprising of Macro overlay and Small Cells underlay

Predictive SON based on EE counters and AI/Machine Learning - FFS

Leveraging Cloud RAN/OpenRAN for better Network energy efficiency - FFS

KPI – Key Performance Indicators; RAT – Radio Access Technology (2G/3G/4G/5G); SON – Self-Organizing Network

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
5G Small Cells / EE Impact Summary

• 5G more energy efficient due to wider bandwidth, better air interface design, more advanced antennas,
base station sleep modes and applying other EE techniques
• 5G Small Cells outperform 5G macro cells due to lower output power, more efficient design and reduced
number of users.

Based on ~60% traffic growth and 5G launch in 2020

ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Conclusions
• Demand for higher capacity and better coverage leading to proliferation of
Small Cells deployments. Energy costs remain a big component of the
Networks OpEx.

• Radio network loading varies across sites and also during different times of
the day. 5G air interface design lends itself to better tuning of power
consumption for different loads and deployments.

• Several load based EE techniques can be adopted in 5G networks to reduce

power consumption at component, box, network level.

• 5G Small Cells promise to provide improved energy efficiency while meeting

the demands of next generation networks.

22
ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
Q&A

Thanks a lot for your time and attention!

Any questions and/or comments?

23
ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.
References
1. Small Cells Forum - https://scf.io/en/documents/194_-
_Deployment_plans_and_business_drivers_for_a_dense_HetNet_SCF_operator_survey.php
2. Nokia RAN Energy Efficiency – link
3. Moore’s law leadership – Intel – link

24
ALL INFORMATION SHALL BE CONSIDERED SPEAKER PROPERTY UNLESS OTHERWISE SUPERSEDED BY ANOTHER DOCUMENT.