Upload
others
View
14
Download
1
Embed Size (px)
Citation preview
5G C-RAN Architectures: A Comparison of Multiple Optical FronthaulNetworksChathurika Ranaweera 1, Elaine Wong 1, Thas A Nirmalathas 1,2, Chamil Jayasundara 1, and Christina Lim 1
1Department of Electrical and Electronic Engineering, The University of Melbourne, Australia2 Melbourne Networked Society Institute, The University of Melbourne, Australia
@nirmalathas
Why 5G?• The development of the fifth generation (5G) wireless
technology is in progress to address– The increasing demands for high capacity,– Low latency, – Ubiquitous mobile access
• 5G expedited to attain – 1000x higher data volume per unit area– 100x higher connecting devices, – 10x longer battery life and – 5x reduced latency
• Demands will be instigated by next-generation mobile and machine-centric applications.
2
Central Office
Macro cell
Key Feature of 5G :Small cells
Small Cells
Capacity Coverage
small cells create an enormous weight in the transport network as it
needs to carry a significant amount of data with a
minimal delay from thousands of cells.
3
Why C-RAN ? • Observation: the evolution of radio access network needs to be
complimented by an evolution of the transport network. • One architectural evolutionary solution: Centralised/Cloud Radio
Access Network (C-RAN) architecture.– significantly lower cost– greener communication– supporting advanced
wireless technologiesEg. Cooperative Multi-Point
(CoMP)
Base-BandUnit(BBU)RemoteRadioHead(RRH)
4
Challenges in C-RAN• In the current C-RAN
architecture (use in LTE-A): the fronthaul network uses common public radio interface (CPRI) [13] over fiber links.
• Ex: supports 150 Mbps of downlink bandwidth in LTE-A, more than 2 Gbps of optical bandwidth When using CPRI.
• If 5G fronthaul networks use CPRI à
5
Architecture Comparison• Three Fronthaul Network Architectures
– CPRI– Physical Layer Split (PLS)– Analogue Radio over Fibre (ARoF)
CPRIandPLS
ARoF
6
Bandwidth Calculations - CPRI
7
Ns isthenumberofsectorsperRRHNa isthenumberofantennasSf isthesamplingfrequency
Sbw isthesamplingbit-width(I/Q)Be istheratioforthecontrollingoverheadLc isthecodinginducedcapacityincrease
Bandwidth Calculations -PLS
8
T T I the transmission time interval (=1mS)M modulation order (=8 i.e. 64QAM)Nsy number of symbol within a TTI (=12) Nsc number of subcarriers (=12)Nrb number of resource blocks (=500)Nmimo number of MIMO streams (=8)
5ResourceBlocksperusersomaximumof100usersconsidered
Comparison of Bandwidth Requirements
• ARoF: required fronthaul bandwidth depends on the wireless carrier frequency and the bandwidth in use, a typical low-cost transceiver can be used to achieve the 5G targeted data rates when 5G uses the frequency range below 6GHz (IF conversion can be used with very low cost component for higher frequencies).
9
Comparison: Delay, Advanced Wireless Functions and RRH complexityDelay AdvancedWirelessFunctions Complexityof RRH
CPRI Afewhundredsofμsincludingthepropagationdelay,roundtripCPRIprocessingdelay,andtheotherfronthaulequipment
processingdelays.
Facilitateadvancewirelesscooperationtechnologies
Simple.However, forhighertransmissiondatarates,Fronthaul
needshighdatarateopticaltransceivers.
PLS AnadditionalprocessingdelayattheRRHcomparedtotheCPRI: symbollevel
processingimplementedinRRH(however,additionaldelayislessthanfewμs).
WirelesscodingfunctionsandMAClayerfunctionsarecentralizedinthe
BBU:canfacilitateadvancewirelesscooperationtechnologies
RRHrelatedequipmentandsoftwareneedtobeupgraded.Networkfunctionvirtualization(NVF)
paradigmcanbeusedtoovercomethedifficultyinupgrading
ARoF RRHwillbemoredelayefficientcomparedtotheCPRI.
Fronthaul linkrange willbe lowcomparedtootherarchitectures(howeverstillfacilitate
fewkmthatconfirmwith5GBBU-RRHrange).
Facilitateadvancewirelesscooperationtechnologies
allthewirelesssignalsareprocessedcentrallyatthe
BBU
Simplearchitecture
10
Optimization framework finds the cost-optimal deployment of
Fibre Fronthaul + RRHs + BBUs
Deployment Cost comparison
Macro cell
Central Office
BBUs
• Optimal Deployment Cost : ILP-based optimization framework
• The total cost consists of the cost of BBU placement, fronthaul and Deploying RRHs.
11
Optimization FrameworkMinimizethetotaldeploymentcostofCRANdeployment
Subject to a range of constraints imposedbyrequirementsofthenetwork such as,• Population coverage• The maximum number of BBUs in one
central office • The maximum distance from a RRH to its
BBU• Fibre connectivity
min 𝜂& + 𝜂() *𝑥)�
)-.
+ (𝜂0+𝜂12)**𝑐),6�
6-.
�
)-.
𝑑),6 + 𝜂1&**𝑦),6�
6-.
�
)-.
𝑑),6 + 𝜂9*𝑥)�
)-.
+ (𝜂2+𝜂2))*𝑍)�
)-.
Equipment&installationcostsofRRHsandFronthaul
Newfibreroutes,andfibrebundles
installationcosts
Fibrepreparation
Fibreconnectionsatthecentralofficeforusingexistingfibrefacility.
CostofBBUsandtheirinstallation
12
Costparameter DescriptionȠs Equipment costofRRHandfronthaulȠri CostofRRHInstallationȠt Fibretrenchingcost/mȠfb Cost ofaFibrebundle/mȠfs CostoffibrepreparationȠe Costofconnectingafibreatexistingfibrefacility
Ƞb Cost ofaBBUȠbi InstallationcostoftheBBUatthefibrefacility
Data Set for Cost Comparison
• 18km2suburbanareainstateofVictoria,Australia
• Population:30,000
13
Cost – Summary of Optimal Solutions
Contributionoffronthaul,RRH,andBBUcostwhenthepopulationcoverageis50%and90%
• ThecostvaluesarenormalizedWRTthetotaldeploymentcostofCPRI-basedC-RANunder90%populationcoveragerequirement.
• Conservativeapproach:CPRIusesa1/2compressiontechniquewithoutanyadditionalcost:CPRIrequire40Gbps transceiversinsteadof100Gbps
• PLSandARoF architecturesuses10Gbps transceivers
NormalizedCostValues*
CPRI:Fronthaul 2912
PLS: Fronthaul 200
ARoF:Fronthaul 200
CPRI:RRH 220
PLS:RRH 300
ARoF:RRH 200
BBU 10000
Fibrebundlecost 1
*Normilized tocostoffiberbundles/m
14
Summary• We investigated the applicability of different optical fronthaul technologies for
CRAN– CPRI, PLS and ARoF
• Comparatively analysed their ability to fulfil requirements of delay, bandwidth,and cost-effectiveness of 5G CRAN, ability to support advanced wirelesscooperation technologies, and complexity of RRH.
• Comparative cost analyses carried out using developed optimization frameworkshowed that cost-effective fronthaul for 5G could be achieved using PLS andARoF architectures
• Overall, our analyses provide insight into how a future proof fronthaul networkcan be realized for 5G C-RAN.
15
Variables
17
Variable Values DescriptionXi 1 Ifith locationisselectedforthe deploymentofaRRH
0 otherwiseZi 1 Ifith fibrefacilityisselectedforthe deploymentofBBU
0 otherwiseCi,j 1 Ifthere isafibreroutefromith nodetojth node
0 otherwiseYi,j integer Numberoffibresinthepathfromith nodetojth nodeBi 1 Ifith householdiscovered
0 otherwise
Antenna Base-Stations• Multiple Sectors, MIMO, Multiple Bands, Carrier Aggregation – leads
to complexities or complex requirements for backhaul/front-haul
CoreNetwork
Interfa
ce
BaseStatio
nCo
ntrol,
Clockand
Managem
ent
Baseband
Processin
gFunctio
ns
DownlinkDigitalIF
TxInterface
RxInterface
UplinkDigitalIF
…
…
GainBlocks
GainBlocks Du
plex/Switch
…AntennaSystem
…
…
…
…
Inter-basestationInterface
Base-bandUnit(BBU) RemoteRadioUnit(RRU)
Backhaul Fronthaul
18
NetworkingofBase-Stations(BSs)Evolutionbasestationarchitectures
Base-Station
SwitchingCentre/
MobileCore
WirelessBack-haul
WirelessFront-haul(RF)
Base-BandProcessing
Unit
RemoteRadioUnit(RRU)
DistributedAntennaSystems
DigitalFiber Links/NetworksFeedingBase-stations
IndustryApproachhasbeenCommonPublicRadioInterface(CPRI)andOpenBase-StationArchitectureInterface
Smallcellsand/orRemoteAntennaTerminals
CloudofBBUs
19
Optical-WirelessIntegrationIncreaseinBase-
stationdensitiesduetomoresmallerCells
>85/25km2
Increasingshareofpowerconsumptionof
cellularnetworks~60%
Rapidlyrisingbandwidth
requirementsforbackhaul/fronthaul
Increasedfootprintofinstalledfiberand
planneddeployment(>>180Mkm )
MaturingofopticalTransceiver
(10Gbpsandbeyond)
Pathwaysforloweringthecostof
deployment
Whyoptical– wirelessnetworkintegration
makessense
20
Cloud/Virtualisation/Software Defined
FullyCentralisedandVirtualPoolofBBUs PartiallyCentralised
ChinaMobile’sC-RANVirtualNetworkFunction
VirtualisationContainer
21
4G-5G:ChallengesandOpportunities
CellDensity10x
RFBandwidth10x
SpectralEfficiency20x
Latency1/10
• AntennaArraysinMassiveMIMO?
• Millimeter-wavesorSpectralfarming
• Virtualisable andSoftwaredefined
• Cloudaccessnetwork• Storageinthenetwork• Secure• Criticalservices
• PhotonicSystemsforMassiveMIMOs
• Microwave/Millimeter-wave/Photonicintegration
• JointNetworkPlanningandOptimisation
• SoftwareDefinedOpticalNetworking
22
5G – Early Demonstrations• Use of 1GHz is useful for coverage (rural and indoor) • above 6GHz is useful for very high data rates and
shorter-range connectivity (15GHz, 28GHz, 60GHz, 70-85GHz)
• Samsung: 1.2Gbps transmission at 110km/h speed using 28GHz frequency. The stationary transmission test is up to 7.5Gbps
• Ericsson: 5 Gbps throughput at 15 GHz
23
Optical Interfaces - CPRICo
reNetwork
Interfa
ce
BaseStatio
nCo
ntrol,
Clockand
Managem
ent
Baseband
Processin
gFunctio
ns
DownlinkDigitalIF
TxInterface
RxInterface
UplinkDigitalIF
…
…
GainBlocks
GainBlocks Du
plex/Switch
…AntennaSystem
…
…
…
…
Inter-basestationInterface
Base-bandUnit(BBU) RemoteRadioUnit(RRU)
Backhaul Fronthaul
Base-BandProcessing
Unit
RemoteRadioUnit(RRU)
CPRI/OBSAIDigitisedIFoverFiber
24
Digital Interface – CPRI/OBSAI
Totalflow=Data+ControlManagement+SYNC
OBSAI/CPRI
DUC
DDC
CFR/DPD
D_Filter
DAC
ADC
ADC
FilterDUC
DDCD_Filter
Filter
LO_1 LO_2
SampleRateConversion Filter
Filter
FilterFilter
FilterFilter
I/Q
I/Q
I/Q
I/Q
RF_1
RF_2
RF_1
RF_2
SFP+
Base-band
AnalogRFunit
25
OBSAI Vs CPRIOBSAI CPRI
8B/10Blinecoding 8B/10Blinecoding
upto6.144/3.072Gbps upto3.072Gbps
FixedIQsampleenvelopesizeat16 Programmable8-20indownlink4-20inuplink
Includetransportandapplicationlayers OnlyPhysicalanddatalinklayers
OBSAI CPRI
UserData(IQ) 80% 93.75%
ControlData(O&M) 4% 6.225%
Synchronization(K-char) 0.25% 0.025%
FixedOverheads 15.75% 0%
OBSAI CPRI
LTE10MHz@12bits 4 6
LTE20MHz@12bits 2 3
LTE10MHz@16bits 4 4
LTE20MHz@16bits 2 2
• OBSAI - higher overheads (+15.75%) for enabling flexibility
• CPRI- capacity advantage (+13.75%) for optimized bandwidth allocation
• OBSAI capacity - independent of sample envelope size;
• CPRI - optimized sample envelope size => more carrier space
26