IEEE ICC’2015 London Panel on Mobile Fronthaul
Mobile Fronthaul Era- Are We There Yet?
Chair: Anthony Magee, 11th June 2015
© 2015 ADVA Optical Networking. All rights reserved. Confidential.2
Agenda
• Introduction of the Panel and Setting the Scene
• ADVAs position on Mobile Fronthaul • Readiness today, research projects & challenges…
• Round Table – Panel member interests in Mobile Generally
• Questions and discussion section
• Summary & Wrap Up
© 2015 ADVA Optical Networking. All rights reserved. Confidential.3
Our Panel
• Chair: Anthony Magee, ADVA Optical Networking• Business Development Director
• Dr Chih-Lin I, China Mobile Research Institute,• Chief Scientist, Wireless Technologies
• Professor Andy Sutton, EE• Principal Network Architect
• Dr Nick Edwards, Openreach, part of BT Group• Wireless Strategy Manager
• Dr Nathan Gomes, University of Kent• Coordinating role for iCIRRUS, Technical Program Chair for IEEE ICC 2015
• Dr Volker Jungnickel, HHI• Head of the metro, access, and in-house systems group
© 2015 ADVA Optical Networking. All rights reserved. Confidential.4
ADVA Solutions Overview
Automated service delivery and assurance from access to core
FSP Service Manager
Private Enterprise Networks
Data Center Connectivity
Carrier Infrastructure
Metro Core
FSP 3000
Broadband BackhaulMetro NetworksLong Haul
Carrier Ethernet Access
FSP 150
Mobile BackhaulBusiness EthernetEthernet Wholesale
© 2015 ADVA Optical Networking. All rights reserved. Confidential.5
Mobile Backhaul Architecture
CE Access IP/MPLSMNO Core
BST CSGW NID Aggregation
• Mobile Backhaul• Carrier Ethernet
• Demarcation and Aggregation
• Timing Distribution
FSP 150 Family
© 2015 ADVA Optical Networking. All rights reserved. Confidential.6
Mobile Fronthaul with Passive Monitoring
Non-intrusive fiber monitoring and optical service assurance
RRH
RRH
RRH
RRHRRHRRH
Small Cell Small Cell Small CellMacro Cell
FSP 3000ALM
FSPService Manager
• CWDM and DWDM filters with an upgrade port are ideally suited for the realization of linear add-drop structures
• Complimentary use of in-path OTDR, the complete link is monitored
• The add-drop filters function as demarcation points
Reflecta
nce
Nx Distance2x1x
BBU Pool
© 2015 ADVA Optical Networking. All rights reserved. Confidential.7
So Why a Panel on Fronthual?
• Mobile Network Evolution• CoMP, LTE-A, Small Cells, 5G• Leading many to look at and evaluate Fronthaul as an enabler
• Serious Concerns Arising from fronthaul• SLA Management• Latency• What architectures will/won’t work?• CPRI Capacity with 5G on the Horizon, what are the alternatives?• Mobile operators, carriers need to be convinced of technology and business case
• Answers need wide consultation and collaboration• ADVA collaborating via
• EU FP7 COMBO – Fixed Mobile Convergence• EU Horizon 2020 iCIRRUS – Ethernet as a basis for C-RAN transport• Interaction with standards bodies
• MEF, ITU-T, FSAN, IEEE 1904.3, IEEE 802.1 TSN
Industry Panel – a good way to gain insight and identify consensus
© 2015 ADVA Optical Networking. All rights reserved. Confidential.8
Introductions and Insights from the Panel
Andy SuttonPrincipal Network ArchitectEE Network Strategy11th June 2015
Mobile Fronthaul Era – Are We There Yet? – panel session:
A Mobile Network Operators Perspective
Contents
10
• Long-term UK market forecasts
• Current network architecture
• Local C-RAN
• Migration scenarios
• Summary
Long-term UK market forecasts
11
-
500
1,000
1,500
2,000
2,500
Dec-15 Dec-20 Dec-25 Dec-30Mo
nth
ly D
ata
Dem
an
d (
PB
)
E-mail/IM/Browsing/Download/MusicAugmented RealityMobile Gaming
0
2
4
6
8
10
12
14
16
18
20
0%
20%
40%
60%
80%
100%
2015 2020 2025 2030
Vid
eo
Data
rate
(M
bp
s)
Vid
eo
reso
luti
on
p
en
etr
ati
on
8K 4K HD SD Video Bitrate
-
200
400
600
800
Dec-15 Dec-20 Dec-25 Dec-30
Mo
nth
ly D
ata
D
em
an
d (
PB
)
Non-Video Data Demand
E-mail/IM/Browsing/Download/Music etc.Augmented RealityMobile Gaming Other services will emerge and drive
new revenue opportunities and
capacity challenges
Streaming and video resolution are key
drivers of long-term growth
22x
68% non-video
demand
2030 76%
DEMAND VIDEO
Current EE Network Architecture
12
Shared mobile backhaul network
Local C-RAN
13
M s
s
s
FronthaulTraditional backhaul M Macro cell s Small cell
Backhaul supports multi-RAT
macro and LTE small cells –
capacity, performance and
sync distribution
considerations
Operators
core
Baseband
capability
supports
macro and
small cells
Wired or
wireless
fronthaul
Local coordination between
macro and co-channel
small cells, support for
features such as dual-
attachment (HetNet CA?)
Combined backhaul and fronthaul requirementsExample use case – 3G MORAN, LTE site and backhaul share
14
FronthaulTraditional backhaul
GSM
MORAN
LTE#2
C
S
G
T
x
C
S
GT
x
All backhaul
Single op LTE C-RAN
All fronthaul*1 - Standalone or 2 x Multi-RAT base stations
LTE#1
GSM
MORAN
LTE#2
LTE#1
C
S
GT
x
Dual op LTE C-RAN GSM
MORAN
LTE#2
LTE#1
GSM
MORAN
LTE#2
T
x
All fronthaul*1
LTE#1
Examples only, not an exhaustive list of scenarios…
Summary
15
• Local C-RAN based on macro and its underlying small cells is an interesting architecture which seems to offer RAN capacity and/or performance benefits - remote sectors…
• The likelihood of a multi-RAT cell site having legacy services on backhaul while migrating LTE to C-RAN is probable and should therefore be supported in equipment
• There is a wide range of site and infrastructure sharing models which should be considered within the context of any network evolution; be this D-RAN, C-RAN or anything in between (in terms of functional split) - will today’s C-RAN scale for 5G?
• Different distributions of functionality will exist in different areas of a national network
• Cost of fronthaul is a significant barrier to adoption
THANK YOU
Mobile Fronthaul Era –
Are We There Yet?
Nick Edwards
Wireless Strategy Manager, Openreach
Introduction to Openreach
• Openreach provides the UK
communications access network (the
“last mile”)
– 30 million residential customers
– served from >5500 exchanges
• via more than 500 communications
providers
• Since 2009, Openreach has been
building a new superfast broadband
coverage
– A layer 2 Ethernet product
– VDSL at up to 80Mbps
– Millions of km of new fibre
– A footprint of more than 22m UK homes
• Openreach also provides fibre Ethernet
circuits (typically 1Gbps symmetric)
– Used by businesses, education,
healthcare, etc
– The majority of mobile base stations
are connected via Openreach access Source: Ofcom infrastructure report 2014
Mobile backhaul
• Mobile backhaul is one of the most significant areas of business for Openreach Ethernet products. Recent developments include:
– Reduction of prices (approximately 50% reduction since 2013)
• A 1Gbps circuit today is a similar cost to a 100Mbps circuit in 2013
– Introduction of synchronised Ethernet for mobile backhaul
– A new VDSL street access product providing up to 80Mbps to a street location
– A new mobile infill product enabling a mobile operator to use a telegraph pole as a small cell site
• Telegraph poles provide a very large single estate of street sites mainly in suburban and rural areas
• Mobile operator requirements for cell site connectivity are evolving
– Ongoing roll-out of LTE macrocells – leveraging our ever growing fibre footprint
– LTE small cells – need low cost connectivity
– Evolution towards LTE Advanced/LTE TDD and in the longer term support of 5G
– Growing interest in options for mobile fronthaul
OPENREACHHANDOVER
POINT
EXISTING PCP
CopperMDF Exchange
Fibre
VDSL CABINET
• 1U space for CP backhaul
equipment
• 240V metered power
8U space for MNO
provided radio equipment
Copperpair
Openreach provided street cabinet stand
alone from existing PCP/NGA cabinet
Openreach provided antenna
RADIO CABINET
Co-axial cable
Fronthaul vs Backhaul
• Fronthaul can offer many benefits to mobile operators and end customers:
• Reduces site acquisition, installation, rental and maintenance costs
• Equipment footprint at the base station is reduced, simplifying plan and build
• Lowers energy consumption for electronics and air conditioning at the cell site
• Pooled baseband can be virtualised, reducing cost and simplifying signalling
• Provide best support for LTE Advanced enhancements to improve cell edge coverage and data rate requiring low latency, high bandwidth, tight synchronisation, e.g.
• Transmission from multiple base stations to one device (CoMP) and interference co-ordination
• However:
• LTE-FDD 2x20MHz has 150Mbps peak rate, but CPRI needs 2.5Gbps/sector
• CPRI is independent of loading so does not benefit from statistical multiplexing
• CPRI bandwidth scales linearly with number of sectors, antennas, channel bandwidth
• Estimates of backhaul requirement for a 5G site may be in excess of 500Gbps
• Each sector needs separate fibre or wavelength
• Daisy chain possible but introduces new point of failure
• Very tight timing requirements make transport of CPRI over Ethernet impractical:
• Latency <100μs round trip, +/- 30ns imbalance
How could Fronthaul be realised?
CPRI fronthaul for macrocells
Would require solutions for efficient use of fibre, e.g. Wavelength Division Multiplexing (WDM)
Openreach is working closely with mobile operators
– Recent lab demonstration of CPRI fronthaul
Baseband
Radio head
Router
Remote
site
4
G
4
G4
G4
G3
G
3
G3
G3
G
Example: WDM with a 4-way split
3 sectors of 4G fronthaul
Ethernet for management+3G/2G
WDM
CPRI fronthaul for small cells
Economics are still very challenging even for conventional Ethernet backhauled small cells
– Fronthauled small cells are even more challenging
New base station architectures providing efficient fronthaul/midhaul would be essential to support 5G bandwidths
– Fronthaul over Ethernet would enable existing high volume Ethernet connectivity to be used
– Maximises benefits of virtualisation of the baseband functions
iCIRRUS- concept
Key: intelligence placed
in fronthaul of C-RAN
http://www.icirrus-5gnet.eu/ http://www.intelligent-nirvana.net/
Added intelligence for
D2D and hetnet
operation
Centralised functions
available to mobile
cloud processing
D2D, mm-wave, mobile
cloud important for 5G
What sort of Ethernet links?
• Ethernet over CPRI? Only real advantage is provision of OAM
• CPRI over Ethernet? Possible
Loss of synchronism through Ethernet switches
Additional framing overhead
• Sample waveform placed directly in Ethernet
frames New standards required
Loss of synchronism
• Bit-rate/bandwidth requirements
BBU pool to RRH connection through switches
Low-latency switching required: Cut-through operation
No contention/queuing
Bit-rate requirements: current and projected
Current CPRI/ORI interfaces Projected requirements
Line rate Example Use Possible uses Approx. line rate*
614.4 Mb/s 10 MHz LTE
channel, with
8B10B encoding
100 MHz, 8
antennas
(sectors/MIMO/C
oMP)
28 Gb/s
4.9152 Gb/s 8 x 10MHz
(multiple
antennas,
8B10B)
500 MHz, 8
antennas
(sectors/MIMO/C
oMP)
141 Gb/s
10.1376 Gb/s 10 x 20 MHz
(multiple
antennas,
64B66B)
500 MHz, 16x8
massive MIMO
2.25 Tb/s
Fronthaul/”fronthaul lite” division
Bandwidth
requirements
may be reduced,
but latency may
become a more
significant
problem
“Fronthaul lite” advantages and challenges
View, Master, Slide Master to change this text to the title of your presentationPage 27
Apart from bit-rate reduction…
Single streams instead of multiple streams for each antenna possible –
weights sent to antenna units/RRHs
Burst-like user data really enables statistical multiplexing gains
CSI sharing for multi-antenna (multi-RRH) techniques needs to take into
account delays
Tight synchronisation will still be a problem, may be even more so
Analogue radio over fibre
Simplest antenna
unit.
Lowest latency in
fibre links?
Lowest energy
consumption?
Performance degradation for high-bandwidth, high-frequency links
- research needed to obtain required/improve/enhance performance
Component costs if not a ubiquitous technology
- integration, photonic integration
Acknowledgments to current projects
Use of Ethernet in the fronthaul
• Use of commodity equipment, or at least lower-cost, industry-standard
equipment
• Sharing of equipment with fixed access networks, enabling greater
convergence and cost reductions
• Ethernet OAM functions standardised
• Use of switches/routers to enable statistical multiplexing gains and lower
the aggregate bit-rate requirements of some links
• Use of standard IP/Ethernet network switching/routing functionality,
including moves to functional virtualisation and overall network
orchestration
• Monitoring through compatible hardware probes.
Fibre transport/C-RANs: bandwidth problem
• Digitised transport preferred:
• 100 MHz bandwidth, 12 bits per sample for OFDM (?), > 2.4 Gb/s per signal
• MIMO/virtual MIMO – multiple signals N x 2.5 Gb/s?
• 1 GHz + bandwidth wireless signals (5G) – 100 Gb/s links to remote radio
heads?
• Compression possible – but only 2 (lossless) or 3 times reduction (China
Mobile)
• Baseband transport
• Order of magnitude reduction in bit-rates
• Co-operation/coordinated transmission?
• Analogue transport (radio over fibre)
• Performance?
“Fronthaul lite” as part of X-haul
Courtesy of Philippe Chanclou, Orange
Labs, OFC 2015, panel presentation
NEW FUNCTIONAL SPLIT
Volker Jungnickel, Luz Fernandez del Rosal
Fraunhofer Heinrich Hertz Institute, Berlin, Germany
Future RAN implementation
• Centralized Radio Access Network (C-RAN)
Fully centralized baseband unit (BBU)
All transport is fronthaul
For operators with own fiber infrastructure
• Distributed Radio Access Network (D-RAN)
Idea is to minimize transport overhead and latency
Data and control are shared among BBUs
Fully distributed processing
For operators with leased infrastructure
• The solution might be between these two extremes
iCirrus proposes Ethernet-based „Fronthaul Lite“
Minimize transport, maximize flexibility
Using new „functional split“ discussed in recent literature
Flexible functional split?References: 35-
38
Figure from P. Rost et al. “Cloud technologies for flexible 5G
radio access networks”, IEEE Communicatinos Magazine,
Vol. 52, No. 5, pp. 68-76
• Old functional split
Digital radio-over-fiber between PHY
and RF
CPRI/ORI: Sampled waveform transport
Huge data rates: No compression, no
statistical multiplexing
• New functional split
Find the right balance between C- and
D-RAN
Compress fronthaul signals more
efficiently
Enable statistical multiplexing
Minimize the added latency
Proposed fixed split is between PHY
and MAC
More intelligence into the remote radio
head
Literature review on
compression (I)Old functional split
New functional
split
Point-to-point (P2P)
DistributedTime-domain Frequency-domain
Fixed Dynamic Fixed Dynamic
Charact
.
Fixed BW
(bandwidth)
reduction
Several
methods
Transmission of
used BW
Adaptive multi-
rate filter
CP & guard
band removal
& several
methods
Transmission of
used BW
Joint processing
of signals
from/to multiple
RRHs
Time-domain
Several
methods
New functional split
between BBU and
RRH
Pros
Statistical
multiplexing
Lossless?
Statistical
multiplexing
Optimal
performance in
network
topologies
Lossless
Statistical
multiplexing
Spatial
compression
Waveform
independent
Cons
Lossy
No statistical
multiplexing
Lossy Lossy
DSP
Redundancy
No statistical
multiplexing
DSP
Redundancy
Lossy
DSP Complexity
No statistical
multiplexing
New architecture
and challenges
Refs 1-10 11-15 16-19 - 20-31 32-41
Compression Methods Overview (II)
Old funtional splitting
New
functional
splitting
Point-to-point (P2P)
DistributedTime-domain Frequency-domain
Fixed Dynamic FixedDynami
c
Compressio
n
ratio (ref)
28.57% (4)
28.08% (9)
Average
37.5%(13)
4% @ QPSK(16)
15% @
64QAM(16)
- Not available 3% @ 10% load(32)
16.6% @ 50%
load(32)
33.3% @ 100%
load(32)
EVM (ref) 2% (4)
1.89% (9)
Average <
1%(13)
< 0.025%(16) - Not available Lossless(32)
Validation (ref)
Real-time (1,2,5,9)
implementation
Real-time(13,15)
implementation
Simulation - Simulation Simulation
Ref 1-10 11-15 16-19 - 20-31 32-41• Proposed new split enables lossless compression w/o added
latency
• Compression ratio depends on served traffic: Statistical Multiplexing
• Independent on the waveform, i.e. applicable to TDMA, CDMA,
OFDMA, 5G
New Functional Split (SISO)
Base station is decomposed into most important blocks
• MAC processor (scheduler) and PHY processor
New functional split lies between PHY and MAC (orange line)
• Downlink: Hard bits for transport blocks plus ressource-map information
• Uplink: Hard bits plus soft bits for transport block plus user ID
Same incl. (massive)
MIMO
The beamformer for (massive) MIMO is shifted into the RRH as well Limit the transport to the number of streams actually used, not antennas
Same incl. MIMO and
CoMP
CoMP and MIMO make the picture complete Downlink data are exchanged over X2++ with adjacent cells (same as in EASY-C
implementation)
Uplink FFT outputs may be exchanged over X2++ with adjacent cells
Small cells are implemented same as macro-cells
Conclusions
• New functional split shifts the fronthaul from below-PHY to below-MAC
• All PHY functions are at the remote radio head
• All MAC functions are in the central office
• Best compression ratio is achieved depending on the traffic load
• Full statistical multiplexing gain: Only scheduled data are transported
• Basic ideas were already discussed in the literature
• Implementation is the next challenge
• The basic principle is independent on the waveform
• Ethernet-based transport is suggested for flexible deployment
• Need for efficient synchronization solutions, e.g. using 1588 PTP and
SynchE
Thank You
IMPORTANT NOTICE
The content of this presentation is strictly confidential. ADVA Optical Networking is the exclusive owner or licensee of the content, material, and information in this presentation. Any reproduction, publication or reprint, in whole or in part, is strictly prohibited.
The information in this presentation may not be accurate, complete or up to date, and is provided without warranties or representations of any kind, either express or implied. ADVA Optical Networking shall not be responsible for and disclaims any liability for any loss or damages, including without limitation, direct, indirect, incidental, consequential and special damages, alleged to have been caused by or in connection with using and/or relying on the information contained in this presentation.
Copyright © for the entire content of this presentation: ADVA Optical Networking.
Some aspects of the work leading to this panel have received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 644526 (iCIRRUS) and were supported by the European Commission under the Seventh Framework Programme (FP7) by the project COMBO (under grant agreement n° 317762)