This seminar will provide the basics of this fascinating technology. After attending this seminar you will understand OFDM-principles,
including SC-FDMA as the transmission scheme of choice for the LTE uplink. Multiple antenna technology (MIMO) is a fundamental
part of LTE and its impact on the design of device and network architecture will be explained. Further LTE-related physical layer
aspects such as channel structure and cell search will be presented with an overview of the LTE protocol structure.
The second part of the seminar provides an overview of the evolution in LTE towards 3GPP specification Release 9 and 10. This
includes features and methods for location based services like GNSS support or time delay measurements and the concept of
multimedia broadcast. Finally, we’ll introduce the main features of LTE-Advanced (3GPP Release-10) including carrier aggregation for
a larger bandwidth and backbone network aspects like self-organizing networks and relaying concepts.
UiPath Community: Communication Mining from Zero to Hero
LTE Evolution: From Release 8 to Release 10
1. UMTS Long Term Evolution
(LTE)
Reiner Stuhlfauth
Reiner.Stuhlfauth@rohde-schwarz.com
Training Centre
Rohde & Schwarz, Germany
Subject to change – Data without tolerance limits is not binding.
R&S® is a registered trademark of Rohde & Schwarz GmbH & Co. KG. Trade names are trademarks
of the owners.
2011 ROHDE & SCHWARZ GmbH & Co. KG
Test & Measurement Division
- Training Center -
This folder may be taken outside ROHDE & SCHWARZ facilities.
ROHDE & SCHWARZ GmbH reserves the copy right to all of any part of these course notes.
Permission to produce, publish or copy sections or pages of these notes or to translate them must first
be obtained in writing from
ROHDE & SCHWARZ GmbH & Co. KG, Training Center, Mühldorfstr. 15, 81671 Munich, Germany
3. 3GPP work plan
GERAN
EUTRAN
New
RAN
Phase 1
Phase 2, 2+ UTRAN
Rel. 95
… Rel. 97
Rel.7 Rel. 99 Up from Rel. 8
…
Rel. 7
Rel. 9
Also contained in
Rel. 10
Evolution
November 2012 | LTE Introduction | 3
4. Overview 3GPP UMTS evolution
HSDPA/ LTE and LTE-
WCDMA
WCDMA HSPA+
HSUPA HSPA+ advanced
3GPP 3GPP Study
3GPP
release 3GPP Release 99/4 3GPP Release 5/6 3GPP Release 7 3GPP Release 8
Item initiated
Release 10
App. year of 2005/6 (HSDPA)
network rollout 2003/4 2008/2009 2010
2007/8 (HSUPA)
Downlink LTE: 150 Mbps* (peak) 100 Mbps high mobility
384 kbps (typ.)
384 kbps (typ.) 14 Mbps (peak) 28 Mbps (peak) HSPA+: 42 Mbps (peak)
28 Mbps (peak)
data rate 1 Gbps low mobility
Uplink LTE: 75 Mbps (peak)
128 kbps (typ.)
128 kbps (typ.) 5.7 Mbps (peak) 11 Mbps (peak) HSPA+: 11 Mbps (peak)
11 Mbps (peak)
data rate
Round
Trip Time ~ 150 ms < 100 ms < 50 ms LTE: ~10 ms
*based on 2x2 MIMO and 20 MHz operation
November 2012 | LTE Introduction | 4
5. Why LTE?
Ensuring Long Term Competitiveness of UMTS
l LTE is the next UMTS evolution step after HSPA and HSPA+.
l LTE is also referred to as
EUTRA(N) = Evolved UMTS Terrestrial Radio Access (Network).
l Main targets of LTE:
l Peak data rates of 100 Mbps (downlink) and 50 Mbps (uplink)
l Scaleable bandwidths up to 20 MHz
l Reduced latency
l Cost efficiency
l Operation in paired (FDD) and unpaired (TDD) spectrum
November 2012 | LTE Introduction | 5
6. Peak data rates and real average throughput (UL)
100
58
11,5 15
10
5,76
Data rate in Mbps
5
2 2 1,8
2
0,947
1
0,473 0,7
0,5
0,174 0,153
0,2
0,1 0,1 0,1
0,1
0,03
0,01
GPRS EDGE 1xRTT WCDMA E-EDGE 1xEV-DO 1xEV-DO HSPA HSPA+ LTE 2x2
(Rel. 97) (Rel. 4) (Rel. 99/4) (Rel. 7) Rev. 0 Rev. A (Rel. 5/6) (Rel. 7) (Rel. 8)
Technology
max. peak UL data rate [Mbps] max. avg. UL throughput [Mbps]
November 2012 | LTE Introduction | 6
8. Round Trip Time, RTT
•ACK/NACK
generation in RNC MSC
TTI
Iub/Iur Iu
~10msec
Serving SGSN
RNC
Node B
TTI
=1msec
MME/SAE Gateway
eNode B
•ACK/NACK
generation in node B
November 2012 | LTE Introduction | 8
9. Major technical challenges in LTE
New radio transmission FDD and
schemes (OFDMA / SC-FDMA) TDD mode
MIMO multiple antenna Throughput / data rate
schemes requirements
Timing requirements Multi-RAT requirements
(1 ms transm.time interval) (GSM/EDGE, UMTS, CDMA)
Scheduling (shared channels, System Architecture
HARQ, adaptive modulation) Evolution (SAE)
November 2012 | LTE Introduction | 9
10. Introduction to UMTS LTE: Key parameters
Frequency
UMTS FDD bands and UMTS TDD bands
Range
1.4 MHz 3 MHz 5 MHz 10 MHz 15 MHz 20 MHz
Channel
bandwidth,
1 Resource 6 15 25 50 75 100
Block=180 kHz Resource Resource Resource Resource Resource Resource
Blocks Blocks Blocks Blocks Blocks Blocks
Modulation Downlink: QPSK, 16QAM, 64QAM
Schemes Uplink: QPSK, 16QAM, 64QAM (optional for handset)
Downlink: OFDMA (Orthogonal Frequency Division Multiple Access)
Multiple Access
Uplink: SC-FDMA (Single Carrier Frequency Division Multiple Access)
Downlink: Wide choice of MIMO configuration options for transmit diversity, spatial
MIMO
multiplexing, and cyclic delay diversity (max. 4 antennas at base station and handset)
technology
Uplink: Multi user collaborative MIMO
Downlink: 150 Mbps (UE category 4, 2x2 MIMO, 20 MHz)
Peak Data Rate 300 Mbps (UE category 5, 4x4 MIMO, 20 MHz)
Uplink: 75 Mbps (20 MHz)
November 2012 | LTE Introduction | 10
13. Orthogonal Frequency Division Multiple Access
l OFDM is the modulation scheme for LTE in downlink and
uplink (as reference)
l Some technical explanation about our physical base: radio
link aspects
November 2012 | LTE Introduction | 13
14. What does it mean to use the radio channel?
Using the radio channel means to deal with aspects like:
C
A
D
B
Transmitter Receiver
MPP
Time variant channel
Doppler effect
Frequency selectivity attenuation
November 2012 | LTE Introduction | 14
15. Still the same “mobile” radio problem:
Time variant multipath propagation
A: free space
A: free space
B: reflection
B: reflection
C C: diffraction
C: diffraction
A
D: scattering
D: scattering
D
B
Transmitter Receiver
Multipath Propagation reflection: object is large
and Doppler shift compared to wavelength
scattering: object is
small or its surface
irregular
November 2012 | LTE Introduction | 15
16. Multipath channel impulse response
The CIR consists of L resolvable propagation paths
L 1
h , t ai t e i
ji t
i 0
path attenuation path phase path delay
|h|²
delay spread
November 2012 | LTE Introduction | 16
17. Radio Channel – different aspects to discuss
Bandwidth or
Wideband Narrowband
Symbol duration
or t
t
Short symbol Long symbol
duration duration
Channel estimation: Frequency?
Pilot mapping Time?
frequency distance of pilots? Repetition rate of pilots?
November 2012 | LTE Introduction | 17
18. Frequency selectivity - Coherence Bandwidth
Here: substitute with single
power Scalar factor = 1-tap
Frequency selectivity
How to combat
channel influence?
f
Narrowband = equalizer
Can be 1 - tap
Wideband = equalizer
Here: find Must be frequency selective
Math. Equation
for this curve
November 2012 | LTE Introduction | 18
19. Time-Invariant Channel: Scenario
Fixed Scatterer
ISI: Inter Symbol
Interference:
Happens, when
Delay spread >
Symbol time
Successive
Fixed Receiver
symbols Transmitter
will interfere Channel Impulse Response, CIR
Transmitter Receiver collision
Signal Signal
t t
Delay Delay spread
→time dispersive
November 2012 | LTE Introduction | 19
20. Motivation: Single Carrier versus Multi Carrier
TSC
|H(f)| f
Source: Kammeyer; Nachrichtenübertragung; 3. Auflage
1
B
TSC
B
t
|h(t)|
t
l Time Domain
l Delay spread > Symboltime TSC
→ Inter-Symbol-Interference (ISI) → equalization effort
l Frequency Domain
l Coherence Bandwidth Bc < Systembandwidth B
→ Frequency Selective Fading → equalization effort
November 2012 | LTE Introduction | 20
21. Motivation: Single Carrier versus Multi Carrier
TSC
|H(f)| f Source: Kammeyer; Nachrichtenübertragung; 3. Auflage
1
B
B TSC
t
|h(t)|
f t
|H(f)|
B 1
f
B N TMC
t
TMC N TSC
November 2012 | LTE Introduction | 21
22. What is OFDM?
Single carrier
transmission,
e.g. WCDMA
Broadband, e.g. 5MHz for WCDMA
Orthogonal
Frequency
Division
Multiplex
Several 100 subcarriers, with x kHz spacing
November 2012 | LTE Introduction | 22
23. Idea: Wide/Narrow band conversion
ƒ
…
S/P
…
…
H(ƒ) t / Tb t / Ts
h(τ) h(τ)
„Channel
Memory“
τ τ
One high rate signal: N low rate signals:
Frequency selective fading Frequency flat fading
November 2012 | LTE Introduction | 23
24. COFDM
Mapper
X
+
X
Data
with
OFDM
FEC Σ
.....
symbol
overhead
Mapper
X
+
X
November 2012 | LTE Introduction | 24
25. OFDM signal generation
00 11 10 10 01 01 11 01 …. e.g. QPSK
h*(sinjwt + cosjwt) h*(sinjwt + cosjwt) => Σ h * (sin.. + cos…)
Frequency
time
OFDM
symbol
duration Δt 2012 |
November LTE Introduction | 25
26. Fourier Transform, Discrete FT
Fourier Transform
H ( f ) h(t )e 2 j ft
dt ;
h(t ) H ( f )e 2 j ft
df ;
Discrete Fourier Transform (DFT)
N 1 N 1 N 1
n n
H n hk e 2 j k n / N
hk cos(2 k ) j hk sin(2 k );
k 0 k 0 N k 0 N
N 1 2 j k
n
1
hk
N
H e
n 0
n
N
;
November 2012 | LTE Introduction | 26
28. Inter-Carrier-Interference (ICI)
10
SMC f
0
-10
-20
-30
xx
S
-40
-50
-60
-70
-1 -0.5 0 0.5 1
f-2 f-1 f0 f1 f2 f
Problem of MC - FDM ICI
Overlapp of neighbouring subcarriers
→ Inter Carrier Interference (ICI).
Solution
“Special” transmit gs(t) and receive filter gr(t) and frequencies fk allows orthogonal
subcarrier
→ Orthogonal Frequency Division Multiplex (OFDM)
November 2012 | LTE Introduction | 28
29. Rectangular Pulse
A(f)
Convolution
sin(x)/x
t
f
Δt
Δf
time frequency
November 2012 | LTE Introduction | 29
31. ISI and ICI due to channel
Symbol l-1 l l+1
h n
n Receiver DFT
Window
Delay spread
fade in (ISI) fade out (ISI)
November 2012 | LTE Introduction | 31
32. ISI and ICI: Guard Intervall
Symbol l-1 l l+1
h n TG Delay Spread
n Receiver DFT
Window
Delay spread
Guard Intervall guarantees the supression of ISI!
November 2012 | LTE Introduction | 32
33. Guard Intervall as Cyclic Prefix
Cyclic Prefix
Symbol l-1 l l+1
h n TG Delay Spread
n Receiver DFT
Window
Delay spread
Cyclic Prefix guarantees the supression of ISI and ICI!
November 2012 | LTE Introduction | 33
34. Synchronisation
Cyclic Prefix
OFDM Symbol : l 1 l l 1
CP CP CP CP
Metric
-
Search window
~
n
November 2012 | LTE Introduction | 34
35. DL CP-OFDM signal generation chain
l OFDM signal generation is based on Inverse Fast Fourier Transform
(IFFT) operation on transmitter side:
Data QAM N Useful
1:N OFDM Cyclic prefix
source Modulator symbol IFFT N:1 OFDM
symbols insertion
streams symbols
Frequency Domain Time Domain
l On receiver side, an FFT operation will be used.
November 2012 | LTE Introduction | 35
36. OFDM: Pros and Cons
Pros:
scalable data rate
efficient use of the available bandwidth
robust against fading
1-tap equalization in frequency domain
Cons:
high crest factor or PAPR. Peak to average power ratio
very sensitive to phase noise, frequency- and clock-offset
guard intervals necessary (ISI, ICI) → reduced data rate
November 2012 | LTE Introduction | 36
37. MIMO =
Multiple Input Multiple Output Antennas
November 2012 | LTE Introduction | 37
38. MIMO is defined by the number of Rx / Tx Antennas
and not by the Mode which is supported Mode
1 1 SISO Typical todays wireless Communication System
Single Input Single Output
Transmit Diversity
1 1 MISO l Maximum Ratio Combining (MRC)
l Matrix A also known as STC
M
Multiple Input Single Output
l Space Time / Frequency Coding (STC / SFC)
Receive Diversity
1 1
SIMO l Maximum Ratio Combining (MRC)
Single Input Multiple Output Receive / Transmit Diversity
M
Spatial Multiplexing (SM) also known as:
l Space Division Multiplex (SDM)
l True MIMO
1 1 MIMO l Single User MIMO (SU-MIMO)
Multiple Input Multiple Output l Matrix B
M M
Space Division Multiple Access (SDMA) also known as:
l Multi User MIMO (MU MIMO)
l Virtual MIMO
Definition is seen from Channel l Collaborative MIMO
Multiple In = Multiple Transmit Antennas Beamforming
November 2012 | LTE Introduction | 38
39. MIMO modes in LTE
-Spatial Multiplexing
-Tx diversity
-Multi-User MIMO
-Beamforming
-Rx diversity
Increased
Increased
Throughput per
Throughput at
Better S/N UE
Node B
November 2012 | LTE Introduction | 39
40. Diversity – some thoughts
The SISO channel:
Fading on the air interface
h11
transmit signal s received signal r
j
r h11 s n h11 e s n
Amplitude phase
scaling rotation
The transmit signal is modified in amplitude and phase
plus additional noise
November 2012 | LTE Introduction | 40
41. Diversity – some thoughts: matched filter
The SISO channel and matched filter (maximum ratio combining):
h11 h*11
received signal r estimated signal ř
transmit signal s
~ h* r h* h s h* n h e j h e j s h* n h 2 s h* n
r 11 11 11 11 11 11 11 11 11
Idea: matched filter multiplies received signal with
conjugate of channel -> maximizes SNR
Transmitted signal s is estimated as:
~
r
~
s 2
h11
November 2012 | LTE Introduction | 41
42. Diversity – some thoughts: performance of SISO
Euclidic distance
Detector
0 threshold
1
Decay with SNR, only
one channel available -
> fading will deteriorate
noise amplitude Bit error rate
distribution
Modulation Bit error Data rate = bits
scheme probability per symbol
BPSK 1/(4*SNR) 1
QPSK 1/(2*SNR) 2
16 QAM 5/(2*SNR) 4
November 2012 | LTE Introduction | 42
43. RX Diversity
Maximum Ratio Combining depends on different fading of the
two received signals. In other words decorrelated fading
channels
November 2012 | LTE Introduction | 43
44. Receive diversity gain – SIMO: 1*NRx
Receive antenna 1
time
r1=h11s+n1
Transmit s NTx NRx
h11
antenna Receive antenna 2
h12
r2=h12s+n2
h1NRx
Receive antenna NRx
rnrx=h1nrxs+nnrx
~ h* s h* s ... h* s
r 11 1 12 2 1nRX nRX
Said:
h11 h12 ... h1nRx s h n h n ... h
2 2 2
* * *
nnRx diversity
11 1 12 2 1nRx
order nRx
signal after maximum ratio combining
1
Pe ~
Probability for errors SNR nRx
November 2012 | LTE Introduction | 44
45. TX Diversity: Space Time Coding
Fading on the air interface
data
The same signal is transmitted at differnet
antennas
space Aim: increase of S/N ratio
increase of throughput
s1 s2
*
S2 *
Alamouti Coding = diversity gain
time
approaches
s2 s1 RX diversity gain with MRRC!
Alamouti Coding -> benefit for mobile communications
November 2012 | LTE Introduction | 45
46. Space Time Block Coding according to Alamouti
(when no channel information at all available at transmitter)
d e1 h1 r1 h2 r2 h1 ² h2 ² d1 n'1
* *
From slide before:
d e 2 h2 r1 h1 r2 h1 ² h2 ² d 2 n'2
* *
Probability for errors (Alamouti) 1
Pe ~
SNR 2
Compare
with Rx diversity
Alamouti coding is full diversity gain
and full rate, but it only works for 1
Pe ~
2 antennas SNR nRx
(due to Alamouti matrix is orthogonal)
November 2012 | LTE Introduction | 46
47. MIMO Spatial Multiplexing
C=B*T*ld(1+S/N)
SISO:
Single Input
Single Output
Higher capacity without additional spectrum!
MIMO: S
C T B ld (1 ) ?
min( N T , N R )
i
i
i 1
N
Multiple Input i
Multiple Output
Increasing
capacity per cell
November 2012 | LTE Introduction | 47
48. Spatial multiplexing – capacity aspects
Ergodic mean capacity of a SISO channel calculated as:
h11
2
C EH {log 2 (1 h11 )}
r s h11 n
Received signal r with
sent signal s, channel h11 and
AWGN with σ=n
P
represents the signal to noise ratio SNR
2
at the receiver branch
Or simplified: S With B = bandwidth
C B * log 2 1 and S/N = signal to
N
noise ratio
November 2012 | LTE Introduction | 48
49. Spatial multiplexing – capacity aspects
Ergodic mean capacity of a MIMO channel is even worse
H
C EH l og 2 det I nR HH
nT
n1 n1
n2 n2
InR is an Identity matrix with size nT x nR nT nR
r sH n
HH is the Hermetian complex
Received signal r with
sent signal s, channel H and
AWGN with σ=n
November 2012 | LTE Introduction | 49
50. Spatial multiplexing – capacity aspects
Some theoretical ideas:
H
C EH l og 2 det I nR HH EH nR * log 2 1
nT
We increase to number of HH H
lim I nR
transmit antennas to ∞, and see: nT nT
So the result is, if the number of Tx antennas is infinity, the
capacity depends on the number of Rx antennas:
After this heavy mathematics the result: If we increase the
number of Tx and Rx antennas, we can increase the capacity!
November 2012 | LTE Introduction | 50
51. The MIMO promise
l Channel capacity grows linearly with antennas
Max Capacity ~ min(NTX, NRX)
l Assumptions
l Perfect channel knowledge
l Spatially uncorrelated fading
l Reality
l Imperfect channel knowledge
l Correlation ≠ 0 and rather unknown
November 2012 | LTE Introduction | 51
52. Spatial Multiplexing
Coding Fading on the air interface
data
data
Throughput: <200%
200%
100%
Spatial Multiplexing: We increase the throughput
but we also increase the interference!
November 2012 | LTE Introduction | 52
53. MIMO – capacity calculations, e.g. 2x2 MIMO
n1
h11
s1 r1
h12
n2
h21
s2 r2
h22
This results in the equations:
Or as matrix:
r1 = s1*h11 + s2*h21 + n1
r1 h11 h12 s1 n1
r2 = s2*h22 + s1*h12 + n2 r h * s n
2 21 h22 2 2
100%
General written as: r = s*H +n
To solve this equation, we have to know H
November 2012 | LTE Introduction | 53
54. Introduction – Channel Model II Correlation of
propagation
h11
pathes
h21
s1 r1
hMR1
h12
s2 h22 r2 estimates
Transmitter hMR2 Receiver
h1MT h2MT
NTx NRx
sNTx hMRMT rNRx
antennas antennas
s H r
Rank indicator
Capacity ~ min(NTX, NRX) → max. possible rank!
But effective rank depends on channel, i.e. the
correlation situation of H
November 2012 | LTE Introduction | 54
55. Spatial Multiplexing prerequisites
Decorrelation is achieved by:
l Decorrelated data content on each spatial stream difficult
l Large antenna spacing Channel
condition
l Environment with a lot of scatters near the antenna
(e.g. MS or indoor operation, but not BS)
Technical
l Precoding assist
But, also possible
that decorrelation
l Cyclic Delay Diversity is not given
November 2012 | LTE Introduction | 55
56. MIMO: channel interference + precoding
MIMO channel models: different ways to combat against
channel impact:
I.: Receiver cancels impact of channel
II.: Precoding by using codebook. Transmitter assists receiver in
cancellation of channel impact
III.: Precoding at transmitter side to cancel channel impact
November 2012 | LTE Introduction | 56
57. MIMO: Principle of linear equalizing
R = S*H + n
Transmitter will send reference signals or pilot sequence
to enable receiver to estimate H.
n H-1
Rx
s r ^
r
Tx H
LE
The receiver multiplies the signal r with the
Hermetian conjugate complex of the transmitting
function to eliminate the channel influence.
November 2012 | LTE Introduction | 57
58. Linear equalization – compute power increase
h11 H = h11
SISO: Equalizer has to estimate 1 channel
h11
h12 h11 h12
H=
h21 h22
h21 h22
2x2 MIMO: Equalizer has to estimate 4 channels
November 2012 | LTE Introduction | 58
59. transmission – reception model
noise
s + r
A H R
transmitter channel receiver
•Modulation, •detection,
•Power •estimation
•„precoding“, •Eliminating channel
•etc. Linear equalization impact
at receiver is not •etc.
very efficient, i.e.
noise can not be cancelled
November 2012 | LTE Introduction | 59
60. MIMO – work shift to transmitter
Channel Receiver
Transmitter
November 2012 | LTE Introduction | 60
61. MIMO Precoding in LTE (DL)
Spatial multiplexing – Code book for precoding
Code book for 2 Tx:
Codebook Number of layers
index
1 2 Additional multiplication of the
1 1 1 0
0 0
2 0 1
layer symbols with codebook
0 1 1 1 entry
1 1
2 1 1
1 1 1 1 1
2
2 1 2 j j
1 1
3 -
2 1
1 1
4 -
2 j
1 1
5 -
2 j
November 2012 | LTE Introduction | 61
62. MIMO precoding
precoding
Ant1
Ant2 t
+
1
2 1 ∑
t
+
1
precoding -1
1
∑=0
t
t
November 2012 | LTE Introduction | 62
63. MIMO – codebook based precoding
Precoding
codebook
noise
s + r
A H R
transmitter channel receiver
Precoding Matrix Identifier, PMI
Codebook based precoding creates
some kind of „beamforming light“
November 2012 | LTE Introduction | 63
64. MIMO: avoid inter-channel interference – future outlook
e.g. linear precoding:
V1,k Y=H*F*S+V
S Link adaptation
+
Transmitter H Space time
F receiver
xk +
yk
VM,k
Feedback about H
Idea: F adapts transmitted signal to current channel conditions
November 2012 | LTE Introduction | 64
65. MAS: „Dirty Paper“ Coding – future outlook
l Multiple Antenna Signal Processing: „Known Interference“
l Is like NO interference
l Analogy to writing on „dirty paper“ by changing ink color accordingly
„Known
„Known „Known „Known
Interference
Interference Interference Interference
is No
is No is No is No
Interference“
Interference“ Interference“ Interference“
November 2012 | LTE Introduction | 65
66. Spatial Multiplexing
Codeword Fading on the air interface
data
Codeword
data
Spatial Multiplexing: We like to distinguish the 2 useful
Propagation passes:
How to do that? => one idea is SVD
November 2012 | LTE Introduction | 66
67. Idea of Singular Value Decomposition
s1 MIMO r1
know
r=Hs+n s2 r2
channel H
Singular Value
Decomposition
~
s1 ~
r1
SISO
wanted ~ ~
s2 r2
~ ~ ~
r=Ds+n
channel D
November 2012 | LTE Introduction | 67
68. Singular Value Decomposition (SVD)
h11 h12
r= H s + n
h21 h22
h11 h0
d1 12
r= U H (V*)T s + n
h0 h22
21 d2
d1 0
(U*)T r= (U*)T U D (V*)T s + (U*)T n
0 d2
~ = (U*)T U d1 0 ~ ~
(U*)T r D (V*)T s + (U*)T n
0 d2
November 2012 | LTE Introduction | 68
69. Singular Value Decomposition (SVD)
r=Hs+n
H = U Σ (V*)T
U = [u1,...,un] eigenvectors of (H*)T H
V = [v1,...,vm] eigenvectors of H (H*)T
1 0 0
0 i eigenvalues of (H*)T H
0 0
2
0 0 3 singular values i i
0
0 ~ = (U*)T r
r
~
~= Σ s + n
r ~ ~ s = (V*)T s
~ = (U*)T n
n
November 2012 | LTE Introduction | 69
70. MIMO and singular value decomposition SVD
Real channel n1
h11
s1 r1
h12
n2
h21
s2 r2
h22
Channel model with SVD
n1
s1 σ1 r1
U Σ VH n2
s2 σ2 r2
SVD transforms channel into k parallel AWGN channels
November 2012 | LTE Introduction | 70
71. MIMO: Signal processing considerations
MIMO transmission can be expressed as
r = Hs+n which is, using SVD = UΣVHs+n
n1
s1 σ1 r1
V U Σ VH n2 UH
s2 σ2 r2
Imagine we do the following:
1.) Precoding at the transmitter:
Instead of transmitting s, the transmitter sends s = V*s
2.) Signal processing at the receiver
Multiply the received signal with UH, r = r*UH
So after signal processing the whole signal can be expressed as:
r =UH*(UΣVHVs+n)=UHU Σ VHVs+UHn = Σs+UHn
=InTnT =InTnT
November 2012 | LTE Introduction | 71
72. MIMO: limited channel feedback
Transmitter H Receiver
n1
s1 σ1 r1
V U Σ VH n2 UH
s2 σ2 r2
Idea 1: Rx sends feedback about full H to Tx.
-> but too complex,
-> big overhead
-> sensitive to noise and quantization effects
Idea 2: Tx does not need to know full H, only unitary matrix V
-> define a set of unitary matrices (codebook) and find one matrix in the codebook that
maximizes the capacity for the current channel H
-> these unitary matrices from the codebook approximate the singular vector structure
of the channel
=> Limited feedback is almost as good as ideal channel knowledge feedback
November 2012 | LTE Introduction | 72
73. Cyclic Delay Diversity, CDD
A2
A1 Amp
litud
D
e
Transmitter B
Delay Spread Time
Delay
Multipath propagation
precoding
+
+
precoding Time
No multipath propagation Delay
November 2012 | LTE Introduction | 73
74. „Open loop“ und „closed loop“ MIMO
Open loop (No channel knowledge at transmitter)
r Hs n Channel
Status, CSI
Rank indicator
Closed loop (With channel knowledge at transmitter
r HWs n Channel
Status, CSI
Rank indicator
Adaptive Precoding matrix („Pre-equalisation“)
Feedback from receiver needed (closed loop)
November 2012 | LTE Introduction | 74
75. MIMO transmission modes
Transmission mode2
Transmission mode3
TX diversity
Transmission mode1 Open-loop spatial
SISO multiplexing
7 transmission
Transmission mode4 Transmission mode7
Closed-loop spatial modes are SISO, port 5
multiplexing defined = beamforming in TDD
Transmission mode6
Transmission mode5 Closed-loop
Multi-User MIMO spatial multiplexing,
using 1 layer
Transmission mode is given by higher layer IE: AntennaInfo
November 2012 | LTE Introduction | 75
76. MIMO transmission modes
the classic:
1Tx + 1RX
Transmission mode1
antenna
SISO
PDCCH indication via
DCI format 1 or 1A
PDSCH transmission via
single antenna port 0 No feedback regarding
antenna selection or
precoding needed
November 2012 | LTE Introduction | 76
77. MIMO transmission modes
Transmission mode 2
Transmit
diversity PDCCH indication via
DCI format 1 or 1A
Codeword is sent
redundantly over several
streams
1 codeword
PDSCH transmission via
2 Or 4 antenna ports No feedback regarding
antenna selection or
precoding needed
November 2012 | LTE Introduction | 77
78. MIMO transmission modes No feedback regarding
antenna selection or
Transmission mode 3 precoding needed
Transmit diversity or Open loop
spatial multiplexing
PDCCH indication via
DCI format 1A
1 codeword PDSCH transmission
Via 2 or 4 antenna ports
PDCCH indication via
DCI format 2A
1 2
codeword codewords
PMI feedback
PDSCH spatial multiplexing PDSCH spatial multiplexing, using CDD
with 1 layer
November 2012 | LTE Introduction | 78
79. MIMO transmission modes Closed loop MIMO =
UE feedback needed regarding
Transmission mode 4 precoding and antenna
Transmit diversity or Closed loop selection
spatial multiplexing
PDCCH indication via
DCI format 1A
1 codeword PDSCH transmission
Via 2 or 4 antenna ports
PDCCH indication via
DCI format 2
precoding
precoding
1 2
codeword codewords
PMI feedback PMI feedback
PDSCH spatial multiplexing PDSCH spatial multiplexing
with 1 layer
November 2012 | LTE Introduction | 79
80. MIMO transmission modes
Transmission mode 5
Transmit diversity or
Multi User MIMO
PDCCH indication via
DCI format 1A
1 codeword PDSCH transmission
Via 2 or 4 antenna ports
PUSCH
UE1
Codeword
PDCCH indication
via DCI format 1D
UE2 PDSCH multiplexing to several UEs.
Codeword PUSCH multiplexing in Uplink
November 2012 | LTE Introduction | 80
81. MIMO transmission modes Closed loop MIMO =
UE feedback needed regarding
Transmission mode 6 precoding and antenna
Transmit diversity or Closed loop selection
spatial multiplexing with 1 layer
PDCCH indication via
DCI format 1A
1 codeword PDSCH transmission
via 2 or 4 antenna ports
PDCCH indication via
DCI format 1B
Codeword is split into
1
streams, both streams have
codeword
to be combined
feedback
PDSCH spatial multiplexing, only 1 codeword
November 2012 | LTE Introduction | 81
82. MIMO transmission modes
Transmission mode 7
Transmit diversity or beamforming
PDCCH indication via
DCI format 1A
1 codeword PDSCH transmission
via 1, 2 or 4 antenna ports
PDCCH indication via
DCI format 1
1
codeword
PDSCH sent over antenna port 5 = beamforming
November 2012 | LTE Introduction | 82
83. Beamforming
Closed loop precoded
Adaptive Beamforming beamforming
•Classic way •Kind of MISO with channel
knowledge at transmitter
•Antenna weights to adjust beam
•Precoding based on feedback
•Directional characteristics
•No specific antenna
•Specific antenna array geometrie
array geometrie
•Dedicated pilots required •Common pilots are sufficient
November 2012 | LTE Introduction | 83
84. Spatial multiplexing vs beamforming
Spatial multiplexing increases throughput, but looses coverage
November 2012 | LTE Introduction | 84
85. Spatial multiplexing vs beamforming
Beamforming increases coverage
November 2012 | LTE Introduction | 85
86. Basic OFDM parameter
LTE
1
f 15 kHz
T
Fs N FFT f
N FFT
Fs 3.84Mcps
256
f
NFFT 2048
Coded symbol rate= R
Sub-carrier CP
S/P Mapping IFFT insertion
N TX Data symbols
Size-NFFT
November 2012 | LTE Introduction | 86
87. LTE Downlink:
Downlink slot and (sub)frame structure
Symbol time, or number of symbols per time slot is not fixed
One radio frame, Tf = 307200Ts=10 ms
One slot, Tslot = 15360Ts = 0.5 ms
#0 #1 #2 #3 #18 #19
One subframe
We talk about 1 slot, but the minimum resource is 1 subframe = 2 slots !!!!!
Ts 1 15000 2048
Ts = 32.522 ns
November 2012 | LTE Introduction | 87
88. Resource block definition
1 slot = 0,5msec
Resource block
=6 or 7 symbols
In 12 subcarriers
12 subcarriers
Resource element
DL UL
N symb or N symb
6 or 7,
Depending on
cyclic prefix
November 2012 | LTE Introduction | 88
89. LTE Downlink
OFDMA time-frequency multiplexing
frequency
QPSK, 16QAM or 64QAM modulation
UE4
1 resource block =
180 kHz = 12 subcarriers UE5
UE3
UE2
UE6
Subcarrier spacing = 15 kHz time
UE1
1 subframe =
*TTI = transmission time interval 1 slot = 0.5 ms = 1 ms= 1 TTI*=
** For normal cyclic prefix duration 7 OFDM symbols** 1 resource block pair
November 2012 | LTE Introduction | 89
90. LTE: new physical channels for data and control
Physical Control Format Indicator Channel PCFICH:
Indicates Format of PDCCH
Physical Downlink Control Channel PDCCH:
Downlink and uplink scheduling decisions
Physical Downlink Shared Channel PDSCH: Downlink data
Physical Hybrid ARQ Indicator Channel PHICH:
ACK/NACK for uplink packets
Physical Uplink Shared Channel PUSCH: Uplink data
Physical Uplink Control Channel PUCCH:
ACK/NACK for downlink packets, scheduling requests, channel quality info
November 2012 | LTE Introduction | 90
91. LTE Downlink: FDD channel mapping example
Subcarrier #0 RB
November 2012 | LTE Introduction | 91
92. LTE – spectrum flexibility
l LTE physical layer supports any bandwidth from 1.4 MHz
to 20 MHz in steps of 180 kHz (resource block)
l Current LTE specification supports only a subset of 6
different system bandwidths
l All UEs must support the maximum bandwidth of 20 MHz
Channel Bandwidth [MHz]
Channel Transmission Bandwidth Configuration [RB]
bandwidth Transmission
BWChannel 1.4 3 5 10 15 20 Bandwidth [RB]
Channel edge
Channel edge
[MHz] Resource block
FDD and
TDD mode 6 15 25 50 75 100
number of resource blocks
Active Resource Blocks DC carrier (downlink only)
November 2012 | LTE Introduction | 92
93. LTE Downlink:
baseband signal generation
code words layers antenna ports
Modulation OFDM OFDM signal
Scrambling
Mapper Mapper generation
Layer
Precoding
Mapper
Modulation OFDM OFDM signal
Scrambling
Mapper Mapper generation
1 stream =
several
Avoid QPSK For MIMO Weighting 1 OFDM subcarriers,
constant 16 QAM Split into data symbol per based on
sequences 64 QAM Several streams for stream Physical
streams if MIMO ressource
needed blocks
November 2012 | LTE Introduction | 93
94. Adaptive modulation and coding
Transportation block size
User data FEC
Flexible ratio between data and FEC = adaptive coding
November 2012 | LTE Introduction | 94
96. Automatic repeat request, latency aspects
•Transport block size = amount of
data bits (excluding redundancy!)
•TTI, Transmit Time Interval = time
duration for transmitting 1 transport
block
Transport block
Round
Trip
Time
ACK/NACK
Network UE
Immediate acknowledged or non-acknowledged
feedback of data transmission
November 2012 | LTE Introduction | 96
97. HARQ principle: Stop and Wait
Δt = Round trip time
Tx Data Data Data Data Data Data Data Data Data Data
ACK/NACK
Demodulate, decode, descramble,
Rx FFT operation, check CRC, etc.
process
Processing time for receiver
Described as 1 HARQ process
November 2012 | LTE Introduction | 97
98. HARQ principle: Multitasking
Δt = Round trip time
Tx Data Data Data Data Data Data Data Data Data Data
ACK/NACK
Demodulate, decode, descramble,
Rx FFT operation, check CRC, etc.
process
ACK/NACK
Processing time for receiver
Rx Demodulate, decode, descramble,
process FFT operation, check CRC, etc.
t
Described as 1 HARQ process
November 2012 | LTE Introduction | 98
100. HARQ principle: Soft combining
lT i is a e am l o h n e co i g
Reception of first transportation block.
Unfortunately containing transmission errors
November 2012 | LTE Introduction | 100
101. HARQ principle: Soft combining
l hi i n x m le f cha n l c ing
Reception of retransmitted
transportation block.
Still containing transmission errors
November 2012 | LTE Introduction | 101
102. HARQ principle: Soft combining
1st transmission with puncturing scheme P1
l T i is a e am l o h n e co i g
2nd transmission with puncturing scheme P2
l hi i n x m le f cha n l c ing
Soft Combining = Σ of transmission 1 and 2
l Thi is an exam le of channel co ing
Final decoding
lThis is an example of channel coding
November 2012 | LTE Introduction | 102
103. Hybrid ARQ
Chase Combining = identical retransmission
Turbo Encoder output (36 bits)
Systematic Bits
Parity 1
Parity 2
Transmitted Bit Rate Matching to 16 bits (Puncturing)
Original Transmission Retransmission
Systematic Bits
Parity 1
Parity 2
Punctured Bit Chase Combining at receiver
Systematic Bits
Parity 1
Parity 2
November 2012 | LTE Introduction | 103
105. LTE Physical Layer:
SC-FDMA in uplink
Single Carrier Frequency Division
Multiple Access
November 2012 | LTE Introduction | 105
106. LTE Uplink:
How to generate an SC-FDMA signal in theory?
Coded symbol rate= R
Sub-carrier CP
DFT Mapping IFFT insertion
NTX symbols
Size-NTX Size-NFFT
LTE provides QPSK,16QAM, and 64QAM as uplink modulation schemes
DFT is first applied to block of NTX modulated data symbols to transform them into
frequency domain
Sub-carrier mapping allows flexible allocation of signal to available sub-carriers
IFFT and cyclic prefix (CP) insertion as in OFDM
Each subcarrier carries a portion of superposed DFT spread data symbols
Can also be seen as “pre-coded OFDM” or “DFT-spread OFDM”
November 2012 | LTE Introduction | 106
107. LTE Uplink:
How does the SC-FDMA signal look like?
In principle similar to OFDMA, BUT:
In OFDMA, each sub-carrier only carries information related to one specific symbol
In SC-FDMA, each sub-carrier contains information of ALL transmitted symbols
November 2012 | LTE Introduction | 107
108. LTE uplink
SC-FDMA time-frequency multiplexing
1 resource block =
180 kHz = 12 subcarriers Subcarrier spacing = 15 kHz
frequency
UE1 UE2 UE3
1 slot = 0.5 ms =
7 SC-FDMA symbols**
1 subframe =
1 ms= 1 TTI*
UE4 UE5 UE6
*TTI = transmission time interval
** For normal cyclic prefix duration
time QPSK, 16QAM or 64QAM modulation
November 2012 | LTE Introduction | 108
109. LTE Uplink:
baseband signal generation
UE specific
Scrambling code
Modulation Transform Resource SC-FDMA
Scrambling element mapper
mapper precoder signal gen.
Mapping on
physical 1 stream =
Discrete
Ressource, several
Avoid QPSK Fourier
i.e. subcarriers,
constant 16 QAM Transform
subcarriers based on
sequences 64 QAM not used for Physical
(optional) reference ressource
signals blocks
November 2012 | LTE Introduction | 109