Current Frontiers in Wireless Communicationswcnc2009.ieee-wcnc.org/downloads/Gerhard.pdf · TU...
Transcript of Current Frontiers in Wireless Communicationswcnc2009.ieee-wcnc.org/downloads/Gerhard.pdf · TU...
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
Current Frontiers in Wireless Communications: Fast & Green & Dirty
Gerhard Fettweis
TU Dresden Gerhard Fettweis Slide 2
TM
Chair and Its Partners
32 Ph.D. students25+ Ms students4 sen. scientists1 post-doc1 professor1 project mgr2 secretaries4 engineers
ProjectsSponsors
IPP 2008Oct-1
Scientific:44 Ph.D. grads175+ Ms. grads500 publications40+ patents
Innovation:8 spinouts200 engineers
Funding:€
+35M Chair€
+40M VC€
250M projects.
Numbers
TU Dresden Gerhard Fettweis Slide 3
The
Vodafone Chair‘s
Startup
History
2002
2007
1999
OnDSP™
based WLAN chip-sets
2000
UMTS/3G network optimization and planning
2003
LTE FDD & TDD test mobile
2004
Module and reference board design
2005 MPSoC
semiconductor IP
2007
Wireless embedded technology
2008 Network performance measurement
2008
LTE Cellular Technology Provider
freedelity
2006
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
FAST
TU Dresden Gerhard Fettweis Slide 5
Embedded Mobile Memroy: Single-Chip Flash Memory Trend
100Gb
10Gb
1Gb
100Mb
10Mb
1Mb
100Kb
10Kb
1995
2000
2005
2010
2015
4Gb
16Gb8Gb
TU Dresden Gerhard Fettweis Slide 6
“The Wireless Roadmap”
TU Dresden Gerhard Fettweis Slide 7
Data Synchronization: Point-to-Point
100Gb/s
10Gb/s
1Gb/s
100Mb/s
10Mb/s
1Mb/s
100Kb/s
10Kb/s
1995
2000
2005
2010
2015
USB 1.0
USB 2.0
USB 3.0
UWB intention802.15.3c
Short links (1m)
TU Dresden Gerhard Fettweis Slide 8
Client Server Applications: WLAN
1995
2000
2005
2010
2015
Short links (1m)
WLAN (10m)802.11vht
802.11n802.11ag
802.11
802.11b
100Gb/s
10Gb/s
1Gb/s
100Mb/s
10Mb/s
1Mb/s
100Kb/s
10Kb/s
TU Dresden Gerhard Fettweis Slide 9
Coverage: Cellular
1995
2000
2005
2010
2015
Short links (1m)
Cellular (100m)
GSM GPRS
HSPA
HSDPALTE
WiMAX
WLAN (10m)
3G R99 / EDGE
LTE Advanced
100Gb/s
10Gb/s
1Gb/s
100Mb/s
10Mb/s
1Mb/s
100Kb/s
10Kb/s
TU Dresden Gerhard Fettweis Slide 10
The Wireless Roadmap
1995
2000
2005
2010
2015
Short links (1m)
Cellular (100m)
WLAN (10m)
100x
10x
ITRS Roadmap: Continues until 2020
Driven By Moore’s Law
100Gb/s
10Gb/s
1Gb/s
100Mb/s
10Mb/s
1Mb/s
100Kb/s
10Kb/s
TU Dresden Gerhard Fettweis Slide 11
The ITU “Van”
Diagram
TU Dresden Gerhard Fettweis Slide 12
2004: Integration Nightmare Roadmap of Wireless Standards into Handsets
2004 2005 2006 2007 2008 2009
Status Today
GSM/UM
TS
Bluetooth
IrDA
+ WiFi 802.11b
+ DVB-T/H
+ WiM
ax+ UWB Wireless USB
+ NFC
+ WiFi 802.11g
+ WiFi 802.11n
+ 802.20
+ ZigBee
WiFiIEEE 802.11b: 11Mb/s wireless LAN @ 2.4GHz (up to 100m)IEEE 802.11a/g: 54Mb/s wireless LAN @ 2.4/5GHz (up to 100m)IEEE 802.11n: 250Mb/s wireless LAN @ 5GHz (up to 30m)
UWB (IEEE 802.15.3a): 480Mb/s wireless USB2.0 (typ. 1m)WiMax (IEEE 802.16d/e): ~10Mb/s broadband wireless accessNFC: Near Field Communications (RF-ID, wireless payment,…)
ZigBee: (IEEE 802.15.4)Sensor networks, remote control
BluetoothVersion 2: TBD?
+ FM Radio
+ GPS
+ Galileo
+ MediaFlo
TU Dresden Gerhard Fettweis Slide 13
2009: Integration Cleanup Roadmap of Wireless Standards into Handsets
2007 2008 2009 2010 2011 2012
Status Today
GSM/UM
TS
Bluetooth
IrDAFM Radio
WiFi 802.11bg
GPS
+ WiM
ax+ W
iFi 802.11n+ Galileo
- FM Radio
- BlueTooth
Status Tomorrow
LTE / WiM
AX
(UWB+??)
(GPS/Galileo)
(WiFi 802.11)
- IrDA
Loosers: MediaFlowDVB T/HNFCZigBee802.20
UWB
TU Dresden Gerhard Fettweis Slide 14
The Only Wireless Features We Need?
Cellular Connectivity Fast
Broadband Local Connectivity Fast
Positioning
TU Dresden Gerhard Fettweis Slide 15
How to Achieve “FAST”
in Cellular
TU Dresden Gerhard Fettweis Slide 16
Shaping the SINR Distribution
Active „shaping“
of SINRInterference CancellationPower Control
AdvantagesHigh average SINR
High spectral efficiencyEnabler for MIMO
Small VarianceHigh grade of fairness
no no
TU Dresden Gerhard Fettweis Slide 17
Fairness & Data Rate & Spectral Efficiency
0.5
0.4
0.3
0.2
0.1
0.0
-10 -5 0 5 10 15 20 25 30SIR in dB
Reuse 1Power Control
E[SIR] = -0.2dB
Reuse 1Power Control
10 Interferers cancelledMacro/Distributed MIMO
E[SIR] = 8.1dB
of SIR
fairn
ess
Interference Cancellation: Fairness & High Data Rate
TU Dresden Gerhard Fettweis Slide 18
CoMP: Coordinated Multi-Point
We thus believe that next generation systems will include multi-cell cooperative signal processing (“network MIMO” or CoMP):
TU Dresden Gerhard Fettweis Slide 19
Roadmap View
1st/2nd generation 3rd generation 4th generation
•
Interference avoidance through high reuse factors
•
Interference shaping and exploitation through distributed MIMO and relaying
•
Interference suppression through classical MIMO
TU Dresden Gerhard Fettweis Slide 20
Potential Gains
of CoMP
Uplink Downlink
Okumura-Hata pathloss model, ITU pedestrian ALink-to-system mapping (MIESM), 8 MCS schemesSpectral eff. losses through guard bands / intervalsAssuming perfect channel est., 2 rx ant. per eNB
Linear joint transmission, assuming perfect channelknowledge at the eNBs
2 tx ant. per eNB
TU Dresden Gerhard Fettweis Slide 21
Limits in Clustered
Virtual
MIMO A Graph Theory
Approach
We can easily observe how many interferers would have to be perfectly cancelled for a user at any location within a cell, to achieve a capacity target:
Target 3 bit/s/Hz/sector Target 4 bit/s/Hz/sector
Note that we are assuming perfect interference cancellation (optimistic), but do not take into account increased diversity (pessimistic)
TU Dresden Gerhard Fettweis Slide 22
Spectral Efficiency
0,1 0,3 0,51,2
3
10
0123456789
10
GSM EDGE HSPA LTE EASY-C Theory
SpectralEfficienncybit/s/Hz/sector GAP
License Cost (GER)1Hz ~ EUR 1000100Mb/s/sector required:- GSM: 1GHz €1T-
LTE: 100MHz €100B-
Theory: 10MHz €10B
20 YEARSOf Engineering
1987 - 2007
20 further YEARSOf Engineering
2007-2027?
Enablers for Ambient Services & Systems Part C: Wide Area Coverage
EASY-C Update
24
Project Partners Coordinator: Gerhard Fettweis
Rosenheim
Dresden
Nürnberg
UlmStuttgart
Darmstadt
Paderborn
Bonn
Düsseldorf
Berlin
München
Aachen
Mainz
Volume:€ +47M project€ +27M funding
25
Current View on Network MIMO
System partitioningReducing
Backhaul
/ Infrastructure
Aspects
Scheduling
Synchronization
in time / frequencyChannel estimation
&Obtaining
transmitter-
side
CSI at eNBs
Impact of network
MIMO
on higher
protocol
layers
See the following challenges connected to network MIMO:
26
Project Workflow
2.1 Specification
1.1 Algorithm & Concept Exploration
1.2 Algorithm Classification & Selection
2.2 Implementation
2.3 Lab Test
2.4 Field Test
3.1 Specification
3.2 SoC Architecture
3.3 Implementation
3.4 Demonstrators
1.3 Assessment & Dissemination
WG 1
Algorithms & Concepts
WG 2Technology Test Bed
WG 3Hardware Architecture
Working Groups Alcatel-LucentBNetzADt. TelekomEricssonFhG HHIIFX/ComneonKathreinNXPQualcommRadioplanSignalionTESTU DresdenUbidyneUni PaderbornVodafone
27
Mini Out-to-Indoor Testbed Dresden TUD Indoor Test, Southern View
Two sectors are installed on the roof of TUD building:
IndoorTerminals
Microwave radio link LAN Network ConnectionOur Tower
Amateur RadioTower
BARKHAUSEN Building
© Microsoft Live
IndoorNLOSHallwayMeasurements
28
Lab Site:2 sites at the chairAntenna: 1700-2700MhzMicrowave link between the sites
Mobile Lab UE:Sorbas UE12V BatteryNotebook
Mini-Testbed Dresden TUD Lab Test
29
Field Test Berlin Overview
In Berlin, a second test bed is beingset up, with a focus on services and applications
AirlaserAirl
aser
optical fiber network
1
23
4
T-Labs TU Berlin
HHI
30
Cellular After 2015
© Google Earth
Hbf-Süd
Karstadt
Postplatz
Lennéplatz
Mitte
Kongresszentrum
Fritz-Förster-Pl.
Strassburger Pl.
WTC
HbfICC 2009Dresden
31
Testbed Dresden Field Test
LTE+ antennas co-located at real 3G sites (downtown, hight of antennas 30-55m)
Support from project partner(Vodafone, T-Mobile)
Backhaul realized withMotorola Canopy PTP600 system (250 Mbit/s TDD system)
TU Dresden Gerhard Fettweis Slide 32
Additional Challenges:“Lessons Learned”
TU Dresden Gerhard Fettweis Slide 33
OFDM for Downlink CoMP
t = t0
-τ t = t0
-τ
t = t0 t = t0
τ
≈
d/cUse CP to compensate?LTE: d=360m
d distance between cooperating base stations
TU Dresden Gerhard Fettweis Slide 34
Carrier Phase Offset, Phase Noise
Carrier Phase Offset (CPO)Caused by LO imperfections
Phase Noise (PN)Caused by LO jitter due to non perfect sine generation
CoMP PN increased by 3-5dB
BPF LNA90°
PLL
LPF
LPF VGA
VGA ADC
ADC
PLL~
Rx DFE
AGC
Rx AFESRC
SRC
BPF PA
Tx AFE
90°
DAC
DAC
PLL PLL~
Tx DFESRC
SRCTS
,fS
,BSTSN
,fSN
,BSN TC
,fC
,BC
h(t,τ) Doppler Spread
θ
‐θ
©
Vincent Kotzsch, 2008
TU Dresden Gerhard Fettweis Slide 35
Sampling Clock Offset
Sampling Clock Offset (SCO)Sampling offset between ADC and DACCaused by LO imperfectionsRange: 10-5 - 50ppm related to sampling rateCoMP “negligible asynchronous OFDM”
BPF LNA90°
PLL
LPF
LPF VGA
VGA ADC
ADC
PLL~
Rx DFE
AGC
Rx AFESRC
SRC
BPF PA
Tx AFE
90°
DAC
DAC
PLL PLL~
Tx DFESRC
SRCTS
,fS
,BSTSN
,fSN
,BSN TC
,fC
,BC
h(t,τ) Doppler Spread
©
Vincent Kotzsch, 2008
TU Dresden Gerhard Fettweis Slide 36
Carrier Frequency Offset
Carrier Frequency Offset (CFO) Continuously phase shift over timeCaused by LO-imperfections and DopplerRange: 10-5 - 50ppm related to fCCoMP “artificial Doppler”
BPF LNA90°
PLL
LPF
LPF VGA
VGA ADC
ADC
PLL~
Rx DFE
AGC
Rx AFESRC
SRC
BPF PA
Tx AFE
90°
DAC
DAC
PLL PLL~
Tx DFESRC
SRCTS
,fS
,BSTSN
,fSN
,BSN TC
,fC
,BC
h(t,τ) Doppler Spread
fC
fC
+ Δ
2πfC
TS
2πfC
2TS
©
Vincent Kotzsch, 2008
TU Dresden Gerhard Fettweis Slide 37
Requirement “FAST”
will not end soon
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
GREEN
cool
TU Dresden Gerhard Fettweis Slide 39
The Server Farms Are Growing
Moore’s law for logic:5x every 5 yearsMoore’s law for flash10x every 5 years
RESULT:2x power consumption every 5 years
TU Dresden Gerhard Fettweis Slide 40
In 2007 the Energy supply of ICT produced CO2
-output at level of 25% of
worldwide cars
The Challenge
Classic Internet (Computing & Mobile)
Cisco IP Traffic Forecast 2008
Internet of Things(networked sensors)
2007 2017
TU Dresden Gerhard Fettweis Slide 41
Energy Costs= Invest
Energy Costs= Personnel2007
Paradigm Change
CellularCustomers (Mio.)
200
400
600
1996
2008
1996
2008
1‘000
2‘000
3‘000
ComputingInternet-Server (Mio.)
QuelleISC
QuelleGSM World
TU Dresden Gerhard Fettweis Slide 42
Server farms & telco infrastructureCurrently 3% of world electric energy consumptionIncreasing by 16-20% / yearDoubling world el. energy consumption by 2030 driven by ICT
ICT: 10% of el. energy demand within industrialized nations900 Bill.. kWh / year = Central and South AmericasWorldwide proliferation: 40% of world el. energy production
Dramatic improvement in energy efficiency needed !
Energy Demand & CO2
of ICT
TU Dresden Gerhard Fettweis Slide 43
Cellular: What Do We Know
Cellular networks: 0.5% of world-wide electrical energyHandsets: 1% of total cellular energy consumptionNetwork: 99% of total cellular energy consumption
Running the network:Expected energy bill: 3x (200%) over next 7 years80% of energy consumed at base station site
50% for coolingE.g. UMTS CDMA: 25% power in “pilot pollution” !!!“Always on” the wrong strategy?!
Need of “border-crossing” optimization: System Engineering
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
CoordinatorProf. Gerhard Fettweis
45
Key Facts of Cool Silicon
Partners–
Companies•
Currently around 40 companies involved–
Universities•
TU Dresden•
TU Chemnitz•
HTW Dresden–
Research Institutes
Money–
EUR 100M funding–
EUR 150M total volume
Duration: Feb-2009 to Feb 2014
TU Dresden Gerhard Fettweis Slide 46
How to Achieve “GREEN”in Handsets (& Base Stations)
TU Dresden Gerhard Fettweis Slide 47
LTE FDD & TDD / WiMAX
Single-Chip SDR: Tomahawk –
Block-Diagram
32
3232
DC_570TRISC No. 1
DC_570TRISC No. 2
GPIO (8)
32 32
32
VGAIF
STREAMIF PCI-EXPRESS
SLAVE SLAVE MASTER MASTER SLAVE
FPGA-BRIDGE
32
COREMANAGER
SLAVEMASTER
SLAVE MASTER
DMADMA
32
MASTER
32
INTBUS (32 BIT)
MASTER MASTER
32 32
SLAVE
MASTER
32
SLAVE
32
VDSPFXT16
VDSPFXT16
32
SLAVE
VDSPFXT16
32
SLAVE
VDSPFXT16
32
SLAVE
RKDSP
32
SLAVE
RKDSP
32
SLAVE
LDPC
32
SLAVE
DEBLOCKER
32
PLD-ApplicationsTensilica
MASTER MASTER
Dresden Silicon
Vodafone Chair
DDR-RAM 16/32 BIT
32
SLAVE
DDR-RAM 16/32 BIT
32
SLAVE
DMA
SLAVE MASTER MASTER
32 32
32
UART 2
32
UART 1
32
SLAVE
Entropy
32
SLAVE
StreamParser
DDR-RAM 16/32 BIT
32
SLAVE
DMA
MASTER
32
MASTER
32
ScratchPad
32
SLAVE SLAVE
32
ASIC-Architectinc
52 4 4338 42 42 913742
EXTBUS (32 BIT)
SLAVE
VDSPFXT16
32
SLAVE
VDSPFXT16
32
BOOT
2
Our Testchip”TOMAHAWK”: running silicon Jan 18, 2008
TensilicaDiamond
212GP
TensilicaDiamond
212GP
TU Dresden Gerhard Fettweis Slide 48
LTE FDD/ TDD/ WiMAX
Single-Chip SDR: Tomahawk -
Die Photo
STA LDPC Decoder/
DeblockerASIP
STA Vector DSP
2x STA SIOUX
Core Manager
3x DDR Controller
10 mm
10 mm2x XtensaDC212GP
Scratchpad
PLL
PLL
PLL
Silicon on Jan 18 2008
In 45 nm CMOS:Complete LTE BB
< 20mm2
< 200mW
130nm UMC 57M transistors
TU Dresden Gerhard Fettweis Slide 49
Area & Power Efficiency
1.
MIMO SVD ASIC2.
Consumer RISC3.
Multimedia DSP4.
FFT Processor5.
RISC CPU with Media Processor
6.
Communications DSP7.
CAM LSI8.
Video Stream Multi-
Processor
9.
Hearing Aid DSP
Source: Markovic et. al., Power and Area Minimization for Multidimensional Signal Processing, IEEE J. Solid-StateCircuits, vol. 42, no. 4, pp. 922-934, April 2007. Results scaled to 90 nm, operations are 12 bit add equivalents
1 2
3
4
5
67
8
9 by 1
TU Dresden Gerhard Fettweis Slide 50
How to Achieve “GREEN” in Networks
TU Dresden Gerhard Fettweis Slide 51
Some Issues
Network topologyNetwork architecture
Network componentsNetwork interconnections
Network supported featuresNetwork delivered bandwidth
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
DIRTY RF
TU Dresden Gerhard Fettweis Slide 53
Dirty RF
Picking up “dirt” Nonlinear
LNAFeedthrough
Coupling
Phase noise
Aperture
JitterAmbiguity
I/Q ImbalanceRRC mismatch
Flicker NoiseDigital noiseNonlinear
PADSP:LivingWithDirtyRF
TU Dresden Gerhard Fettweis Slide 54
Limited Tx-Rx Isolation due to miniaturization and frequency agility→ Tx Leakage (TxL)→ Strong out-of-band interference in Rx branch
Transmit Leakage in FDD and Direct Conversion
Approaches for Solution:
Problem: Tx
Leakage in FDD Terminals
Additional bandpass filterAdaptive filter in parallel to duplexerDirty RF: Compensation of analog impairments by digital signal processing+ Reconfigurable / standard independent / relaxed RF requirements
-
Limited reconfigurability-
Increased analog complexity
DL signal
TU Dresden Gerhard Fettweis Slide 55
Dup
lexe
r
Tx
Leakage in Zero-IF Receivers
Nonlinearity of I/Q down converter Intermodulation product 2nd order of Tx Leakage (TxL-IM2)
Low freq. TxL-IM2 interfering desired, down converted DL signal
Tx
Leakage intermodulation
products 2nd order (TxL-IM2)
DL signal
TxL- IM2
TxL channel
DC-
TU Dresden Gerhard Fettweis Slide 56
Antenna nearfield distorted bymetal interrupter moved withindistance [0.036, 0.303] m to the antenna
Tx-Rx
Isolation
Measurement
with
UMTS SAW Duplexer
B7632 (EPCOS AG)
UL-Band DL-Band
B
Frequenz [GHz]
Tx-R
x Is
olat
ions
däm
pfun
g [d
B]
1.9 1.95 2 2.05 2.1 2.15 2.2-70
-65
-60
-55
-50
meanmin / max
Frequency
[GHz]
Tx-R
xis
olat
ion
[dB
]
Tx-Rx isolation affected by nearfielddistortionsOnly small variations in UL-Band (ca. 2.2 dB)TxL Estimation must be adaptive
TU Dresden Gerhard Fettweis Slide 57
SNR loss
due
to TxL
-5 0 5 10 15 20 25 30
0
2
4
6
8
10
12
14
16
18
20
γ̃TxL[dB]
∆γ̃w.r.tw/ oTxL[dB]
w/o TxL
CmpGenie TxL
CmpML, τ=3.5LMS
τ=3.5meas.τ=1.5
dig. TxL
Cmp
required
TxL
negligible
analog. TxL
Cmp
req.
OFDM susceptible to TxL for γTxL < 30dBdigital TxL compensationsuitable for γTxL > 0dBTxL can be mitigated digitally up to 30 dB
Simulation Parameters:DL / UL: 802.11a, 16QAMDL channel: HiperLAN A CSF EQ in TxL Est.analog DC-1
ADC: 8-bit lin. mid-rise, η=2TxL: - TxL ch: 6-tap exp PDP
and meas. IR- TxL SIR γTxL = 10dB- no TxL I/Q mismatch
TxL Est: - NB=4000,
ML reaches almost Genie Cmp. limit, but suffers from high min. SNR lossLMS performance stronglydepending on TxL channel property
TU Dresden Gerhard Fettweis Slide 58
Phase Noise and Clippping
-1.5 -1 -0.5 0 0.5 1 1.5-1.5
-1
-0.5
0
0.5
1
1.5
In
Qua
• Phase Noise:• Kalman
Tracking• Complete Sync
-1.5 -1 -0.5 0 0.5 1 1.5-1.5
-1
-0.5
0
0.5
1
1.5
InQ
ua
• Clipping:• Analogue PAPR• Saleh
Model• AM/PM distortion
-1.5 -1 -0.5 0 0.5 1 1.5-1.5
-1
-0.5
0
0.5
1
1.5
In
Qua
• Clipping & PN:• Rate
TU Dresden Gerhard Fettweis Slide 59
Phase Noise in OFDM
TU Dresden Gerhard Fettweis Slide 60
Setup:IEEE 802.11a SISOExponential PDP, 8 taps64-QAM f = 4GHz, IBO = 0dB
Rate under Phase Noise & Clipping
-10 0 10 20 30 400
1
2
3
4
5
6
SNR [dB]
Rat
e [b
its]
no distortionno compensationIa
= 0.2, ICI1
Ia
= 0.4, ICI2
Ia
= 0.6, ICI3
Ia
= 0.8, ICI4
TU Dresden Gerhard Fettweis Slide 61
Many Further Issues
“Estimation and Detection”
algorithm designNonlinearitiesPhase noiseI/Q imbalanceSampling jitterTransmit leakage…
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
Conclusions
TU Dresden Gerhard Fettweis Slide 63
FAST increasing data transmission rates
GREEN much lower-power “cool communications”
RF impairment estimation with DIRTY RF approach
We need much more further research !
Vodafone Chair Mobile Communications Systems, Prof. Dr.-Ing. G. Fettweis
Thanks !
Thank you Vodafone for the continued support of the chair
www.vodafone-chair.com