Open Loop versus Closed Loop
Performance in LTE
Lutz Schönerstedt (Presenter), Andreas Weber, Michael Ohm, Thorsten Wild
Bell Labs Germany, Stuttgart
12.2.2009 (VDE/ITG-Fachgruppe 5.2.4 at ComNets RWTH Aachen)
Agenda
1. Open And Closed Loop Transmission in LTE
2. Physical Layer Results
3. System Level Performance
4. Conclusion
1
Open And Closed Loop Transmission in LTE
Pilot Feedback: PMI, CQI, Rank
SFBC and PARC, TX-Diversity (Closed Loop) and PSRC
3 | Presentation Title | Month 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Pilot Feedback
eNodeB
UE
Uplink feedback: PMI, CQI, Rank
Comment
Freq.
Resolution
Available in
open loop
Available in
closed loop
PMI
Prefered
precoding
matrix indicator
Index of best Tx
weight
Subband
-
X
CQI
Channel quality
indicator
Supported
transport format
Subband
X
X
No. of spatial
streams supported
Full band
X
X
Rank
4 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Downlink MIMO Modes
LTE-MIMO
2 x 2 and 4 x 2 (TX x RX)
Closed loop
Open loop
Single stream
Multi stream
Rank 1
Rank 2-4
OL
Beamforming
OL Tx Diversity
Codebook-based
linear precoding
SFBC
CDD
PARC
Space Frequency
Block Coding
Cyclic Delay
Diversity
(Per Antenna
Rate Control)
5 | Open Loop versus Closed Loop Performance in LTE| February 2009
Single & multi stream
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Rank 1: CL Tx Diversity
Rank 2-4: PSRC (Per
Stream Rate Control)
Comparing Downlink MIMO Modes
Single Stream
OL SFBC
Uses diversity by orthogonally transmitting one data stream over two transmit
antennas, to reduce dynamic of the received power.
CL Tx Diversity
Tries to maximize the received power at the mobile by applying precoding (based on
PMI feedback). Sends correlated symbols on transmit antennas.
Dual Stream
OL PARC
Uses diversity to transmit two data streams over two transmit antennas.
CL PSRC
Uses diversity to transmit two data streams over two transmit antennas.
Tries to maximize the received signal quality of the data streams at the mobile by
applying precoding (based on PMI feedback).
6 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Linear Precoding (for PARC, PSRC and CL Tx Diversity)
Complex linear transmit antenna weights
Codebook
index
Distributes data streams over the antennas
Number of layers υ
1
2
0
1 1

2 1
1
1 1
 
2 −1
1 1 0


2 0 1
1 1 1 


2 1 −1
2
1 1
 
2  j
1 1 1 


2  j − j
3
1 1
 
2 − j 
-
2 Tx with 2 layers example for OFDM:
R8 codebook for 2 Tx
(3GPP TS 36.211)
7 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
2
Physical Layer Results
Open Loop Single Layer
8 | Presentation Title | Month 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
SFBC 2x1, 2x2 versus Single Antenna 1x1, 1x2
Legend:
0
VehA, 120km/h, ρ =0.1, ρ =0.1, QPSK, R=1/3
tx
rx
10
SFBC, 1RX: SFBC 2x1
SFBC, 2Rx: SFBC 2x2
BLER
SISO: 1x1
SIMO: 1x2
-1
10
Gain of SFBC at BLER = 0.1:
SFBC, 2 Rx
SFBC, 1 Rx
SIS O
SFBC 2x2 is 1.1dB better than 1x2.
SIMO, MRC
-2
SFBC 2x1 is 1.8dB better than 1x1.
10
-10
-5
0
SNR [dB]
9 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
5
10
3
System Level Performance
Open Loop and Closed Loop Single/Dual Layer
10 | Presentation Title | Month 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Cell border throughput over spectral efficiency 3km/h
Simulation Assumptions:
ISD 500m
46dBm per Antenna
10MHz bandwidth
Round Robin
Scheduler
C ell B order T hroug hput [kbits /s ]
210 UEs in 21 sectors
300
280
260
240
S ing le Antenna 3 km/h
S F B C 3 k m/h
220
S F B C _P AR C 3 km/h
T x D iv 3 km/h
200
T x D iv _P S R C 3 k m/h
Single Antenna 1x2
SFBC 2x2
180
0.95
1.00
1.05
1.10
1.15
S pec tral E ffic ienc y [bits /s /Hz ]
CL Tx Diversity 2x2
11 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
1.20
1.25
Cumulative Probability of SINR
1.0
C umulative P robability
0.9
0.8
S in g le A n ten n a
0.7
SFBC
C L T X D iv
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-20
-10
0
10
20
P er S u bc arrier S INR meas u red at D ec oder In pu t [dB ]
12 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
30
Cell border throughput over spectral efficiency 250km/h
C ell B order T hroug hput [k bits /s ]
150
S ing le Antenna 250 km/h
S F B C 250 k m/h
T x D iv 250 km/h
145
140
135
0.5
0.55
0.6
0.65
S pec tral E ffic ienc y [bits /s /Hz ]
13 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
0.7
SFBC versus CL Tx Diversity
220
SFBC
SFBC
SFBC
SFBC
SFBC
T x D iv
T x D iv
T x D iv
T x D iv
T x D iv
C ell B order T hroug h put [kbits /s ]
210
200
190
180
170
160
30 km/h
60 km/h
90 km/h
120 k m/h
250 k m/h
30 k m/h
60 k m/h
90 k m/h
120 k m/h
250 k m/h
150
140
130
120
0.5
0.6
0.7
S pec tral E ffic ienc y [bits /s /Hz ]
14 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
0.8
4
Conclusion
15 | Presentation Title | Month 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
Conclusion
SFBC gains from physical layer simulation couldn’t be retrieved in system level
simulation.
Especially at low speed, a transmission technique with higher channel quality
variance (with the same mean quality) handles more throughput. This is due to
the non linear mapping of channel quality and throughput.
Frequency selective schedulers, which can take advantage of situations with
high SINR, promise to improve system performance even more.
With a round robin scheduler and velocities higher than 30 km/h, SFBC
becomes attractive as a fallback mode for closed loop transmit diversity.
16 | Open Loop versus Closed Loop Performance in LTE| February 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX
www.alcatel-lucent.com
www.alcatel-lucent.com
17 | Presentation Title | Month 2009
All Rights Reserved © Alcatel-Lucent 2009, XXXXX