MM

YS
Very Large MIMO Systems:
Opportunities and Challenges
Erik G. Larsson
March 21, 2012
Div. of Communication Systems
Dept. of Electrical Engineering (ISY)
Linkoping University
Linkoping, Sweden
www.commsys.isy.liu.se
With thanks to my team and collaborators:
Hien Q. Ngo (LiU, Sweden)
Antonios Pitarokoilis (LiU)
Saif Mohammed (LiU)
Daniel Persson (LiU)
Fredrik Rusek (Lund, Sweden)
Ove Edfors (Lund)
Buon Kiong Lau (Lund)
Fredrik Tufvesson (Lund)
Thomas L. Marzetta (Bell Labs/Alcatel-Lucent, USA)
Christoph Studer (Rice Univ., USA)
1/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Why MIMO

Array gain (beamforming)

Spatial division mult. access

Spatial multiplexing

Rate min (n
r
, n
t
) log (1 + SNR)

Reliability p
e
SNR
n
r
n
t
2/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Very Large MIMO
M=
x100
antennas!
K
terminals
k=1
k=K

We think of very large (multiuser) MIMO as a system with

M K 1

coherent, but simple, processing

Potential to improving rate & reliability dramatically

Potential to scaling down TX power drastically

3/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large MIMO Arrays

Reduce bulky items (coax)

Each antenna unit simple (low accuracy)

Resilience against individual failures (hotswapping)

Potential economy of scale in manufacturing

Enable for mmWave radio (60-300 GHz)?

4/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large MIMO: Some Known Facts
(Notation: M antennas, K terminals, power per terminal P)

Linear processing (MRC/MRT, ZF, ... ) nearly optimal as

M K 1

P and M large enough pilot contamination limits

performance

Scaling down P with M noise will limit performance.

Perfect CSI & optimal processing P can be scaled as 1/M

Given linear processing and imperfect CSI, in a MU system,

P can be scaled as 1/

M
5/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large MIMO: Some Known Facts, cont.
System performance can be limited by

intracell interference

pilot contamination

thermal noise + intercell interference

so there are several possible operating points depending on

number of antennas

available TX power

choice of receiver/precoder algorithm

coherence time (dictates ultimately the number of users served)

6/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large MIMO: Some Beliefs/Speculation

Not enough time for CSI feedback, so must operate in TDD mode.

Not enough pilots

Not enough resources for fast CSI feedback

System will operate in nearly-noise limited regime (1 bpcu/term)

Very aggressive spatial multiplexing; aggregate eciency K bpcu

Each user could get the full bandwidth

simple MAC, little or no control signaling

Impairments, e.g., multiuser interference (almost) drown in noise

linear or nearly-linear receivers

May even get away with equalization-free (matched lter only) SC

transmission

Vast excess (M K) of degrees of freedom:

use for HW-friendly signal shaping and smart receiver algorithms

Per-antenna constant envelope or low-PAR multiuser precoding

Channel estimation exploiting subspaces

Channel will harden (random matrix theory)

Larger array reveals new propagation phenomena

7/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large MIMO: Some of the Most Important Questions

Processing will have to be simple (linear). How good is this?

Non-CSI@TX operation:
broadcasting (control signaling) and acquisition

Hardware imperfections: phase noise, I/Q imbalance, A/D, PAs

Synchronization at low SNR

TDD will bring us pilot contamination in the downlink.

How bad is this really in practice?

TDD will require reciprocity calibration.

How, when and at what cost?
8/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Favorable Propagation and
Its Implications
9/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Favorable propagation and Rate

Consider M K, MIMO link, channel H, M K. Rate:

R =
K

k=1
log
2
_
1 +
SNR
K

2
k
_
, SNR =
P
tot
N
0

If |H
ij
| 1,

K
k=1

2
k
= H
2
MK, so
log
2
(1 +MSNR)
. .
rank-1 channel (LoS)

2
1
=MK,
2
2
==
2
K
=0
R K log
2
_
1 +
M
K
SNR
_
. .
H
H
HI (full rank channel)

2
1
==
2
K
=M
Favorable propagation

H i.i.d. and M K favorable propagation.

In MU-MIMO (H G), favorable propagation if

G
H
G
M

_

1
0 0
0
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
0 0
K
_

_
D, M K
10/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Favorable propagation and ideal channels
40 30 20 10 0 10 20 30
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
P
r
o
b
(

a
b
s
c
i
s
s
a
)
ordered singular values [dB]


i.i.d 6x128
i.i.d. 6x6
11/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Do we have favorable propagation in practice?

Our partners at Lund Univ., Sweden have conducted unique

measurements [RPL2011+,GERT2011].

Indoor 128-ant. (4x16 dual-pol.) array. 3 users indoor, 3 outdoor.

2.6 GHz CF, 50 MHz BW, 100 snapshots (10m).

Normalized to retain only small-scale fading.

12/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Lund measurements, example of results
40 30 20 10 0 10 20 30
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
P
r
o
b
(

a
b
s
c
i
s
s
a
)
ordered singular values [dB]


meas 6x128
meas 6x6
13/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Spectral/energy eciency tradeos
14/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Uplink single-cell, model assumptions

g
k
=

k
h
k
.

h
k
: small scale fading;
E[|h
mk
|
2
] = 1; zero mean.

k
: path loss+shadowing

SNR for kth terminal: P

k

RX signal:
y =

P
K

k=1
g
k
..
M1
x
k
+n, E[|x
k
|
2
] = 1, n
i
CN(0, 1)

Write as
y =

P G
..
MK
x
..
K1
+n, G = HD
1/2
, H [h
1
, . . . , h
K
], D
_

1
.
.
.

K
_

_
15/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Spectral-energy eciency tradeo

Sum-spectral eciency:
R =
_
_
_
_
1

T
_
K log
2
_
1 +
(M1)P
2
(K1)P
2
+(+K)P+1
_
, for MRC
_
1

T
_
K log
2
_
1 +
(MK)P
2
(+K)P+1
_
, for ZF

Energy eciency: =
R
P

Reference mode: K = 1, M = 1
arg max
1T

Single-user system: K = 1, M xed

arg max
P,
, s.t. S = const., 1 T

Multi-user system: M xed

arg max
P,K,

s.t. S = const., K T
(K M for ZF)
16/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Cellular UL, (BT)
coh.
= 14 14, M = 1 [NLM2011]
0 10 20 30 40 50 60 70 80 90
10
-1
10
0
10
1
10
2
10
3
10
4
K=1, M=1
20 dB
10 dB
0 dB
-10 dB

R
e
l
a
t
i
v
e

E
n
e
r
g
y
-
E
f
f
i
c
i
e
n
c
y

(
b
i
t
s
/
J
)
/
(
b
i
t
s
/
J
)

Spectral-Efficiency (bits/s/Hz)
17/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Cellular UL, (BT)
coh.
= 14 14, M = 100 [NLM2011]
0 10 20 30 40 50 60 70 80 90
10
-1
10
0
10
1
10
2
10
3
10
4
K=1, M=1
20 dB
10 dB
0 dB
-10 dB

R
e
l
a
t
i
v
e

E
n
e
r
g
y
-
E
f
f
i
c
i
e
n
c
y

(
b
i
t
s
/
J
)
/
(
b
i
t
s
/
J
)

Spectral-Efficiency (bits/s/Hz)
K=1, M=100
18/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Cellular UL, (BT)
coh.
= 14 14, M = 100 [NLM2011]
0 10 20 30 40 50 60 70 80 90
10
-1
10
0
10
1
10
2
10
3
10
4
K=1, M=1
MRC
20 dB
10 dB
0 dB
-10 dB
-20 dB

R
e
l
a
t
i
v
e

E
n
e
r
g
y
-
E
f
f
i
c
i
e
n
c
y

(
b
i
t
s
/
J
)
/
(
b
i
t
s
/
J
)

Spectral-Efficiency (bits/s/Hz)
K=1, M=100
M = 100
ZF
19/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Cellular UL, (BT)
coh.
= 14 14, M = 50 [NLM2011]
0 10 20 30 40 50 60 70 80 90
10
-1
10
0
10
1
10
2
10
3
10
4
-20 dB
M=50
ZF
MRC
20 dB
10 dB
0 dB
-10 dB

R
e
l
a
t
i
v
e

E
n
e
r
g
y
-
E
f
f
i
c
i
e
n
c
y

(
b
i
t
s
/
J
)
/
(
b
i
t
s
/
J
)

Spectral-Efficiency (bits/s/Hz)
K=1, M=50
K=1, M=1
20/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Large-Scale MU-MIMO
Downlink Precoding
21/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Downlink precoding - general remarks

In the downlink, the are M K unused degrees of freedom.

These excess DoF could be used to

Null out interference

Shape the transmitted signals in a hardware-friendly way

An excess in the number of antennas also means that simple

precoders (MRT, time-reversal) may be used to combat frequency
selectivity
22/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Constant Envelope MU-MIMO Downlink Precoding

Constant-envelope transmission [SM2011a,SM2011b]

x =
_
P
M
_

_
e
j
1
.
.
.
e
j
M
_

Insensitive to PA non-linearity.

How well can we approximate a desired waveeld at K locations, by

varying only the phase of the transmitted signals?

Not to be confused with equal-gain transmission (something entirely

dierent)!
23/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Constant Envelope versus Beamforming
u

P
h

1
h

P
h

m
h

P
h

M
h
Amplitude range = [0

P|u|]
e
j
u
1
e
j
u
m
e
j
u
M
Constant amplitude =
_
P
M
_
P
M
_
P
M
_
P
M
Antenna 1 Antenna 1
Antenna m Antenna m
Antenna M Antenna M
Beamforming Constant-envelope transmission
24/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Constant-env. MU-MIMO precoding, Gauss. symbols
0 50 100 150 200 250 300
12
9
6
3
0
3
6
9
12
15 15
No. of Base Station Antennas (M)

M
i
n
.

r
e
q
d
.

P
T
/

2

(
d
B
)

t
o

a
c
h
i
e
v
e

a

p
e
r

u
s
e
r

r
a
t
e

o
f

2

b
p
c
u


K = 10, Proposed CE Precoder (CE)
K = 10, ZF Phaseonly Precoder (CE)
K = 10, GBC Sum Cap. Upp. Bou. (APC)
K = 40, Proposed CE Precoder (CE)
K = 40, ZF Phaseonly Precoder (CE)
K = 40, GBC Sum Cap. Upp. Bou. (APC)
1.7 dB
25/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
PAR-Aware MU-MIMO OFDM Downlink [SL2012]

Precoder design problem:

(PMP)
_

_
minimize
a
1
,..., a
N
max
_
a
1

, . . . , a
N

_
subject to s
w
= H
w
f
w
( a
1
, . . . , a
N
), w T
0
N1
= f
w
( a
1
, . . . , a
N
), w T
c
.

H
w
: time-frequency channel

f
w
() linear function; includes OFDM modulation and S/P
conversion

T : used subcarriers; T
c
: null subcarriers

Convex optimization problem, fast algorithm; relaxation

26/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
PAR-aware MU-MIMO OFDM DL (99% percentiles)
27/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Literature
RPL+2011 F. Rusek, D. Persson, B. K. Lau, E. G. Larsson, T. L.
Marzetta, O. Edfors, and F. Tufvesson, Scaling up MIMO:
Opportunities and Challenges with Large Arrays,
arXiv:1201.3210, 2011.
NLM2011 H.Q. Ngo, E.G. Larsson and T. Marzetta, Energy and
Spectral Eciency of Very Large Multiuser MIMO
Systems, arXiv:1112.3810, 2011.
LM2011a E.G. Larsson and S.K. Mohammed, Per-antenna Constant
Envelope Precoding for Large Multi-User MIMO Systems,
arXiv:1201.1634v1, 2011.
LM2011b E.G. Larsson and S.K. Mohammed, Single-User
Beamforming in Large-Scale MISO Systems...: The
Doughnut Channel, arXiv:1111.3752, 2011.
SL2012 C. Studer and E.G. Larsson, PAR-Aware Large-Scale
Multi-User MIMO-OFDM Downlink, arXiv:1202.4034,
2012.
HBD2011 J. Hoydis, S. ten Brink, M. Debbah, Massive MIMO: How
Many Antennas do We Need?, arXiv:1107.1709, 2011.
GERT2011 X. Gao, O. Edfors, F. Rusek, and F. Tufvesson, Linear
pre-coding performance in measured very-large MIMO
channels, IEEE VTC 2011
Mar2010 T. L. Marzetta, Noncooperative MU-MIMO with unlimited
numbers of base station antennas, IEEE Trans. Wireless.
Comm. 2010.
JAMV2011 J. Jose, A. Ashikhmin, T. L. Marzetta, and S. Vishwanath,
Pilot contamination and precoding in multi-cell TDD
systems, IEEE Trans. Wireless Commun., 2011.
GJ2011 B. Gopalakrishnan and N. Jindal, An Analysis of Pilot
Contamination on Multi-User MIMO Cellular Systems with
Many Antennas, IEEE SPAWC 2011.
NML2011 H. Q. Ngo, T. Marzetta and E. G. Larsson, Analysis of the
pilot contamination eect in very large multicell multiuser
MIMO systems for physical channel models, IEEE ICASSP
2011.
PML2012 A. Pitarokoilis, S. K. Mohammed and E. G. Larsson, On
the optimality of single-carrier transmission in large scale
antenna systems, IEEE Wireless Communication Letters,
submitted, 2012.
CMT2004 G. Caire, R. M uller and T. Tanaka, Iterative multiuser
joint decoding: optimal power allocation and low-complexity
implementation, IEEE Trans. IT, 2004.
ZLW2007 H. Zhao, H. Long and W. Wang, Tabu search detection for
MIMO systems, IEEE PIMRC 2007.
VMCR2008 K. Vishnu Vardhan, S. Mohammed, A. Chockalingam, and
B. Sundar Rajan, A low-complexity detector for large
MIMO systems and multicarrier CDMA systems, IEEE
JSAC 2008.
Sun2009 Y. Sun, A family of likelihood ascent search multiuser
detectors: an upper bound of bit error rate and a lower
bound of asymptotic multiuser eciency, IEEE JSAC 2009.
LH1999 A. Lampe and J. Huber, On improved multiuser detection
with iterated soft decision interference cancellation, in
Proc. IEEE Communication Theory Mini-Conference, 1999.
28/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS
Thank You
29/29
Erik G. Larsson
Very Large MIMO Systems
Communication Systems
Link oping University
MM
YS