1
MIMO Radar Diversity Means Superiority
Jian Li and Petre Stoica
Abstract A MIMO (multi-input multi-output) radar system,
unlike a standard phased-array radar, can transmit via its
antennas multiple probing signals that may be correlated or
uncorrelated with each other. We provide herein an overview of
our recent result, showing that this waveform diversity enables
the MIMO radar superiority in several fundamental aspects,
including: 1) significantly improved parameter identifiability,
2) direct applicability of adaptive arrays for target detection
and parameter estimation, and 3) much enhanced flexibility for
transmit beampattern design. Specifically, we show that 1) the
maximum number of targets that can be uniquely identified by
the MIMO radar is up to Mt times that of its phased-array
counterpart, where Mt is the number of transmit antennas, 2)
the echoes due to targets at different locations can be linearly
independent of each other, which allows the direct application
of many adaptive techniques to achieve high resolution and
excellent interference rejection capability, and 3) the probing
signals transmitted via its antennas can be optimized to obtain
several transmit beampattern designs with superior performance.
For example, the covariance matrix of the probing signal vector
transmitted by the MIMO radar can be optimized to maximize
the power around the locations of the targets of interest and also
to minimize the cross-correlation of the signals reflected back to
the radar by these targets, thereby significantly improving the
performance of adaptive MIMO radar techniques. Additionally,
we demonstrate the advantages of several MIMO transmit
beampattern designs, including a beampattern matching design
and a minimum sidelobe beampattern design, over their phased-
array counterparts.
I. I NTRODUCTION
MIMO radar is an emerging technology that is attracting the
attention of researchers and practitioners alike. A MIMO radar
system, unlike a standard phased-array radar, can transmit via
its antennas multiple probing signals that may be quite dif-
ferent but correlated with each other. This waveform diversity
enables superior capabilities compared with a standard phased-
array radar; see, e.g., [1][7].
We provide herein an overview of our recent results showing
that this waveform diversity enables the MIMO radar superior-
ity in several fundamental aspects. Without loss of generality,
we consider targets associated with a particular range bin and
a single pulse is transmitted by each transmit antenna. Targets
in adjacent range bins contribute as interferences to the range
bin of interest.
First, we address one of the most basic issues of MIMO
radar its parameter identifiability, which is the maximum
number of targets that can be uniquely identified by the radar.
We show that the waveform diversity afforded by MIMO radar
enables a much improved parameter identifiability over its
Jian Li is with the Department of Electrical and Computer Engineering,
P.O. Box 116130, University of Florida, Gainesville, FL 32611-6130, USA;
Phone: (352) 392-2642, Fax: (352) 392-0044, Email: li@dsp.ufl.edu.
Petre Stoica is with the Department of Information Technology, Uppsala
University, Uppsala, Sweden.
phased-array counterpart, i.e., the maximum number of targets
that can be uniquely identified by the MIMO radar is up to
Mt times that of its phased-array counterpart, where Mt is
the number of transmit antennas. The parameter identifiability
is further demonstrated in a numerical study using both the
Cramer-Rao bound (CRB) and a least-squares method for
target parameter estimation.
Second, we consider an adaptive MIMO radar scheme that
can be used to deal with multiple targets. Linearly independent
waveforms can be transmitted simultaneously via the multiple
transmit antennas of a MIMO radar. Due to the different
phase shifts associated with the different propagation paths
from the transmitting antennas to targets, these independent
waveforms are linearly combined at the targets with different
phase factors. As a result, the signal waveforms reflected from
different targets are linearly independent of each other, which
allows for the direct application of Capon and of other adaptive
array algorithms. We consider herein applying the Capon
algorithm to estimate the target locations and an approximate
maximum likelihood (AML) method recently introduced in [8]
to determine the reflected signal amplitudes.
Finally, we show that the probing signal vector transmitted
by a MIMO radar system can be designed to approximate a
desired transmit beampattern and also to minimize the cross-
correlation of the signals bounced from various targets of
interest an operation that would be hardly possible for
a phased-array radar. An efficient Semi-definite Quadratic
Programming (SQP) algorithm can be used to solve the signal
design problem in polynomial time. Using this design, we can
significantly improve the parameter estimation accuracy of the
adaptive MIMO radar techniques. In addition, we consider a
minimum sidelobe beampattern design. We demonstrate the
advantages of these MIMO transmit beampattern designs over
their phased-array counterparts. Due to the significantly larger
number of degrees of freedom of a MIMO system, we can
achieve much better transmit beampatterns with a MIMO
radar, under the practical uniform elemental transmit power
constraint, than with its phased-array counterpart.
II. P ROBLEM F ORMULATION
Consider a MIMO radar system with Mt transmit antennas
and Mr receive antennas. Let xm (n) denote the discrete-
time baseband signal transmitted by the mth transmit antenna.
Also, let denote the location parameter(s) of a generic
target, for example, its azimuth angle and its range. Then,
under the assumption that the transmitted probing signals are
narrowband and that the propagation is non-dispersive, the
baseband signal at the target location can be described by the
 2

expression (see, e.g., [3] and Chapter 6 in [9]): Using the results of [11], [12] we can show that a sufficient
Mt and almost necessary condition for parameter identifiability
X 4
ej2f0 m () xm (n) = a ()x(n), n = 1, , N, (1) is:
Mt + Mr Mt Mr + 1
m=1 Kmax , , (7)
2 2
where f0 is the carrier frequency of the radar, m () is the
time needed by the signal emitted via the mth transmit antenna depending on the array geometry and on how many antennas
to arrive at the target, () denotes the conjugate transpose, that the transmit and receive arrays share [10].
N denotes the number of samples of each transmitted signal Furthermore, generically speaking, (i.e., for almost any vec-
pulse, tor ), the identifiability can be ensured under the following
T weaker condition [10], [11]:
x(n) = x1 (n) x2 (n) xMt (n) , (2)
2(Mt + Mr 1) 2Mt Mr
Kmax , . (8)
and 3 3
T
a() = ej2f0 1 () ej2f0 2 () ej2f0 Mt ()
, For a phased-array radar, the condition similar to (7) is
(3)
Mr + 1
with ()T denoting the transpose. By assuming that the trans- Kmax < , (9)
2
mit array of the radar is calibrated, a() is a known function
of . and that similar to (8) is
Let ym (n) denote the signal received by the mth receive 2Mr
antenna; let Kmax < . (10)
3
T
y(n) = y1 (n) y2 (n) yMr (n) , n = 1, , N, Hence, the maximum number of targets that can be uniquely
(4) identified by a MIMO radar can be up to Mt times that of
and let its phased-array counterpart. When a ULA is used for both
T transmitting and receiving, which appears to be the worst
b() = ej2f0 1 () ej2f0 2 () ej2f0 Mr () , MIMO radar scenario from the parameter identifiability stand-
(5) point, the maximum number of targets that can be identified
where m () is the time needed by the signal reflected by the by the MIMO radar is about twice that of its phased-array
target located at to arrive at the mth receive antenna. Then, counterpart.
under the simplifying assumption of point targets, the received We present several numerical examples to demonstrate the
data vector can be described by the equation (see, e.g., [6], parameter identifiability of MIMO radar, as compared to its
[7]): phased-array counterpart. First, consider a MIMO radar system
K
X where a ULA with M = Mt = Mr = 10 antennas and half-
y(n) = k bc (k )a (k )x(n) + (n), n = 1, , N, wavelength spacing between adjacent antennas is used both for
k=1 transmitting and for receiving. The transmitted waveforms are
(6)
orthogonal to each other. Consider a scenario in which K tar-
where K is the number of targets that reflect the signals back
gets are located at 1 = 0 , 2 = 10 , 3 = 10 , 4 = 20 ,
to the radar receiver, {k } are the complex amplitudes propor-
5 = 20 , 6 = 30 , 7 = 30 , , with identical complex
tional to the radar-cross-sections (RCSs) of those targets, {k }
amplitudes 1 = = K = 1. The number of snapshots
are their location parameters, (n) denotes the interference-
is N = 256. The received signal is corrupted by a spatially
plus-noise term, and ()c denotes the complex conjugate. The
and temporally white circularly symmetric complex Gaussian
unknown parameters, to be estimated from {y(n)}N n=1 , are noise with mean zero and variance 0.01 (i.e., SNR=20 dB)
{k }K K
k=1 and {k }k=1 . and by a jammer located at 45 with an unknown waveform
(uncorrelated with the waveforms transmitted by the radar)
III. PARAMETER I DENTIFIABILITY with a variance equal to 1 (i.e., INR = 20 dB).
Parameter identifiability is basically a consistency aspect: Consider the Cramer-Rao bound (CRB) of {k }, which
we want to establish the uniqueness of the solution to gives the best performance of an unbiased estimator. By as-
the parameter estimation problem as either the signal-to- suming that {(n)}N n=1 in (6) are independently and identically
interference-plus-noise ratio (SINR) goes to infinity or the distributed (i.i.d.) circularly symmetric complex Gaussian ran-
snapshot number N goes to infinity [10]. It is clear that in dom vectors with mean zero and unknown covariance Q,
either case, assuming that the interference-plus-noise term the CRB for {k } can be obtained using the Slepian-Bangs
(n) is uncorrelated with x(n), the identifiability property formula [9]. Figure 1(a) shows the CRB of 1 for the MIMO
of the first term in (6) is not affected by the second term. radar as a function of K. For comparison purposes, we also
In particular, it follows that asymptotically we can handle provide the CRB of its phased-array counterpart, for which all
any number of interferences; of course, for a finite snapshot the parameters are the same as for the MIMO radar except that
number N and a finite SINR, the accuracy will degrade as the Mt = 1 and that the amplitude of the transmitted waveform
number of interferences increases, but that is a different issue is adjusted so that the total transmission power does not
the parameter identifiability is not affected. change. Note that the phased-array CRB increases rapidly as
 3

1 1
10 10
PhasedArray PhasedArray
MIMO Radar MIMO Radar
0 0
10 10

1 1
10 10
CRB

CRB
2 2
10 10

3 3
10 10

4 4
10 10

2 4 6 8 10 12 2 4 6 8 10 12 14 16
K K

(a) (a)
1.5 2

1.8
Modulus of Complex Amplitude

Modulus of Complex Amplitude

1.6

1.4
1
1.2

0.8
0.5
0.6

0.4

0.2

0 0
80 60 40 20 0 20 40 60 80 80 60 40 20 0 20 40 60 80
Angle (deg) Angle (deg)

(b)
Fig. 1. Performance of a MIMO radar system where a ULA with M = 10
antennas and 0.5-wavelength inter-element spacing is used for both transmit-
ting and receiving. (a) Cramer-Rao bound of 1 versus K and (b) LS spatial
spectrum when K = 12.
(b)
Fig. 2. Performance of a MIMO radar system with Mt = Mr = 5 antennas,
and with half-wavelength inter-element spacing for the receive ULA and
2.5-wavelength inter-element spacing for the transmit ULA. (a) Cramer-Rao
bound of 1 versus K and (b) LS spatial spectrum when K = 16.

K increases from 1 to 6. The corresponding MIMO CRB, IV. A DAPTIVE MIMO R ADAR
however, is almost constant when K is varied from 1 to 12 (but
The signals reflected back by the targets due to MIMO
becomes unbounded for K > 12). Both results are consistent
probing can be linearly independent of each other, which is
with the parameter identifiability analysis (see [10] for details):
another significant advantage of MIMO radar. Let
Kmax 6 for the phased-array radar and Kmax 12 for the

MIMO radar. = 1 a(1 ) 2 a(2 ) a(K ) .
A (11)
K
We next consider a simple semi-parametric least-squares
(LS) method [6] for MIMO radar parameter estimation. Figure Then the sample covariance matrix of the target
1(b) shows the LS spatial spectrum as a function , when reflected is A R
xx A,
where R xx =
PN waveforms
K = 12. Note that all 12 target locations can be approximately (1/N ) n=1 x(n)x (n). When orthogonal waveforms
determined from the peak locations of the LS spatial spectrum. are used for MIMO probing, for example, and N Mt ,
Consider now a MIMO radar system with Mt = Mr = 5 R xx is a scaled identity matrix. Then A R xx A
has full
antennas. The distance between adjacent antennas is 0.5- rank, i.e., the target reflected waveforms are not completely
wavelength for the receiving ULA and 2.5-wavelength for correlated with each other (or coherent), if the columns of A
the transmitting ULA. We retain all the simulation parameters are linearly independent of each other, which requires that
corresponding to Figure 1 except that the targets are located K Mt . The fact that the target reflected waveforms are
at 1 = 0 , 2 = 8 , 3 = 8 , 4 = 16 , 5 = 16 , non-coherent allows the direct application of many adaptive
6 = 24 , 7 = 24 , in this example. Figure 2(a) shows techniques for target localization [6].
the CRB of 1 , for both the MIMO radar and the phased-array We demonstrate the performance of the Capon method for
counterpart, as a function of K. Again, the MIMO CRB is target localization. Consider the a scenario of a MIMO radar
much lower than the phased-array CRB. The behavior of both with a uniform linear array (ULA) comprising M = Mt =
CRBs is consistent with the parameter identifiability analysis: Mr = 10 antennas with half-wavelength spacing between
Kmax 3 for the phased-array radar and Kmax 16 for the adjacent antennas. This array is used both for transmitting
MIMO radar. Moreover, the parameters of all K = 16 targets and for receiving. Without loss of generality, the total transmit
can be approximately determined with the simple LS method, power is set to 1. Assume that K = 3 targets are located at
as shown in Figure 2(b). 1 = 40 , 2 = 0 , and 3 = 40 with complex amplitudes
 4

2
V. P ROBING S IGNAL D ESIGN
The power of the probing signal at a generic focal point
Capon Spectrum 1.5 with location is given by (see (1)):
P () = a ()Ra(), (12)
1
where R is the covariance matrix of x(n), i.e.,
0.5 R = E{x(n)x (n)}. (13)
The spatial spectrum in (12), as a function of , will be
0
50 0 50
called the transmit beampattern.
Angle (degree) The first problem we will consider in this section consists
(a) of choosing R, under a uniform elemental power constraint,
2 c
Rmm = , m = 1, , M ; with c given, (14)
M
1.5 where M is a short notation for Mt , Rmm denotes the
(m, m)th element of R, to achieve the following goals:
(a) Control the spatial power at a number of given target
GLRT

1
locations by matching (or approximating) a desired
transmit beampattern.
0.5 (b) Minimize the cross-correlation between the prob-
ing signals at a number of given target locations;
note from (1) that the cross-correlation between the
probing signals at locations and is given by
0
50 0 50
Angle (degree)
a ()Ra().
(b)
Let () denote a desired transmit beampattern, and let
Fig. 3. The Capon spatial spectrum and the GLRT pseudo-spectrum as {l }Ll=1 be a fine grid of points that cover the location sectors
functions of , for the initial omnidirectional probing. (a) Capon and (b)
GLRT. of interest. We assume that the said grid contains points which

are good approximations of the locations {k }K k=1 of the
targets of interest, and that we dispose of (initial) estimates

{k }K K
k=1 of {k }k=1 , where K denotes the number of targets
equal to 1 = 2 = 3 = 1. There is a strong jammer at 25 of interest that we wish to probe further. We can obtain ()

with an unknown waveform (uncorrelated with the transmitted and {k }K k=1 using the Capon and GLRT approaches presented
MIMO radar waveforms) with a power equal to 106 (60 dB). in Section IV.
Each transmitted signal pulse has N = 256 samples. The As stated above, our goal is to choose R such that the
received signal is also corrupted by a zero-mean circularly transmit beampattern, a ()Ra(), matches or rather approx-
symmetric spatially and temporally white Gaussian noise with imates (in a least squares (LS) sense) the desired transmit
variance 2 . beampattern, (), over the sectors of interest, and also such
Since we do not assume any prior knowledge about the that the cross-correlation (beam)pattern, a ()Ra() (for 6=
target locations, orthogonal waveforms are used for MIMO
), is minimized (once again, in a LS sense) over the set

probing. (We refer to this as initial probing, since after we {k }K
k=1 . Mathematically, we want to solve the following
get the target location estimates with this probing, we can op- problem:
timize the transmitted beampattern to improve the estimation ( L
accuracy, as shown in Section V.) Using the data collected as 1X 2
min wl [(l ) a (l )Ra(l )]
a result of this initial probing, we can obtain the Capon spatial ,R L
l=1
spectrum and the generalized likelihood ratio test (GLRT)

2wc X X 2
K1 K
function [6]. An example of the Capon spectrum for 2 = 10
+ a (k )Ra(p )
dB is shown in Figure 3(a), where very narrow peaks occur K2 K
k=1 p=k+1
around the target locations. Note that in Figure 3(a), a false
c
peak occurs around = 25 due to the presence of the very s.t. Rmm = , m = 1, , M
M
strong jammer. The corresponding GLRT pseudo-spectrum as R 0, (15)
a function of is shown in Figure 3(b). Note that the GLRT
is close to one at the target locations and close to zero at where is a scaling factor, wl 0, l = 1, , L, is the weight
any other locations including the jammer location. Therefore, for the lth grid point and wc 0 is the weight for the cross-
the GLRT can be used to reject the jammer peak in the Capon correlation term. The value of wl should be larger than that
spectrum. The remaining peak locations in the Capon spectrum of wk if the beampattern matching at l is considered to be
are the estimated target locations. more important than the matching at k . Note that by choosing
 5

0
4.5 10
wc=0
4 wc=1
3.5
2
10
3
Beampattern

2.5

MSE
4
2 10
1.5

1
6
10
0.5 Optimal Beampattern Matching
Omnidirectional Beampattern
0
50 0 50 20 10 0 10 20
Angle (degree) Reciprocal of Noise Level (dB)

Fig. 4. MIMO beampattern matching designs for = 5 and c = 1. The (a)

beampatterns are obtained using wc = 0 or wc = 1. 10
2

maxl wl > wc we can give more weight to the first term in 10

4

the design criterion above, and viceversa for maxl wl < wc .

We show in [13] that this design problem can be efficiently

MSE
6
10
solved in polynomial time as a semi-definite quadratic program
(SQP).
8
To illustrate the beampattern matching design, consider the 10

example considered in Figure 3. The initial target location Optimal Beampattern Matching
estimates obtained using Capon or GLRT can be used to derive 10
Omnidirectional Beampattern
10
a desired beampattern. In the following numerical examples, 20 10 0 10 20
Reciprocal of Noise Level (dB)
we form the desired beampattern by using the dominant peak
locations of the GLRT pseudo-spectrum, denoted as 1 , , (b)
K , as follows (with K being the resulting estimate of K, and Fig. 5. MSEs of (a) the location estimates and of (b) the complex
amplitude estimates for the first target, as functions of 10 log10 2 , obtained
K = K): with initial omnidirectional probing and with probing using the beampattern
matching design for = 5 , wc = 1, and c = 1.
1, [k , k + ], k = 1, , K,

() = (16)
0, otherwise,
where 2 is the chosen beamwidth for each target ( should (b) Achieve a predetermined 3 dB main-beam width.
be greater than the expected error in {k }). Figure 4 is This problem can be formulated as follows:
obtained using (16) with = 5 in the beampattern matching
design in (15) along with a mesh grid size of 0.1 , wl = 1, min t
t,R
l = 1, , L, and either wc = 0 or wc = 1. Note that the
s.t. a (0 )Ra(0 ) a (l )Ra(l ) t, l
designs obtained with wc = 1 and with wc = 0 are similar
to one another. However, the cross-correlation behavior of the a (1 )Ra(1 ) = 0.5a (0 )Ra(0 )
former is much better than that of the latter in that the reflected a (2 )Ra(2 ) = 0.5a (0 )Ra(0 )
signal waveforms corresponding to using wc = 1 are almost R0
uncorrelated with each other. c
Rmm = , m = 1, , M, (17)
Next, we examine the MSEs of the location estimates M
obtained by Capon and of the complex amplitude estimates where 2 1 (with 2 > 0 and 1 < 0 ) determines the
obtained by AML [8]. In particular, we compare the MSEs ob- 3 dB main-beam width and denotes the sidelobe region of
tained using the initial omnidirectional probing with those ob- interest. As shown in [13], this minimum sidelobe beampattern
tained using the optimal beampattern matching design shown design problem can be efficiently solved in polynomial time
in Figure 4 with = 5 and wc =1. Figures 5(a) and 5(b) as a semi-definite program (SDP).
show the MSE curves of the location and complex amplitude Finally, consider the conventional phased-array beampattern
estimates obtained for the target at 40 from 1000 Monte- design problem in which only the array weight vector can
Carlo trials (the results for the other targets are similar). The be adjusted and therefore all antennas transmit the same
estimates obtained using the optimal beampattern matching differently scaled waveform. We can readily modify the pre-
design are much better: the SNR gain over the omnidirectional viously described beampattern matching or minimum sidelobe
design is larger than 10 dB. beampattern designs for the case of phased-arrays by adding
Another beampattern design problem we consider consists the constraint rank(R) = 1. However, due to the rank-one
of choosing R, under the uniform elemental power constraint constraint, both these originally convex optimization problems
in (14), to achieve the following goals: become non-convex. The lack of convexity makes the rank-one
(a) Minimize the sidelobe level in a prescribed region. constrained problems much harder to solve than the original

convex optimization problems [14]. Semi-definite relaxation
(SDR) is often used to obtain approximate solutions to such
rank-constrained optimization problems [15]. Typically, the
SDR is obtained by omitting the rank constraint. Hence,
interestingly, the MIMO beampattern design problems are
SDRs of the corresponding phased-array beampattern design
problems.
In the numerical examples below, we have used the Newton-
like algorithm presented in [14] to solve the rank-one con-
strained design problems for phased-arrays. The said algorithm
uses SDR to obtain an initial solution, which is the exact solu-
tion to the corresponding MIMO beampattern design problem.
Although the convergence of the Newton-like algorithm is not
guaranteed [14], we did not encounter any apparent problem
in our numerical simulations.
Consider the minimum sidelobe beampattern design prob-
lem in (17), with the main-beam centered at 0 = 0 , with a 3
dB width equal to 20 (2 = 1 = 10 ), and with c = 1, for
the same MIMO radar scenario as the one considered in Figure
3. The sidelobe region is = [90 , 20 ] [20 , 90 ].
The MIMO minimum-sidelobe beampattern design is shown
in Figure 6(a). Note that the peak sidelobe level achieved
by the MIMO design is approximately 18 dB below the
mainlobe peak level. Figure 6(b) shows the corresponding
phased-array beampattern obtained by using the additional
constraint rank(R) = 1. The phased-array design fails to
provide a proper mainlobe (it suffers from peak splitting)
and its peak sidelobe level is much higher than that of its
MIMO counterpart. We note that, under the elemental power
constraint, the number of degrees of freedom (DOF) of the
phased-array that can be used for beampattern design is
equal to only M 1 (real-valued parameters); consequently,
it is difficult for the phased-array to synthesize a proper
beampattern. The MIMO design, on the other hand, can be
used to achieve a much better beampattern due to its much
larger number of DOF, viz. M 2 M .
VI. C ONCLUSIONS
We have shown that the waveform diversity offered by a
MIMO radar system enables significant superiority over its
phased-array counterpart. We hope that this overview of our
recent results on the MIMO radar, along with the related
results obtained by our colleagues, will stimulate the interest
deserved by this topic in both academia and government
agencies as well as industry.
R EFERENCES
[1] E. Fishler, A. Haimovich, R. Blum, D. Chizhik, L. Cimini, and R. Valen-
zuela, MIMO radar: an idea whose time has come, Proceedings of the
IEEE Radar Conference, pp. 7178, April 2004.
[2] E. Fishler, A. Haimovich, R. Blum, L. Cimini, D. Chizhik, and
R. Valenzuela, Performance of MIMO radar systems: advantages of
angular diversity, 38th Asilomar Conference on Signals, Systems and
Computers, Pacific Grove, CA, vol. 1, pp. 305309, Nov. 2004.
[3] D. R. Fuhrmann and G. San Antonio, Transmit beamforming for MIMO
radar systems using partial signal correlation, 38th Asilomar Conference
on Signals, Systems and Computers, Pacific Grove, CA, vol. 1, pp. 295
299, Nov. 2004.
6
20
10
Beampattern (dB)
0
10
20
30
50 0 50
Angle (degree)
(a)
20
10
Beampattern (dB)
0
10
20
30
50 0 50
Angle (degree)
(b)
Fig. 6. Minimum sidelobe beampattern designs, under the uniform elemental
power constraint, when the 3 dB main-beam width is 20 . (a) MIMO and (b)
phased-array.
[4] F. Robey, S. Coutts, D. Weikle, J. McHarg, and K. Cuomo, MIMO radar
theory and experimental results, 38th Asilomar Conference on Signals,
Systems and Computers, Pacific Grove, CA, vol. 1, pp. 300304, Nov.
2004.
[5] K. Forsythe and D. Bliss, Waveform correlation and optimization issues
for MIMO radar, 39th Asilomar Conference on Signals, Systems and
Computers, Pacific Grove, CA, pp. 13061310, Nov. 2005.
[6] L. Xu, J. Li, and P. Stoica, Radar imaging via adaptive MIMO
techniques, EUSIPCO (invited), Florence, Italy, 2006. (Available:
ftp://www.sal.ufl.edu/xuluzhou/EUSIPCO2006.pdf).
[7] E. Fishler, A. Haimovich, R. Blum, L. Cimini, D. Chizhik, and R. Valen-
zuela, Spatial diversity in radars - models and detection performance,
IEEE Transactions on Signal Processing, vol. 54, pp. 823838, March
2006.
[8] L. Xu, P. Stoica, and J. Li, A diagonal growth curve model and
some signal processing applications, to appear in IEEE Transactions on
Signal Processing, (available: ftp://www.sal.ufl.edu/xuluzhou/DGC.pdf).
[9] P. Stoica and R. Moses, Spectral Analysis of Signals. Upper Saddle
River, NJ: Prentice Hall, 2005.
[10] J. Li, P. Stoica, L. Xu, and W. Roberts, On parameter identifiability of
MIMO radar, IEEE Signal Processing Letters, submitted.
[11] M. Wax and I. Ziskind, On unique localization of multiple sources by
passive sensor arrays, IEEE Transactions on Acoustics, Speech, and
Signal Processing, pp. 9961000, July 1989.
[12] A. Nehorai, D. Starer, and P. Stoica, Direction-of-arrival estimation in
applications with multipath and few snapshots, Circuits, Systems, and
Signal Processing, vol. 10, pp. 327342, 1991.
[13] J. Li, P. Stoica, and Y. Xie, On probing signal design for MIMO radar,
40th Asilomar Conference on Signals, Systems and Computers (invited),
Pacific Grove, CA, 2006.
[14] R. Orsi, U. Helmke, and J. B. Moore, A Newton-like method for solving
rank constrained linear matrix inequalities, Proceedings of the 43rd
IEEE Conference on Decision and Control, pp. 31383144, 2004.
[15] S. Boyd and L. Vandenberghe, Convex Optimization. Cambridge, UK:
Cambridge University Press, 2004.