MIT Lincoln Laboratory

mimoASAP-1
bliss
Robust MIMO Wireless
Communication in the Presence
of Interference Using Ad Hoc
Antenna Arrays
Dr. Daniel W. Bliss
& Amanda M. Chan
MIT Lincoln Laboratory
bliss@ll.mit.edu
This work was sponsored by the U.S. Air Force under Air Force contract F19628000-C-0002. Opinions,
interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed
by the United States Government.
MIT Lincoln Laboratory
mimoASAP-2
bliss
Topics
MIMO Communication
Introduction
Military wireless communication
MIMO definition
Ad hoc antenna networks
MIMO Theory
Phenomenology
Receiver
MIT Lincoln Laboratory
mimoASAP-3
bliss
Advanced Military
Wireless Communications
High Data Rate
Applications
Non-Line-of-Sight
Complicated Multipath
Non-Line-of-Sight
Complicated Multipath
Urban
Environment
Real-time tactical
information for
war-fighters
Ad Hoc Distributed
Short Range Network
Reach-back
Link
Remote
Controlled
Vehicles
Forested
Environment
High Power
Jammer
Low Cost
Close Approach
Jammers
MIT Lincoln Laboratory
mimoASAP-4
bliss
Complicated Multipath
Environment
MIMO Communication
Multiple-Input Multiple-Output
Single transmitted data stream
Single received data stream
Employ array of antennas at both
transmitter and receiver
Employ multiple modes through
environment
Not just point-to-point beamforming
Data
01110111001
Data
01110111001
Transmit
Antenna
Array
Space-Time
Coding
Multiple
Input
Receive
Antenna
Array
bounce off
building
Space-Time
Receiver
Multiple
Output
MIT Lincoln Laboratory
mimoASAP-5
bliss
Advantages of MIMO Communication
Coherent receive beamforming
Gain
Jammer mitigation
Transmit spatial diversity
Fading mitigation
Shadowing mitigation
Jammer avoidance
Enables high spectral efficiency
Enables high data rates given
limited bandwidths
Low duty cycle communication
Transmitter
Receiver
MIMO Communication
Multiple-Input Multiple-Output
Transmit
Array
Receive
Array
SISO Communication
Single-Input Single-Output
bounce off
building
MIT Lincoln Laboratory
mimoASAP-6
bliss
Distributed Ad Hoc Antenna Arrays
Multiple-Input Multiple-Output
Single transmit data stream
Single received data stream
Employ users as antenna array
Coherently process received
signal
Use local network to move
distributed data to/from
interested user
Inter-Group
Link
Has
Information
Needs
Information
Intra-Group
Link
Issues
Local networking
Relative local
oscillator errors
Issues
Local networking
Relative local
oscillator errors
MIT Lincoln Laboratory
mimoASAP-7
bliss
Topics
MIMO Communication
Introduction
MIMO Theory
Capacity
Phenomenology
Interference Mitigation
Space-Time Coding
Phenomenology
Receiver
MIT Lincoln Laboratory
mimoASAP-8
bliss
MIMO Capacity Bound
MIMO bound follows different theoretical limit
Divide total energy amongst transmitters
avoiding compressive regime of SISO
Shannon limit
Spectral Efficiency(b/s/Hz)
E
b
/
N
0
(
d
B
)
Shannon Limit
SISO
4x4 MIMO
8x8 MIMO
Transmit
Power
Matrix Channel
Matrix
Determinant
Channel
Loss
Bandwidth
Normalized
Capacity
Transmit Power
(noise normalized)
Assuming Flat
MIMO Channel
Assuming Flat
MIMO Channel
MIT Lincoln Laboratory
mimoASAP-9
bliss
MIMO Channel Knowledge
Channel knowledge affects MIMO capacity and coding
Informed Transmitter Uninformed Transmitter
Determinant

C
IT
? max
tr{P}? P
o
log
2
I ? HPH

Channel
Capacity
(b/s/Hz)
Transmit
Power Matrix
(noise-normalized)
Channel
Matrix

C
UT
= log
2
I +
P
o
n
Tx
HH

Transmitter
Channel
Knowledge
Total Power
(noise-normalized)
Number
Of Transmitters
MIT Lincoln Laboratory
mimoASAP-10
bliss
Channel Matrix
8 x 8 MIMO Example
Line-of-Sight
Random Scattering
C
a
p
a
c
i
t
y

(
b
/
s
/
H
z
)
Line-of-Sight
Random
Scattering
Channel Capacity
Mean SISO SNR (dB)
Uninformed
Transmitter
Line-of-Sight
Random
Scattering
Eigenvalues of HH

R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
Eigenvalue #
Channel matrix, H, contains
complex attenuation between each
transmit and receive antenna
MIT Lincoln Laboratory
mimoASAP-11
bliss
Jammer Mitigation & Avoidance
SINR Loss
Transmitter
Transmit
Array
Interference
Interference
Receive
Array
SIMO MIMO
95% Outage
Capacity
Adaptive performance in
the presence of Jammer
MIMO has better outage
capacity performance
Assumptions
Single high power jammer
I.I.D. random Gaussian
channel
MIMO uninformed
transmitter
MIT Lincoln Laboratory
mimoASAP-12
bliss
Space-Time Coding
Space-time coding converts information bits to waveform
distributed amongst antennas
Space-time coding analogous to conventional (SISO)
coding approaches
Trellis
Low density parity check
Turbo

Coding
&
Modulation
Information
0110
Space-Time Turbo Code Example
Total data rate 2 b/s/Hz
4 transmit antennas
4096 bit interleavers
QPSK constellations
Uninformed transmitter
Space-Time Turbo Code Example
Total data rate 2 b/s/Hz
4 transmit antennas
4096 bit interleavers
QPSK constellations
Uninformed transmitter
Different
Signals
MIT Lincoln Laboratory
mimoASAP-13
bliss
Topics
MIMO Communication
Introduction
MIMO Theory
Phenomenology
Experimental setup
Phenomenology
Receiver
MIT Lincoln Laboratory
mimoASAP-14
bliss
MIMO Experiment
Summer 2002
2 Groups of 4, or
8 Coherent
Transmitters
Near PCS band
2 Groups of 4, or
8 Coherent
Transmitters
Near PCS band
16-Channel
Hi-Fidelity
Data Recording
System
16-Channel
Hi-Fidelity
Data Recording
System
4 Transmit
Antennas
Investigate channel
phenomenology
Study space-time coding
Explore transmitter
coherence requirements
Demonstrate robustness to
Jamming
Cochannel interference
MIT Lincoln Laboratory
mimoASAP-15
bliss
Channel Modes
Experimental Results
Receive
Array
M
IT
Cambridge
R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
Mode #
Random
Measured
Line-of-sight
10
0
2 3 4
-10
1
Random
Measured
Line-of-sight
2 3 4
Mode #
10
0
-10
R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
1
Random
Measured
Line-of-sight
2 3 4
Mode #
10
0
-10
R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
1
Transmit
Array
MIT Lincoln Laboratory
mimoASAP-16
bliss
CDF of
Channel Stationarity
CDFs of Power Weighted Mean cos
2
?
n
Stationary Transmitter
Moving Transmitter (5-10 m/s)
0.1
0.9
Indoor
Outdoor
0.1
0.9
MIT Lincoln Laboratory
mimoASAP-17
bliss
Delay-Frequency Correlations
Experimental Data
#1 #2
Delay (resolution cells, 8? s)
F
r
e
q
u
e
n
c
y

O
f
f
s
e
t

(
r
e
s
o
l
u
t
i
o
n

c
e
l
l
s
,

6
0
H
z
)

Time-Frequency Pulse Response
Receive
Array
Moving
Transmitter
Delayed and
Doppler Shifted
Signal
#1
#2
0
-10
-20
-30
-40
R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
Offset
In Delay
And
Frequency
Offset
In Delay
And
Frequency
MIT Lincoln Laboratory
mimoASAP-18
bliss
Topics
MIMO Communication
Introduction
MIMO Theory
Phenomenology
Receiver
Space-time-frequency
adaptive processing
Multiuser detection
MCMUD
Experimental performance
MIT Lincoln Laboratory
mimoASAP-19
bliss
Receive
Array
Moving
Transmitter
Delayed and
Doppler
Shifted Signal
Adaptive Beamforming in Multipath
Space-Time-Frequency Adaptive Processing

? ?
? ? ?
?
?f ?f
?f ?f ?f

?f
?f ?f ?f
A
d
a
p
t
e
d
C
o
e
f
f
i
c
i
e
n
t
s

?
Delay Taps
Frequency Taps

Filters weights
Jointly take into account
space-time-frequency
correlations
Space-Time-Frequency
Filter Cube
Space
4 Antennas
T
i
m
e

s
F
r
e
q
u
e
n
c
y

2
0
0

H
z
MIT Lincoln Laboratory
mimoASAP-20
bliss
Notional Multiuser Detection
Signal 1 + Signal 2
Demodulate
Signal 1
Demodulate
Signal 1
Remodulate
Signal 1
Remodulate
Signal 1
?
?
-
+
Bits
Signal 1
Signal 2
Signal 1
Signal 2
+
Generic Stage of Successive Decoding
MIT Lincoln Laboratory
mimoASAP-21
bliss
T
e
m
p
o
r
a
l
S
u
b
t
r
a
c
t
i
o
n
T
u
r
b
o
E
n
c
o
d
e
r
Channel
Estimation
Block for Each Transmitter
S
p
a
c
e
-
T
i
m
e
F
r
e
q
u
e
n
c
y
A
d
a
p
t
i
v
e
B
e
a
m
f
o
r
m
e
r
Info
Bits
MCMUD
for Space-Time Turbo Code
Multichannel Multiuser Detector
(MCMUD, pat. pending)
Iterative decoder
Channel estimate
Training-based
Data-directed
Estimation subtraction
(multiuser detection)
Space-time-frequency
adaptive beamformers
S
p
a
c
e
-
T
i
m
e
M
u
l
t
i
p
l
e
x
e
r
S
p
a
c
e
-
T
i
m
e
D
e
m
u
l
t
i
p
l
e
x
e
r
T
u
r
b
o
D
e
c
o
d
e
r
MIT Lincoln Laboratory
mimoASAP-22
bliss
T
e
m
p
o
r
a
l
S
u
b
t
r
a
c
t
i
o
n
T
u
r
b
o
E
n
c
o
d
e
r
Channel
Estimation
Block for Each Transmitter
S
p
a
c
e
-
T
i
m
e
F
r
e
q
u
e
n
c
y
A
d
a
p
t
i
v
e
B
e
a
m
f
o
r
m
e
r
Info
Bits
MCMUD
for Space-Time Turbo Code
Multichannel Multiuser Detector
(MCMUD, pat. pending)
Iterative decoder
Channel estimate
Training-based
Data-directed
Estimation subtraction
(multiuser detection)
Space-time-frequency
adaptive beamformers
S
p
a
c
e
-
T
i
m
e
M
u
l
t
i
p
l
e
x
e
r
-
=
S
p
a
c
e
-
T
i
m
e
D
e
m
u
l
t
i
p
l
e
x
e
r
T
u
r
b
o
D
e
c
o
d
e
r
MIT Lincoln Laboratory
mimoASAP-23
bliss
T
e
m
p
o
r
a
l
S
u
b
t
r
a
c
t
i
o
n
T
u
r
b
o
E
n
c
o
d
e
r
Channel
Estimation
Block for Each Transmitter
S
p
a
c
e
-
T
i
m
e
F
r
e
q
u
e
n
c
y
A
d
a
p
t
i
v
e
B
e
a
m
f
o
r
m
e
r
Info
Bits
MCMUD
for Space-Time Turbo Code
Multichannel Multiuser Detector
(MCMUD, pat. pending)
Iterative decoder
Channel estimate
Training-based
Data-directed
Estimation subtraction
(multiuser detection)
Space-time-frequency
adaptive beamformers
S
p
a
c
e
-
T
i
m
e
M
u
l
t
i
p
l
e
x
e
r
S
p
a
c
e
-
T
i
m
e
D
e
m
u
l
t
i
p
l
e
x
e
r
T
u
r
b
o
D
e
c
o
d
e
r
I
Q
MIT Lincoln Laboratory
mimoASAP-24
bliss
Adaptive Spatial Processing
Iteration #1
Iteration #2
Iteration #3
Experimental Results
Successive MCMUD Iterations
Tx #1 Tx #2
Tx #3 Tx #4
T
r
a
n
i
n
g
-
b
a
s
e
d

S
p
a
c
e
-
F
r
e
q
u
e
n
c
y

F
i
l
t
e
r
Iteration #1
Tx #1 Tx #2
Tx #3 Tx #4
D
a
t
a
-
D
i
r
e
c
t
e
d

S
p
a
c
e
-
T
i
m
e
-
F
r
e
q
u
e
n
c
y

F
i
l
t
e
r
Iteration #2
Tx #1 Tx #2
Tx #3 Tx #4
S
p
a
c
e
-
T
i
m
e
-
F
r
e
q
u
e
n
c
y

F
i
l
t
e
r
W
i
t
h

M
u
l
t
i
u
s
e
r

D
e
t
e
c
t
i
o
n
Iteration #3
Receiver Bit Error Rate
Mean SISO SNR (dB)
MIT Lincoln Laboratory
mimoASAP-25
bliss
4x4 MIMO Performance
Motion, J ammers, and LO Errors
2 Noise Jammers (25 dB JNR)
Moving transmitter (25 mph)
Artificial relative local
oscillator error (? 80 Hz)
Jammer Spatial
Mode Distribution
R
e
l
a
t
i
v
e

P
o
w
e
r

(
d
B
)
Mode #
Receive
Array
25
mph
Jammers
25 dB JNR
Experimental MIMO Performance
Error-free 2b/s/Hz data-link
Near performance of
jammer-free environment!
20 dB Better Than
SISO Theoretic Limit
20 dB Better Than
SISO Theoretic Limit
MIT Lincoln Laboratory
mimoASAP-26
bliss
Summary
MIMO provides robust communication links
New receiver design concepts (MCMUD) enable
communication in complicated environments
Demonstrated dramatic performance advantages
using experimental data
MCMUD enables coherent use of ad hoc distributed
networks for MIMO communication
MIT Lincoln Laboratory
mimoASAP-27
bliss
Acknowledgements
MIT Linconln Laboratory
New Technology Initiative Board
Experiment team
Sean Tobin, Jeff Nowak, Lee Duter, John Mann,
Bob Downing, Peter Priestner, Bob Devine,
Tony Tavilla, Andy McKellips, Gary Hatke
Code, algorithm and experiment design
Keith Forsythe, Peter Wu, Ali Yegulalp
Analysis support
Amanda Chan
Students
Nick Chang (U. Mich),
Naveen Sunkavally (MIT)
MIT Lincoln Laboratory
mimoASAP-28
bliss
Backup Slides
Advanced Shoe-Phone Technology
MIT Lincoln Laboratory
mimoASAP-29
bliss
MIMO Ground-to-Ground Example
Non-Line-Of-Sight
Receive
Array
MIT
Transmit
Array
Cambridge
Boston
University
4 x 4 MIMO performance
dramatically better than SISO
6 times bit rate
Fading resistance
Significant Doppler introduced
by fast vehicles on Storrow Dr.
MIMO vs. SISO
Mean SISO SNR
Simulated SISO
Block Fading
1/3 b/s/Hz
Diversity 1
Experimental
4x4 MIMO
2 b/s/Hz
Charles
River
Simulated
MIT Lincoln Laboratory
mimoASAP-30
bliss
Space-Time Codes Used in Experiment
4 Transmitters
Alamouti (2 Tx) , ? = 2
Block, ? = 3
Turbo, ? = 2
Turbo, ? = 4
CDMA, ? = 12/256
LDPC, ? = 1
LDPC, ? = 2
Trellis (Chen) , ? = 2
8 Transmitters
Channel probe
2+2+2+2 Trellis, ? = 6
Block, ? = 3
Turbo, ? = 4
Turbo, ? = 8
CDMA, ? = 18/256
CDMA, ? = 20/256
LDPC, ? = 2
Space-Time Code Source
New Designs
Provided by campus
Literature
? Spectral Efficiency (b/s/Hz)
MIT Lincoln Laboratory
mimoASAP-31
bliss
MCMUD Detector
Progressive Complexity
Joint channel and data estimation
First iteration access to limited training data or channel
estimate from previous frame
Increase detector complexity with iteration
Increase number of turbo iterations with number of
detector iterations
Training
Based
Coarse
Space
Frequency
Beamformer
P
e
r
f
o
r
m
a
n
c
e
Decision
Directed
Space
Time
Frequency
Beamformer
Decision
Directed
Space
Time
Frequency
Beamformer
Multiuser
Detection

Detector Iteration
MIT Lincoln Laboratory
mimoASAP-32
bliss
History of Wireless Communication
Year
1
8
6
5
1
8
8
7
Hertz
Confirms
E&M
Maxwell
Electro-
Magnetics
Berrou, et al
Turbo Codes
1
9
9
3
1
9
9
6
Foschini
MIMO
1
9
9
8
Tarokh
Space-Time
Trellis Codes
Liu, et al
Space-Time
Turbo Codes
1
9
5
8
Price&Green
(Lincoln)
Rake Receiver
QuickTime and a TIFF (Uncompressed) decompressor are needed to see this picture.
Cooper, et al
(Motorola)
Handheld
Cell Phone
1
9
7
3
1
9
8
2
GSM
Digital Cellular
Development
Begins
1
9
8
7
SINCGARS
Production
QuickTimeand a TIFF (Uncompressed) decompressor are needed to see this picture.
Bell-Lab
Car Radio
Phone
1
9
0
1
Marconi
Practical
Wireless
System
1
9
2
4
1
9
4
2
First US
Military
Handheld
BC-611
Shannon
Information
Theory
1
9
4
8
Hamming
Error-
Correcting
Codes
1
9
5
0
MIT Lincoln Laboratory
mimoASAP-33
bliss
Important Antenna Array Concepts
Diversity
Adaptive Spatial
Beamforming
User of
Interest
Interfering
User
Null in
Beam Pattern
Transmitter
Receive
Array
Adaptive
Spatial
Filter
MIT Lincoln Laboratory
mimoASAP-34
bliss
The Channel Matrix
A Toy Model
Toy MIMO channel model
2x2
line of sight
Resolving individual antennas
increases eigenvalue
MIMO systems in real
environments employ scatterers to
increase effective aperture
Toy MIMO channel model
2x2
line of sight
Resolving individual antennas
increases eigenvalue
MIMO systems in real
environments employ scatterers to
increase effective aperture
Channel Matrix,
Eigenvalues of Channel Matrix
2x2 MIMO System
V
a
r
y

A
p
e
r
t
u
r
e
??
b ?
2
?
arccos
??
h
1

??
h
2
??
h
1
??
h
2
??
H ?
??
h
1
??
h
2
? ?
? 2 a
??
v
1
??
v
2
? ?
Unit norm
steering vector
MIT Lincoln Laboratory
mimoASAP-35
bliss
Space-Time Turbo Code
Block diagram for
space-time turbo code
Rate 2 b/s/Hz
123 kChip/s
4 Tx antennas
4096 bit interleavers
QPSK constellation
Optional training data
Info
Bits
Convolutional
Encoder
Convolutional
Encoder
Interleaver
#1
Interleaver
#2
Alternating
Interleaver
Interleaver
#1
Interleaver
#2
?
Space-Time
Turbo Code
i
i
?
B
l
o
c
k
W
i
s
e
MIT Lincoln Laboratory
mimoASAP-36
bliss
Uncooperative External Interference
Effective Loss of Complexity
Uncooperative interference is
equivalent to spatially
correlated noise
Covariance of interference plus
noise
Maximize capacity by
decorrelating channel matrix
with respect to interference
Estimate using new
Modes near interference energy
become less useful
Effectively reduces the
environmental complexity
Uncooperative interference is
equivalent to spatially
correlated noise
Covariance of interference plus
noise
Maximize capacity by
decorrelating channel matrix
with respect to interference
Estimate using new
Modes near interference energy
become less useful
Effectively reduces the
environmental complexity
Channel Capacity in
Interference
Noise-Normalized
Transmit
Covariance
Matrix
Interference Whitened
Channel Matrix
Informed Transmitter (IT)
Uninformed Transmitter (UT)
R

H ? R
?1/ 2
H

P

H


C
IT
? max
tr{

P }? P
o
log
2
I ?

H

P

H

C
UT
? log
2
I ?
P
o
n
Tx

H

H

MIT Lincoln Laboratory

mimoASAP-37
bliss
The Channel Matrix
Channel matrix, H, contains complex
attenuation between each transmit and
receive antenna
Large channel eigenvalues of HH

are useful
R
e
l
a
t
i
v
e

P
o
w
e
r
Channel Eigenvalues
High
Capacity
Environment Low
Capacity
Environment
Sorted by Eigenvalue Strength
Few Useful Modes
Many Useful Modes
Many Useful Modes
Scatterers