Professional Documents
Culture Documents
Stefan Schwarz
sschwarz@nt.tuwien.ac.at
March 27, 2014
institute of
telecommunications
y Slide 2 / 31 y y Contentsy
Contents
Application Scenario
Conclusions
Stefan Schwarz
y Slide 3 / 31 y y Overview of CoMP in LTEy
CoMP Motivation
0
2010 2011 2012 2013 2014 2015 2016 2017 2018
Year
Estimated growth of mobile traffic (1 Exabyte = 1018 bytes); Ericsson traffic exploration tool [Ericsson, 2013b]
Stefan Schwarz
y Slide 4 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 4 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 4 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 5 / 31 y y Overview of CoMP in LTEy
CoMP Principles
I Definition of CoMP:
I Coordinated transmission/reception of data among several
transmission/reception points to reduce or even exploit interference
I Transmission/reception points:
base stations, relays, access points, remote radio heads, user equipments
Stefan Schwarz
y Slide 5 / 31 y y Overview of CoMP in LTEy
CoMP Principles
I Definition of CoMP:
I Coordinated transmission/reception of data among several
transmission/reception points to reduce or even exploit interference
I Transmission/reception points:
base stations, relays, access points, remote radio heads, user equipments
Stefan Schwarz
y Slide 6 / 31 y y Overview of CoMP in LTEy
3GPP Time-Line
2008 2010 2012 2014 2016
Rel 8 NOW
LTE
Rel 9
LTE Enhancements
Rel 10
LTE-A
CoMP
Stefan Schwarz
y Slide 6 / 31 y y Overview of CoMP in LTEy
3GPP Time-Line
2008 2010 2012 2014 2016
Rel 8 NOW
LTE
Rel 9
LTE Enhancements
Rel 10
LTE-A
CoMP
Rel 11
LTE-A Enhancements
CoMP
Stefan Schwarz
y Slide 6 / 31 y y Overview of CoMP in LTEy
3GPP Time-Line
2008 2010 2012 2014 2016
Rel 8 NOW
LTE
Rel 9
LTE Enhancements
Rel 10
LTE-A
CoMP
Rel 11
LTE-A Enhancements
CoMP
Rel 12
Imperfect Backhaul
Stefan Schwarz
y Slide 7 / 31 y y Overview of CoMP in LTEy
I Coordinated beamforming:
I Spatial interference mitigation
I Signal to leakage and noise ratio (SLNR) Coordinated
scheduling
[Sadek et al., 2007]
I Advantage: good trade-off (CSI only)
I Joint transmission:
I Exploitation of interference
Base station User
I Distributed antenna system (DAS) X2 interface or low-latency high-bandwidth dedicated connection
I Advantage: potentially highest performance
I Disadvantage: overhead (CSI and data)
Stefan Schwarz
y Slide 7 / 31 y y Overview of CoMP in LTEy
I Coordinated beamforming:
I Spatial interference mitigation
I Signal to leakage and noise ratio (SLNR) Coordinated
scheduling
[Sadek et al., 2007]
I Advantage: good trade-off (CSI only)
Coordinated
beamforming
I Joint transmission:
I Exploitation of interference
Base station User
I Distributed antenna system (DAS) X2 interface or low-latency high-bandwidth dedicated connection
I Advantage: potentially highest performance
I Disadvantage: overhead (CSI and data)
Stefan Schwarz
y Slide 7 / 31 y y Overview of CoMP in LTEy
I Coordinated beamforming:
I Spatial interference mitigation
I Signal to leakage and noise ratio (SLNR) Coordinated
scheduling
[Sadek et al., 2007]
I Advantage: good trade-off (CSI only)
Coordinated
beamforming
I Joint transmission:
I Exploitation of interference
Base station User
I Distributed antenna system (DAS) X2 interface or low-latency high-bandwidth dedicated connection
I Advantage: potentially highest performance
I Disadvantage: overhead (CSI and data)
Stefan Schwarz
y Slide 8 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 8 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 8 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 8 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 8 / 31 y y Overview of CoMP in LTEy
Stefan Schwarz
y Slide 9 / 31 y y Overview of CoMP in LTEy
H2
H1
H3
CSI for H1 H2 H3
Stefan Schwarz
y Slide 10 / 31 y y Multi-User MIMO Transmissiony
Contents
Application Scenario
Conclusions
Stefan Schwarz
y Slide 11 / 31 y y Multi-User MIMO Transmissiony
Stefan Schwarz
y Slide 11 / 31 y y Multi-User MIMO Transmissiony
yu = GH H H H
Fs xs + GH
X
u Hu Fu xu + Gu Hu u zu (1)
s∈S
| {z } | {z }
intended signal s6=u noise
| {z }
interference
Stefan Schwarz
y Slide 12 / 31 y y Multi-User MIMO Transmissiony
I Single-user MIMO:
I Number of parallel users: S = 1
I Questions:
I How to select the users S?
Stefan Schwarz
y Slide 12 / 31 y y Multi-User MIMO Transmissiony
I Single-user MIMO:
I Number of parallel users: S = 1
I Questions:
I How to select the users S?
Stefan Schwarz
y Slide 13 / 31 y y Multi-User MIMO Transmissiony
Transceiver Design
I Block-diagonalization precoding
I Iterative joint optimization, e.g., based on MMSE
criteria [Shi et al., 2008]
Stefan Schwarz
y Slide 13 / 31 y y Multi-User MIMO Transmissiony
Transceiver Design
I Block-diagonalization precoding
I Iterative joint optimization, e.g., based on MMSE
criteria [Shi et al., 2008]
Stefan Schwarz
y Slide 14 / 31 y y Multi-User MIMO Transmissiony
Heff
u = Hu Gu ∈ C
Nt ×L
versus Hu ∈ CNt ×Nr (2)
Stefan Schwarz
y Slide 15 / 31 y y Multi-User MIMO Transmissiony
S
yu = Heff
H
Fu xu + Heff
H X
u u Fs x s + GH
u zu
s=1
s6=u
Heff
H
s Fu = 0, ∀s, u ∈ S and s 6= u, (3)
H
rank Heff
u Fu = L, ∀u ∈ S (4)
Stefan Schwarz
y Slide 15 / 31 y y Multi-User MIMO Transmissiony
S
yu = Heff
H
Fu xu + Heff
H X
u u Fs x s + GH
u zu
s=1
s6=u
Heff
H
s Fu = 0, ∀s, u ∈ S and s 6= u, (3)
H
rank Heff
u Fu = L, ∀u ∈ S (4)
h iH
H̄u = Heff eff eff eff
1 , . . . , Hu−1 , Hu+1 , . . . , HS ∈ C(S−1)L×Nt ,
Stefan Schwarz
y Slide 15 / 31 y y Multi-User MIMO Transmissiony
S
yu = Heff
H
Fu xu + Heff
H X
u u Fs x s + GH
u zu
s=1
s6=u
Heff
H
s Fu = 0, ∀s, u ∈ S and s 6= u, (3)
H
rank Heff
u Fu = L, ∀u ∈ S (4)
Fu ∈ null H̄u ,
iH
h (5)
H̄u = Heff eff eff eff
1 , . . . , Hu−1 , Hu+1 , . . . , HS ∈ C(S−1)L×Nt ,
rank (Fu ) = L (6)
Stefan Schwarz
y Slide 16 / 31 y y Multi-User MIMO Transmissiony
Nt
⇒ Nt − (S − 1)L = L ⇒ S = (7)
L
yu = heff
H
fu xu + heff
H X
u u fs xs + gH
u zu (8)
s
Stefan Schwarz
y Slide 16 / 31 y y Multi-User MIMO Transmissiony
Nt
⇒ Nt − (S − 1)L = L ⇒ S = (7)
L
yu = heff
H
fu xu + heff
H X
u u fs xs + gH
u zu (8)
s
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
f1
h3 I Intended signal power:
2
P1 = hH
1 f1
y
I ⇒ Orthogonal user selection
x h2
P
RZF ∝ Nt log 1 + log U ∝ RDPC
Nt
y1 = hH H H
1 f1 x1 +h1 f2 x2 + h1 f3 x3
y2 = hH H H
2 f1 x1 +h2 f2 x2 + h2 f3 x3
[Yoo and Goldsmith, 2006,
Boccardi and Huang, 2007]
y3 = hH H
3 f1 x1 +h3 f2 x2 + hH
3 f3 x3
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
f1
h3 I Intended signal power:
2
P1 = hH
1 f1
y
I ⇒ Orthogonal user selection
x h2
P
RZF ∝ Nt log 1 + log U ∝ RDPC
Nt
y1 = hH H H
1 f1 x1 +h1 f2 x2 + h1 f3 x3
y2 = hH H H
2 f1 x1 +h2 f2 x2 + h2 f3 x3
[Yoo and Goldsmith, 2006,
Boccardi and Huang, 2007]
y3 = hH H
3 f1 x1 +h3 f2 x2 + hH
3 f3 x3
Stefan Schwarz
y Slide 17 / 31 y y Multi-User MIMO Transmissiony
f1
h3 I Intended signal power:
2
P1 = hH
1 f1
y
I ⇒ Orthogonal user selection
x h2
P
RZF ∝ Nt log 1 + log U ∝ RDPC
Nt
y1 = hH H H
1 f1 x1 +h1 f2 x2 + h1 f3 x3
y2 = hH H H
2 f1 x1 +h2 f2 x2 + h2 f3 x3
[Yoo and Goldsmith, 2006,
Boccardi and Huang, 2007]
y3 = hH H
3 f1 x1 +h3 f2 x2 + hH
3 f3 x3
Stefan Schwarz
y Slide 18 / 31 y y Multi-User MIMO Transmissiony
h iH
Fu ∈ null H̄u , H̄u = Heff eff eff eff
1 , . . . , Hu−1 , Hu+1 , . . . , HS
I What channel state information does the base station need, i.e., what
feedback information do the users have to provide?
I Notice, Heff can be replaced with any matrix spanning the same subspace
j
Heff
j ≡ H̃j ∈ C
Nt ×L
⇔ span Heff
j = span H̃j , (11)
Heff
H
j Fu = 0 ⇔ H̃H
j Fu = 0 (12)
⇒ the users have to convey span Heff
j
to the base station
Stefan Schwarz
y Slide 18 / 31 y y Multi-User MIMO Transmissiony
h iH
Fu ∈ null H̄u , H̄u = Heff eff eff eff
1 , . . . , Hu−1 , Hu+1 , . . . , HS
I What channel state information does the base station need, i.e., what
feedback information do the users have to provide?
I Notice, Heff can be replaced with any matrix spanning the same subspace
j
Heff
j ≡ H̃j ∈ C
Nt ×L
⇔ span Heff
j = span H̃j , (11)
Heff
H
j Fu = 0 ⇔ H̃H
j Fu = 0 (12)
⇒ the users have to convey span Heff
j
to the base station
Stefan Schwarz
y Slide 19 / 31 y y Multi-User MIMO Transmissiony
span Heff
j = span H̃j , H̃H
j H̃j = IL (13)
1
[Ravindran and Jindal, 2008, Schwarz et al., 2013a, Schwarz et al., 2013b]
Stefan Schwarz
y Slide 19 / 31 y y Multi-User MIMO Transmissiony
span Heff
j = span H̃j , H̃H
j H̃j = IL (13)
1
[Ravindran and Jindal, 2008, Schwarz et al., 2013a, Schwarz et al., 2013b]
Stefan Schwarz
y Slide 20 / 31 y y Multi-User MIMO Transmissiony
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Stefan Schwarz
y Slide 21 / 31 y y Multi-User MIMO Transmissiony
Hu = Uu Σu VH
(MET)
u ⇒ Gu = Vu (:, 1 : L), (16)
Heff = Uu (:, 1 : L) diag σ1,u , . . . , σL,u
u (17)
Theorem (MET)
S users are served with L spatial streams each over i.i.d. Rayleigh fading channels,
using b bits to quantize and feedback the channel state information.
L
(`)
X
RMET − RMET-Quant ≤ log2 1 + ρ cMET DMET , (18)
`=1
− L(N b−L) P
DMET ∝ 2 t , ρ= , ρdB = 10 log10 (ρ) (19)
σz2 S L
∂b
MET feedback bit-scaling for constant rate offset: ∝ L (Nt − L) (20)
∂ρdB
Hu = Uu Σu VH
(MET)
u ⇒ Gu = Vu (:, 1 : L), (16)
Heff = Uu (:, 1 : L) diag σ1,u , . . . , σL,u
u (17)
Theorem (MET)
S users are served with L spatial streams each over i.i.d. Rayleigh fading channels,
using b bits to quantize and feedback the channel state information.
L
(`)
X
RMET − RMET-Quant ≤ log2 1 + ρ cMET DMET , (18)
`=1
− L(N b−L) P
DMET ∝ 2 t , ρ= , ρdB = 10 log10 (ρ) (19)
σz2 S L
∂b
MET feedback bit-scaling for constant rate offset: ∝ L (Nt − L) (20)
∂ρdB
n o
= argmin d2c Heff 2
(SQBC) (SQBC)
u , Qj = argmin dc Hu G, Qj
Gu , Ĥu (21)
G,Qj ∈Qu G,Qj ∈Qu
Theorem (SQBC)
S users are served with L spatial streams each over i.i.d. Rayleigh fading channels,
using b bits to quantize and feedback the channel state information.
b
− L(N −N
(L) (L,N ) (L,N )
r
RBD − RSQBC ≤ L log2 (1 + ρ cSQBC DSQBC ) + dSQBCr , DSQBC ∝ 2 t r) (22)
∂b
SQBC feedback bit-scaling: ∝ L (Nt − Nr ), MET: L (Nt − L) (23)
∂ρdB
n o
= argmin d2c Heff 2
(SQBC) (SQBC)
u , Qj = argmin dc Hu G, Qj
Gu , Ĥu (21)
G,Qj ∈Qu G,Qj ∈Qu
Theorem (SQBC)
S users are served with L spatial streams each over i.i.d. Rayleigh fading channels,
using b bits to quantize and feedback the channel state information.
b
− L(N −N
(L) (L,N ) (L,N )
r
RBD − RSQBC ≤ L log2 (1 + ρ cSQBC DSQBC ) + dSQBCr , DSQBC ∝ 2 t r) (22)
∂b
SQBC feedback bit-scaling: ∝ L (Nt − Nr ), MET: L (Nt − L) (23)
∂ρdB
40
35
30
25
20
15
10
5
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Achievable rate with perfect CSIT
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
40
35
30
25
20
15
10
5
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Achievable rate with perfect CSIT
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
50
95% confidence interval 6xN_flat_2_streams
Nr = 4
MET
45 Nr = 2
SQBC
Achievable sum rate [bits/s/Hz]
40 Nr = 4
35
30
25
20
15
10
5
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Achievable rate with perfect CSIT
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
40
Nr = 5
35
30
(L,Nr)
25 dSQBC
20
15
10
5
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Achievable rate with perfect CSIT
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
40 70
Nr = 5
35
60
30
(L,Nr)
50
25 dSQBC
40
20
30
15
10 20
5 10
0 0
0 3 6 9 12 15 18 21 24 27 30 10 12 14 16 18 20 22 24 26 28 30
SNR [dB] SNR [dB]
Achievable rate with perfect CSIT Feedback bit-scaling to achieve a loss of 1 bit/s/Hz
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
40 70
Nr = 5
35
60 Nr = 3
30
(L,Nr)
50
25 dSQBC
40
20
30
15
10 20
5 10
0 0
0 3 6 9 12 15 18 21 24 27 30 10 12 14 16 18 20 22 24 26 28 30
SNR [dB] SNR [dB]
Achievable rate with perfect CSIT Feedback bit-scaling to achieve a loss of 1 bit/s/Hz
Stefan Schwarz
y Slide 23 / 31 y y Multi-User MIMO Transmissiony
40 70
Nr = 5
35
60
30
(L,Nr)
50
25 dSQBC
40
20
30 L (Nt-Nr)
15
10 20
Nr = 5
5 10
0 0
0 3 6 9 12 15 18 21 24 27 30 10 12 14 16 18 20 22 24 26 28 30
SNR [dB] SNR [dB]
Achievable rate with perfect CSIT Feedback bit-scaling to achieve a loss of 1 bit/s/Hz
Stefan Schwarz
y Slide 24 / 31 y y Multi-User MIMO Transmissiony
35
30
25
20
15
10
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
35
30
Nr = 4
25
20
15
10
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
35
30
Nr = 4
25
20
15
Nr = 2
10
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
35
30
Nr = 4
25
20
15
Nr = 2
10
0
0 3 6 9 12 15 18 21 24 27 30
SNR [dB]
Contents
Application Scenario
Conclusions
Stefan Schwarz
y Slide 26 / 31 y y Application Scenarioy
Base station Remote radio unit User Low-latency high-bandwidth dedicated connection
Base station Remote radio unit User Low-latency high-bandwidth dedicated connection
Base station Remote radio unit User Low-latency high-bandwidth dedicated connection
Stefan Schwarz
y Slide 28 / 31 y y Application Scenarioy
21 predictive quant. 21
18 18
memoryless quant.
15 15
12 12
9 9
6 memoryless quant. 6
3 3
0 0
0 4 8 12 16 20 24 28 32 0 4 8 12 16 20 24 28 32
Number of users per sector Number of users per sector
Stefan Schwarz
y Slide 29 / 31 y y Application Scenarioy
15
12
6
Macro only
3 predictive quant.
memoryless quant.
0
0 2 4 6 8 10
Number of feedback bits per user
Stefan Schwarz
y Slide 29 / 31 y y Application Scenarioy
9 12
9
6
Macro only 6
3 predictive quant. Macro only
3 predictive quant.
memoryless quant. memoryless quant.
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Number of feedback bits per user Number of feedback bits per user
Stefan Schwarz
y Slide 30 / 31 y y Conclusionsy
Contents
Application Scenario
Conclusions
Stefan Schwarz
y Slide 31 / 31 y y Conclusionsy
Stefan Schwarz
y Slide 31 / 31 y y Conclusionsy
Stefan Schwarz
Coordinated Multi-Point (CoMP) in LTE
Wireless Communications Seminar
Stefan Schwarz
sschwarz@nt.tuwien.ac.at
March 27, 2014
institute of
telecommunications
y Slide 33 / 31 y y Referencesy
References I
Andrews, J. (2013).
Seven ways that HetNets are a cellular paradigm shift.
IEEE Communications Magazine, 51(3):136–144.
Costa, M. (1983).
Writing on dirty paper (corresp.).
IEEE Transactions on Information Theory, 29(3):439 – 441.
Ericsson (2013a).
Ericsson mobility report.
white paper.
Ericsson (2013b).
Traffic exploration tool.
http://www.ericsson.com/TET.
[Online; accessed 03-July-2013].
Stefan Schwarz
y Slide 34 / 31 y y Referencesy
References II
Marzetta, T. (2010).
Noncooperative cellular wireless with unlimited numbers of base station antennas.
IEEE Transactions on Wireless Communications, 9(11):3590–3600.
Rappaport, T., Sun, S., Mayzus, R., Zhao, H., Azar, Y., Wang, K., Wong, G., Schulz, J., Samimi, M., and
Gutierrez, F. (2013).
Millimeter wave mobile communications for 5G cellular: It will work!
IEEE Access, 1:335–349.
Stefan Schwarz
y Slide 35 / 31 y y Referencesy
References III
Sadek, M., Tarighat, A., and Sayed, A. (2007).
A leakage-based precoding scheme for downlink multi-user MIMO channels.
IEEE Transactions on Wireless Communications, 6(5):1711–1721.
Stefan Schwarz
y Slide 36 / 31 y y Referencesy
References IV
Telatar, I. (1999).
Capacity of multi-antenna Gaussian channels.
European Transactions on Telecommunications, 10(6).
Stefan Schwarz