xv

Statistical MIMO Radar
Abstract Inspired by recent advances in multiple-input multiple-output (MIMO)
communications, we introduce the statistical MIMO radar concept. Unlike
beamforming, array radar, or STAP, which presuppose a high correlation
between signals either transmitted or received by an array, the proposed
MIMO radar exploits the independence between signals at the array elements.
Whereas correlation-based array techniques are capable of providing degrees
of freedom for spatial fltering, they have no bearing on the effects of target
scattering. Radar targets generally consist of many small elemental scatterers
that are fused by the radar waveform and the processing at the receiver to
result in echoes with fuctuating amplitude and phase. In conventional radar,
target radar cross-section (RCS) fuctuations are regarded as a nuisance that
degrades radar performance. The novelty of statistical MIMO radar is that it takes
the opposite view, namely, it capitalizes on target RCS scintillations and glint to
improve the radar’s performance. MIMO radar utilizes multiple antennas at both
the transmitter and receiver. It can be applied in monostatic or bistatic modes.
The antennas at each end of the radar system have to be suffciently separated
such that the target provides uncorrelated refection coeffcients between each
transmit/receive pair of antennas. We demonstrate that the MIMO radar greatly
improves detection and estimation performance due to the absence of target
fades. Specifcally, statistical MIMO radar overcomes target RCS fuctuations
by averaging over many decorrelated channels between transmit and receive
antennas. Subsequently, the received signal is a superposition of independently
faded signals, and the average SNR of the received signal is more or less
constant. This is equivalent to converting a Swerling case I RCS to a Swerling
case II, but without the loss of time. Moreover, MIMO spatial diversity also
eliminates the deep interference nulls in the elevation coverage due to surface
multipath refection.
Alex Haimovich and Eran Fishler
New Jersey Institute of Technology
phone: 973-596-3534
email: haimovic@njit.edu
email: eran.fshler@njit.edu
Rick Blum
Lehigh University
email: rblum@eecs.lehigh.edu
Len Cimini
University of Delaware
email: cimini@ece.udel.edu
Dmitry Chizhik and Reinaldo Valenzuela
Bell Labs—Lucent Technologies
email: chizhik@lucent.com
email: rav@lucent.com
Abstract Presentation Back to TOC
Next Abstract
Report Documentation Page
Form Approved
OMB No. 0704-0188
Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information,
including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington
VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it
does not display a currently valid OMB control number.
1. REPORT DATE
20 DEC 2004
2. REPORT TYPE
N/A
3. DATES COVERED
-
4. TITLE AND SUBTITLE
Statistical MIMO Radar
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
New Jersey Institute of Technology; Lehigh University
8. PERFORMING ORGANIZATION
REPORT NUMBER
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
11. SPONSOR/MONITOR’S REPORT
NUMBER(S)
12. DISTRIBUTION/AVAILABILITY STATEMENT
Approved for public release, distribution unlimited
13. SUPPLEMENTARY NOTES
See also, ADM001741 Proceedings of the Twelfth Annual Adaptive Sensor Array Processing Workshop,
16-18 March 2004 (ASAP-12, Volume 1)., The original document contains color images.
14. ABSTRACT
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF
ABSTRACT
UU
18. NUMBER
OF PAGES
13
19a. NAME OF
RESPONSIBLE PERSON
a. REPORT
unclassified
b. ABSTRACT
unclassified
c. THIS PAGE
unclassified
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39-18
Motivation
• Radar targets provide a rich
scattering environment.
• Conventional radars experience
target fluctuations of 5-25 dB.
• Slow RCS fluctuations (Swerling
I model) cause long fades in
target RCS, degrading radar
performance.
• In statistical MIMO the angular
spread of the target backscatter
is exploited in a variety of ways
to extend the radar’s
performance envelope.
Backscatter as a function of azimuth angle,
10-cm wavelength [Skolnik 2003].
Abstract Presentation Back to TOC
Next Abstract
The S-MIMO Concept
• Statistical-MIMO radar offers the potential for significant
gains:
• Detection/estimation performance
• Resolution performance
• Here, we focus only on detection performance
• Our results question the common belief that one should
maximize the coherent processing gain.
• With S-MIMO a very sparse array of sensors transmits a set
of orthogonal waveforms.
• By using this approach, we create many "independent"
radars, that average out target scintillations.
Signal Model
• Point source assumption dominates current models used in
radar theory.
• This model is not adequate for an array of sensors with large
spacing between the array elements.
• Distributed target model
Many
random
scatterers
Signal Model (Cont.)
( )
{ }
o
o
o o ì
ì o o
-
- >
> 
1 2
1 2
Denote by the gain between the th transmitter and
th receiver. It can be shown that ~ 0,1 .
Take and . We can show that if either /
or ' ' / ' , then 0,
jk
jk
jk il c
H
c jk il
k
j CN
d d d
d d d E
{ }
o o =
and otherwise
1
H
jk il
E
r1
r2
d
d
1
d
2
d’
1
t1 t2
d’
2
d’
Target beamwidth
Phased Array Radar
( )
-
-
-
Phased array radars consist of closely spaced sensors.
The gain between each transmitter receiver pair is the same.
Transmitted waveform is
This gives rise to the following received signal m
t s
( ) ( ) ( ) ( ) ( )
( ) ( )
o t
o
= ÷ +
-
=
0 0 0 0
2
0 0 0
odel
, ,
If beamformer is applied at both the transmitter and the receiver,
then the received signal at the output of the beamformer equals
, ,
H E
t x y x y t t
M
E
y t x y x
M
r a b s n
a b
( ) ( ) ( )
t
'
÷ +
2
0
y s t n t
S-MIMO Radar
( ) ( ) ( ) ( ) ( )
t
-
-
= ÷ +
-
In S-MIMO radar, the inter element spacing is large. The gain between every
transmitter receiver pair is different.
The received signal is given by
vec ~ CN ,
Each
E
t t t
M
r Hs n H 0 I
( )
-
transmitting element transmits one of M orthogonal waveforms.
By matched filtering the received signal at each sensor with each of the
transmitted waveforms we can reconstract

ji
r t
( ) ( )
o t = ÷ +
- Therefore, instead of coherent gain of , we created independent radars.
ji i ji
E
s t n t
M
MN MN
The Radar Detection Problem
t
t
-
-
0
1
The radar detection problem:
: Target does not exists at delay
: Target exists at delay
Assume that all the parameters are known. The optimal detector is the LRT
detector, and it
H
H
( ) ( )
( ) ( )
o
>
=
<
0
1
1
0
|
lo
is given by,
g
|
H
H
f t H
T
f t H
r
r
S-MIMO Radar
( )
;
t
o
o o
÷
>
= = ÷
<
-
1
2
2
0
2
2
1
Denote by the vector that contains the output of a bank of matched filters
sampled at . The op
,
timal detector is
whe re 1
2
MN
H
n
FA
H
T F P
x
x
( )
; ;
o
o
÷
-
| |
|
= ÷ ÷
|
|
+
|
\ .
2 2
2 2
2
1
2
It is possible to compute the probability of detection as a
function of the probability of false alarm, and i

t equa s
1 1
l
MN MN
n
D FA
n
P F F P
E
M
( ) ( ) ( )
( )
( )
;
; ;
o
o o
o
t
o
÷
÷ ÷
>
= = ÷
<
| |
= ÷ ÷
|
- =
+
\
÷
.
í
0
2
2
1
2 2
2 2
2
2
0 0
1
2
1 1
2
Let , . The optimal detector:
| | 1
2
1 1
H
n
FA
H
n
D FA
H
n
x t x y s t dt
N
T x F P
P F F P
EN
r a
Phased Array Radar
The Invariance Detector
-
-
-
2
Assume access to a vector that contains samples of
the noise process.
Note that is the ML estimate of the noise level.
The optimal detector whose performance depends only on
SNR
L
/
(not
L y
y
o =
-
>
<
1
0
2
2
on the noise level)

This test statistic is very intuitive. It normalizes the UMP
test by the best estimate of the noise level.
H
H
T
x
y
Example: Miss Probability
• Assume a system with four receiving and one or two transmitting
antennas, M=2, N=4, and the probability of false alarm is 1e-6
0 5 10 15 20 25 30
10
-3
10
-2
10
-1
10
0
SNR
P
M
D
S-MIMO
Phas ed Array
I-S-MIMO L=64
I-Phas ed Array L=64
Example: ROC
• The following figure depicts the ROC. SNR=10dB.
10
-10
10
-9
10
-8
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
P
FA
P
D
S-MIMO
Phas ed Array
I-S-MIMO L=64
I-Phas ed Array L=64
Concluding Remarks
• S-MIMO is a new concept for radar systems.
• This concept utilizes spatial diversity in order to overcome
target scintillations.
• At 90% probability of detection, the proposed system
outperform phased array radars by 5 dB, which is equivalent
to almost twice the range.
• The S-MIMO radar can be shown to have superior
performance in range estimation and resolution as well.
Abstract Presentation Back to TOC
Next Abstract

ABSTRACT c. PERFORMING ORGANIZATION REPORT NUMBER 10. Send comments regarding this burden estimate or any other aspect of this collection of information. DISTRIBUTION/AVAILABILITY STATEMENT Approved for public release. THIS PAGE 17. AUTHOR(S) 5d. SPONSOR/MONITOR’S ACRONYM(S) 11. REPORT TYPE 3. WORK UNIT NUMBER 7. DATES COVERED 20 DEC 2004 4. SUBJECT TERMS 16. gathering and maintaining the data needed. SECURITY CLASSIFICATION OF: a. TASK NUMBER 5f. 8-98) Prescribed by ANSI Std Z39-18 . REPORT b. ABSTRACT 15. Directorate for Information Operations and Reports. Lehigh University 9. 0704-0188 Public reporting burden for the collection of information is estimated to average 1 hour per response. CONTRACT NUMBER 5b. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 8.Report Documentation Page Form Approved OMB No. REPORT DATE 2. 14. NUMBER OF PAGES 19a. LIMITATION OF ABSTRACT 18. 1. GRANT NUMBER 5c. The original document contains color images. PROJECT NUMBER 5e. searching existing data sources. NAME OF RESPONSIBLE PERSON unclassified unclassified unclassified UU 13 Standard Form 298 (Rev. and completing and reviewing the collection of information. 1215 Jefferson Davis Highway. SUPPLEMENTARY NOTES See also. Respondents should be aware that notwithstanding any other provision of law. Suite 1204. SPONSOR/MONITOR’S REPORT NUMBER(S) 12. including the time for reviewing instructions. including suggestions for reducing this burden. PROGRAM ELEMENT NUMBER Statistical MIMO Radar 6. ADM001741 Proceedings of the Twelfth Annual Adaptive Sensor Array Processing Workshop. Arlington VA 22202-4302. Volume 1). TITLE AND SUBTITLE N/A 5a.. 16-18 March 2004 (ASAP-12. no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) New Jersey Institute of Technology. distribution unlimited 13. to Washington Headquarters Services.

Slow RCS fluctuations (Swerling I model) cause long fades in target RCS. 10-cm wavelength [Skolnik 2003]. Conventional radars experience target fluctuations of 5-25 dB. In statistical MIMO the angular spread of the target backscatter is exploited in a variety of ways to extend the radar’s performance envelope. degrading radar performance. . • • • Backscatter as a function of azimuth angle.Abstract Presentation Back to TOC Next Abstract Motivation • Radar targets provide a rich scattering environment.

that average out target scintillations. we create many "independent" radars.The S-MIMO Concept • Statistical-MIMO radar offers the potential for significant gains: • Detection/estimation performance • Resolution performance • Here. . • By using this approach. • With S-MIMO a very sparse array of sensors transmits a set of orthogonal waveforms. we focus only on detection performance • Our results question the common belief that one should maximize the coherent processing gain.

• This model is not adequate for an array of sensors with large spacing between the array elements. • Distributed target model Many random scatterers .Signal Model • Point source assumption dominates current models used in radar theory.

We can show that if either d1  d c / d 2 or d '1  d ' c / d '2 . and otherwise E  jk  ilH  1     d’2 d2 Target beamwidth d d’ d1 r1 r2 t1 d’1 t2 . then E  jk  ilH  0.1 .  Take  jk and  il .Signal Model (Cont.)  Denote by  jk the gain between the kth transmitter and jth receiver. It can be shown that  jk ~ CN  0.

r t   then the received signal at the output of the beamformer equals y t   2 2 E  a  x0 . y 0  b  x0 .  Transmitted waveform is s  t   This gives rise to the following received signal model E H x0 . y 0  b  x0 . y 0  s  t     n  t  a  M  If beamformer is applied at both the transmitter and the receiver.Phased Array Radar  Phased array radars consist of closely spaced sensors. y 0  s  t     n   t  M . The gain between each transmitter receiver pair is the same.

we created MN independent radars. . The gain between every transmitter receiver pair is different. the inter element spacing is large. instead of coherent gain of MN.S-MIMO Radar  In S-MIMO radar.  The received signal is given by E r t   Hs  t     n  t  M vec H ~ CN  0.  By matched filtering the received signal at each sensor with each of the transmitted waveforms we can reconstract E r ji  t    ji si  t     n ji  t  M  Therefore.I  Each transmitting element transmits one of M orthogonal waveforms.

f r  t  | H1  H0 T  log  H1 f  r  t  | H0  S-MIMO Radar  Denote by x the vector that contains the output of a bank of matched filters sampled at  . T  x F 2 1  PFA   H0 2  2 MN . and it is given by. The optimal detector is H1 2  n 1 2  where   . The optimal detector is the LRT detector.The Radar Detection Problem  The radar detection problem: H0 : Target does not exists at delay  H1 : Target exists at delay   Assume that all the parameters are known.

 It is possible to compute the probability of detection as a function of the probability of false alarm. The optimal detector:  H0 T | x |2  H1 2 N n 1 F 2 1  PFA    2 2 2  n  1 1 PD  1  F 2  2 F 2 1  PFA   2 2    n  EN . and it equals   2  n  1 PD  1  F 2  F 2 1  PFA   2 MN 2 MN  E  2    n M  Phased Array Radar  Let x   r H  t  a  x0 . y 0  s  t    dt .

The Invariance Detector  Assume access to a vector y that contains L samples of the noise process. . 2  The optimal detector whose performance depends only on SNR (not on the noise level) 2 H x  1 T 2  y  H0  This test statistic is very intuitive. It normalizes the UMP test by the best estimate of the noise level.  Note that y / L is the ML estimate of the noise level.

M=2. N=4. and the probability of false alarm is 1e-6 10 0 S -MIMO P has e d Array I-S -MIMO L=64 I-P has ed Array L=64 10 -1 P MD 10 -2 10 -3 0 5 10 15 S NR 20 25 30 .Example: Miss Probability • Assume a system with four receiving and one or two transmitting antennas.

5 0. SNR=10dB.8 0.Example: ROC • The following figure depicts the ROC.6 0.9 0.4 S -MIMO P ha s e d Arra y I-S -MIMO L=64 I-P has ed Array L=64 10 -9 0.3 -10 10 10 -8 10 -7 10 -6 10 P FA -5 10 -4 10 -3 10 -2 10 -1 .7 PD 0. 1 0.

• The S-MIMO radar can be shown to have superior performance in range estimation and resolution as well. • This concept utilizes spatial diversity in order to overcome target scintillations. which is equivalent to almost twice the range. • At 90% probability of detection. .Abstract Presentation Back to TOC Next Abstract Concluding Remarks • S-MIMO is a new concept for radar systems. the proposed system outperform phased array radars by 5 dB.

Sign up to vote on this title
UsefulNot useful