Lecture PPT Stocia

Periodogram
and
Correlogram
Methods
Lecture 2
Lecture notes to accompany Introduction to Spectral Analysis Slide L2–1

by P. Stoica and R. Moses, Prentice Hall, 1997
Periodogram
Recall 2nd definition of (! ):
N
8
2 9
(!) = Nlim E 1 y (t )e ;i!t

>
<
X

>
=
!1 N t=1 >
:

>
;
Given : fy(t)gNt=1
Drop “ lim
N !1
” and “E fg” to get

N

2
^p(!) = N1

X
y(t)e;i!t

t=1

Natural estimator
Used by Schuster (1900) to determine “hidden
periodicities” (hence the name).

Correlogram
Recall 1st definition of (! ):
1
(!) = r(k)e;i!k
X
k=;1
Truncate the “ ” and replace “r
P
(k)” by “^r(k)”:
;1
NX
^c(!) = ^r(k)e;i!k
k=;(N ;1)

Covariance Estimators
(or Sample Covariances)
Standard unbiased estimate:

1 N
^r(k) = N ; k y(t)y(t ; k); k 0
X
t=k+1
Standard biased estimate:

1 N
r^(k) = N y(t)y(t ; k); k 0
X
t=k+1
For both estimators:
^r(k) = ^r(;k); k < 0

Relationship Between
^ (! ) and
p ^ (! )
c
If: the biased ACS estimator r ^(k) is used in ^c(!),

Then:

N

2
^p(!) = 1

y (t )e ;i!tX

N t=1

;1
NX
= ^r(k)e;i!k
k=;(N ;1)
= ^c(!)
^p(!) = ^c(!)
Consequence:
^( )
Both p ! and c ^ (!) can be analyzed simultaneously.

Statistical Performance of
^ (! ) and
p ^ (! )
c
Summary:
Both are asymptotically (for large N ) unbiased:

E ^p(!) ! (!) as N ! 1
n o
Both have “large” variance, even for large N .
Thus, p ^ (!) and ^c(!) have poor performance.

Intuitive explanation:
r^(k) ; r(k) may be large for large jkj
Even if the errors f^r(k) ; r(k)gNjkj;=01 are small,

there are “so many” that when summed in
[ ^ ( ) ; ( )]
p ! ! , the PSD error is large.
Bias Analysis of the Periodogram
;1
NX
E ^p(!) = E ^c(!) = E f^r(k)g e;i!k
n o n o
k=;(N ;1)
N ;1 j k j !
= 1 ; N r(k)e;i!k
X
k=;(N ;1)
1
= wB (k)r(k)e;i!k
X
k=;1
8
1 ; jNkj ; jkj N ; 1

<
wB (k) = :
0; jkj N
= Bartlett, or triangular, window
Thus,
^ 1 Z
E p(!) = 2 ; ( )WB (! ; ) d

n o
Ideally: WB (!) = Dirac impulse (!).

Bartlett Window WB (! )
1 sin( !N= 2)
"
2 #
WB (!) = N sin(!=2)
WB (!)=WB (0), for N = 25

0
−10
−20
dB
−30
−40
−50
−60
−3 −2 −1 0 1 2 3
ANGULAR FREQUENCY
Main lobe 3dB width 1=N .

For “small” N , WB (!) may differ quite a bit from (!).

Smearing and Leakage
Main Lobe Width: smearing or smoothing
Details in (!) separated in f by less than 1=N are not

resolvable.
φ(ω) ^φ(ω)
smearing
<1/Ν ω ω
Thus: Periodogram resolution limit = 1=N .

Sidelobe Level: leakage
φ(ω)≅δ(ω) ^φ(ω)≅WB(ω)
leakage
ω ω

Periodogram Bias Properties
Summary of Periodogram Bias Properties:
For “small” N , severe bias

As N ! 1, WB (!) ! (!),
^(!) is asymptotically unbiased.
so

Periodogram Variance
As N !1
E ^p(!1) ; (!1) ^p(!2) ; (!2)
nh ih io
2(!1); !1 = !2
(
= 0; !1 6= !2
Inconsistent estimate
Erratic behavior
^φ(ω)
^φ(ω)
asymptotic mean = φ(ω)
- 1 st. dev = φ(ω) too

+
ω
Resolvability properties depend on both bias and variance.

Improved Periodogram-Based
Methods
Lecture 3

Blackman-Tukey Method
Basic Idea: Weighted correlogram, with small weight

^( )
applied to covariances r k with “large” k . jj
;1
MX
^BT (!) = w(k)^r(k)e;i!k
k=;(M ;1)
fw(k)g = Lag Window
w(k)
1
-M M k

Blackman-Tukey Method, con't
1 Z
^BT (!) = 2 ; ^p( )W (! ; )d
W (!) = DTFT fw(k)g

= Spectral Window
Conclusion: BT ^ (!) = “locally” smoothed periodogram

Effect:
Variance decreases substantially

Bias increases slightly
By proper choice of M :
MSE = var + bias2 ! 0 as N ! 1

Window Design Considerations
Nonnegativeness:
1 Z
^BT (!) = 2 ; ^p( ) W (! ; )d

0
| {z }
If W (!) 0 (, w(k) is a psd sequence)

Then: ^BT (!) 0 (which is desirable)
Time-Bandwidth Product
;1
MX
w (k )
Ne = k=;(Mw(0)
;1) = equiv time width
1 W (!)d!
Z
2 ;
e = W (0) = equiv bandwidth
Ne e = 1
Window Design, con't
e = 1=Ne = 0(1=M )
is the BT resolution threshold.
As M increases, bias decreases and variance

increases.
) Choose M as a tradeoff between variance and

bias.
Once M is given, Ne (and hence e) is essentially

fixed.
) Choose window shape to compromise between

smearing (main lobe width) and leakage (sidelobe
level).
The energy in the main lobe and in the sidelobes

cannot be reduced simultaneously, once M is given.

Window Examples
Triangular Window, M
0
= 25
−10
−20
dB
−30
−40
−50
−60
−3 −2 −1 0 1 2 3
ANGULAR FREQUENCY
Rectangular Window, M
0
= 25
−10
−20
dB
−30
−40
−50
−60
−3 −2 −1 0 1 2 3
ANGULAR FREQUENCY

Bartlett Method
Basic Idea:
1 Μ 2Μ ... Ν
^φ1(ω) ^φ2(ω) ^φL(ω)

...
average
^φB(ω)
Mathematically:
yj (t) = y((j ; 1)M + t) t = 1; : : : ; M
= the j th subsequence
4 [N=M ])
(j = 1; : : : ; L =

M

2
^j (!) = 1

M
X
yj (t)e;i!t

t=1

1 L
^B (!) = L ^j (!)
X
j =1
Comparison of Bartlett
and Blackman-Tukey Estimates
8 9
1 L > ;1
MX >
^B (!) = L ^rj (k)e;i!k

X < =
j =1 k=;(M ;1)
>
: >
;
;1 < 1 X
8 9
MX L
^rj (k) e;i!k
=
=
k=;(M ;1) L j =1
: ;
M ;1
' ^r(k)e;i!k
X
k=;(M ;1)
Thus:
^B (!) ' ^BT (!) with a rectangular
lag window wR k ()
^( )
Since B ! implicitly uses fwR(k)g, the Bartlett
method has
High resolution (little smearing)

Large leakage and relatively large variance
Welch Method
Similar to Bartlett method, but
allow overlap of subsequences (gives more

subsequences, and thus “better” averaging)
use data window for each periodogram; gives

mainlobe-sidelobe tradeoff capability
1 2 N
. . .
subseq subseq subseq
#1 #2 #S
=
Let S # of subsequences of length M .
(Overlapping means S > N=M “better[ ])
averaging”.)
Additional flexibility:
The data in each subsequence are weighted by a temporal

window
Welch is approximately equal to BT ^ (!) with a

non-rectangular lag window.
Daniell Method
By a previous result, for N 1,

f^p(!j )g are (nearly) uncorrelated random variables for
2 N ;1
! = j N j =0 j
Idea: “Local averaging” of (2J + 1) samples in the
frequency domain should reduce the variance by about
(2 + 1)
J .
1 k +J
^D (!k ) = 2J + 1 ^p (!j )
X
j =k;J

Daniell Method, con't
As J increases:
Bias increases (more smoothing)

Variance decreases (more averaging)
Let = 2J=N . Then, for N 1,

^D(!) ' 1 ^p(!)d!
Z
2 ;
^D (!) ' ^BT (!) with a rectangular spectral
Hence:
window.

Summary of Periodogram Methods
Unwindowed periodogram
– reasonable bias
– unacceptable variance
Modified periodograms
– Attempt to reduce the variance at the expense of (slightly)
increasing the bias.
BT periodogram
– Local smoothing/averaging of p ^ (!) by a suitably selected
spectral window.
– Implemented by truncating and weighting r ^(k) using a lag
window in c ! ^( )
Bartlett, Welch periodograms
^ ()
– Approximate interpretation: BT ! with a suitable lag
window (rectangular for Bartlett; more general for Welch).
– Implemented by averaging subsample periodograms.
Daniell Periodogram
– Approximate interpretation: ^BT (!) with a rectangular
spectral window.
– Implemented by local averaging of periodogram values.

Parametric Methods
for
Rational Spectra
Lecture 4

Basic Idea of Parametric Spectral Estimation
Observed
Data
Estimate ^θ ^
^φ(ω) =φ(ω,θ)
Estimate
parameters
PSD
Assumed in φ(ω,θ)
functional
form of φ(ω,θ)
possibly revise assumption on φ(ω)
Rational Spectra
j k jm k e;i!k
P
(!) = P
jkjn k e;i!k
(!) is a rational function in e;i! .
( )
By Weierstrass theorem, ! can approximate arbitrarily
well any continuous PSD, provided m and n are chosen
sufficiently large.
Note, however:
choice of m and n is not simple

some PSDs are not continuous
AR, MA, and ARMA Models
By Spectral Factorization theorem, a rational (!) can be

factored as
B ( ! ) 2
(!) = A(!) 2

A(z) = 1 + a1z;1 + + anz;n

B (z) = 1 + b1z;1 + + bmz;m
and, e.g., A(! ) = A(z )jz =ei!
Signal Modeling Interpretation:

e(t) - B (q) y(t)-
e(!) = 2 A(q) y (!) = B (!) 2 2

A(!)

white noise ltered white noise
ARMA: A(q)y(t) = B (q)e(t)

AR: A(q)y(t) = e(t)
MA: y(t) = B (q)e(t)

ARMA Covariance Structure
ARMA signal model:

n m
y(t)+ aiy(t ; i) = bj e(t ; j ); (b0 = 1)
X X
i=1 j =0
Multiply by y (t ; k) and take E fg to give:

n m
r (k ) + air(k ; i) = bj E fe(t ; j )y(t ; k)g
X X
i=1 j =0
m
= 2 bj hj ;k
X
j =0
= 0 for k > m
( ) 1
B q
where H (q ) = A(q ) = hk q;k ; (h0 = 1)
X
k=0

AR Signals: Yule-Walker Equations
AR: m = 0.
Writing covariance equation in matrix form for
k =1 : : : n:
r(0) r(;1) : : : r(;n)
2 32
1 3
22 3
r(1) r(0)
6
6
.. 7
7
6
6 a1 = 0
7
7
6
6
7
7
6
4
.. . . . r (;1) 7
5
6
4
..7
5
6
..
4
7
5
r(n) : : : r(0) an 0
1 2 " # " #
R = 0
These are the Yule–Walker (YW) Equations.

MA Signals
MA: n =0
y(t) = B (q)e(t)
= e(t) + b1e(t ; 1) + + bme(t ; m)
Thus,
r(k) = 0 for jkj > m
and
m
2 2
(!) = jB (!)j = r(k)e;i!k
X
k=;m

MA Spectrum Estimation
Two main ways to Estimate (! ):
1. Estimate fbkg and 2 and insert them in

(!) = jB (!)j22
nonlinear estimation problem
^(!) is guaranteed to be 0
2. Insert sample covariances f^r(k)g in:

m
(!) = r(k)e;i!k
X
k=;m
^ (!) with a rectangular lag window of
This is BT
2 + 1.
length m
^(!) is not guaranteed to be 0
Both methods are special cases of ARMA methods

described below, with AR model order n . =0
ARMA Signals
ARMA models can represent spectra with both peaks

(AR part) and valleys (MA part).
A(q)y(t) = B (q)e(t)
B (! ) 2 m

e

;i!k P
(!) = 2 A(!) = jA(!)j2

k

= ;m k

where
k = E f[B (q)e(t)][B (q)e(t ; k)]g

= E f[A(q)y(t)][A(q)y(t ; k)]g
n n
= aj ap r(k + p ; j )
X X
j =0 p=0

ARMA Spectrum Estimation
Two Methods:
2
(! )
1. Estimate fai; bj ; 2g in (! ) = 2 A(! )
B

nonlinear estimation problem; can use an approximate

linear two-stage least squares method
^(!) is guaranteed to be 0
m ;i!k
k=;m k e
P
2. Estimate fai; r(k)g in (!) = jA(!)j2

linear estimation problem (the Modified Yule-Walker
method).
^(!) is not guaranteed to be 0

Modified Yule-Walker Method
ARMA Covariance Equation:

n
r(k) + air(k ; i) = 0; k > m
X
i=1
In matrix form for k = m + 1; : : : ; m + M
2 3 2 3
r(m) : : : r(m ; n + 1) "
a1 # r(m + 1)
6 r(m + 1) r(m ; n + 2) 7 .. = ;6 r(m + 2) 7
4 .. ... .. 5 . 4 .. 5
.
r(m + M ; 1) : : : r(m ; n + M )
.
an .
r(m + M )
Replace fr(k)g by f^r(k)g and solve for faig.
If M =
n, fast Levinson-type algorithms exist for
obtaining ai . f^ g
If M > n overdetermined Y W system of equations; least
squares solution for ai . f^ g
Note: For narrowband ARMA signals, the accuracy of
f^ g
ai is often better for M > n

Summary of Parametric Methods for Rational Spectra
Computational Guarantee
Method Burden Accuracy ^(!) 0 ?
Use for
AR: YW or LS low medium Yes Spectra with (narrow) peaks but
no valley
MA: BT low low-medium No Broadband spectra possibly with
valleys but no peaks
ARMA: MYW low-medium medium No Spectra with both peaks and (not
too deep) valleys
ARMA: 2-Stage LS medium-high medium-high Yes As above

Parametric Methods
for
Line Spectra — Part 1
Lecture 5

Line Spectra
Many applications have signals with (near) sinusoidal

components. Examples:
communications
radar, sonar
geophysical seismology
ARMA model is a poor approximation
Better approximation by Discrete/Line Spectrum Models
6
(!)
2 6(223)
6(21)
6(222)
2
; !1 !2 !3
- !
An “Ideal” line spectrum
Line Spectral Signal Model
Signal Model: Sinusoidal components of frequencies

!k and powers 2k , superimposed in white noise of
f g f g
power 2 .
y(t) = x(t) + e(t) t = 1; 2; : : :

n
x(t) = k ei(!k t+k)
X
k=1 xk (t)
| {z }
Assumptions:
A1: k > 0 !k 2 [;; ]

(prevents model ambiguities)
A2: f'k g = independent rv's, uniformly
distributed on [;; ]
(realistic and mathematically convenient)
A3: e(t) = circular white noise with variance 2

E fe(t)e(s)g = 2t;s E fe(t)e(s)g = 0
(can be achieved by “slow” sampling)

Covariance Function and PSD
Note that:
E
n
i' ;i'
e p e j = 1, for p = j
o
i' ;i' i'

E e pe j = E e p E e j
n o n o n
;i' o
2
1
Z
i'
= 2 ; e d' = 0, for p 6= j

Hence,
E xp(t)xj (t ; k) = 2p ei!pk p;j

n o
r(k) = E fy(t)y(t ; k)g

= np=1 2p ei!pk + 2k;0
P
and
n
(!) = 2 2p (! ; !p) + 2
X
p=1

Parameter Estimation
Estimate either:
f!k ; k; 'k gnk=1; 2 (Signal Model)
f!k ; 2k gnk=1; 2 (PSD Model)
Major Estimation Problem: f!

^ kg
Once f!^kg are determined:

f^2k g can be obtained by a least squares method from
n
r^(k) = 2p ei!^pk + residuals
X
p=1
OR:
Both f^kg and f'^kg can be derived by a least

squares method from
n
y(t) = k ei!^k t + residuals
X
k=1
with k = k ei'k .

Nonlinear Least Squares (NLS) Method
N

n

2
min y(t) ; kei(!k t+'k )
X X

f! ; ;' g
k k k t=1

k=1{z

| }
F (!;;')
Let:
k = k ei'k
= [1 : : : n]T
Y = [y(1) : : : y(N )]T
ei!1 ei!n
2 3
B = .. 6
4
.. 7
5
eiN!1 eiN!n

Nonlinear Least Squares (NLS) Method, con't
Then:
F = (Y ; B )(Y ; B ) = kY ; B k2
= [ ; (BB);1BY ][BB]
[ ; (BB);1BY ]
+Y Y ; Y B(BB);1BY
This gives:
^ = (B B );1B Y !=^!

and
!^ = arg max Y B (B B );1B Y
!

NLS Properties
Excellent Accuracy:
6 2
!k ) = N 32 (for N 1)
var (^
k
Example: N = 300
SNRk = 2 k = 2 = 30 dB
(^!k ) 10;5.
q
Then var
Difficult Implementation:
The NLS cost function F is multimodal; it is difficult to

avoid convergence to local minima.

Parametric Methods
for
Line Spectra — Part 2
Lecture 6

The Covariance Matrix Equation
Let:
a(!) = [1 e;i! : : : e;i(m;1)! ]T
A = [a(!1) : : : a(!n)] (m n)
Note: rank (A) = n (for m n)
Define
2
y(t) 3
4 6
y~(t) = 6 6
y(t ; 1) 7
7
7 = Ax~(t) + ~e(t)
4
.. 5
y(t ; m + 1)
where
x~(t) = [x1(t) : : : xn(t)]T
~e(t) = [e(t) : : : e(t ; m + 1)]T
Then
4
R= E fy~(t)~y(t)g = APA + 2I
with
21 0
2 3
P = E fx~(t)~x(t)g = 6 ... 7
2n
4 5
0
Eigendecomposition of R and Its Properties
R = APA + 2I (m > n)

Let:
1 2 : : : m: eigenvalues of R
fs1; : : : sng: orthonormal eigenvectors associated
with f1; : : : ; ng
fg1; : : : ; gm;ng: orthonormal eigenvectors associated

with fn+1; : : : ; m g
S = [s1 : : : sn] (m n)
G = [g1 : : : gm;n] (m (m ; n))
Thus,
1
2 3
S
" #
R = [S G] 6
4
... 7
5
G
m
Eigendecomposition of R and Its Properties,
con't
As rank (APA) = n:
k > 2 k = 1; : : : ; n
k = 2 k = n + 1; : : : ; m
1
2
; 2 0
3
= 6 ... = nonsingular 7
n ; 2
4 5
0
Note:
1 0
2 3
2
RS = APA S + S = S 4
6 ... 7
5
0 n
; 1 4
S = A(PA S ) = AC
with C j j 6= 0
(since rank S ( ) = rank(A) = n).
Therefore, since S G , =0
AG = 0

MUSIC Method
2

a (!1)
3
AG = 6
4
.. 7
5 G=0
a(!n)
) fa(!k )gnk=1 ? R(G)
Thus,
f!kgnk=1 are the unique solutions of

a(!)GGa(!) = 0.
Let:
1 N
R^ = N y~(t)~y(t)
X
t=m
S^; G^ = S; G made from the
^
eigenvectors of R

Spectral and Root MUSIC Methods
Spectral MUSIC Method:
f!^kgnk=1 = the locations of the n highest peaks of the

“pseudo-spectrum” function:
1 ; ! 2 [ ; ; ]
^ ^
a (!)GG a(!)
Root MUSIC Method:
f!^kgnk=1 = the angular positions of the n roots of:
aT (z;1)G^G^a(z) = 0
that are closest to the unit circle. Here,
a(z) = [1; z;1; : : : ; z;(m;1)]T
Note: Both variants of MUSIC may produce spurious

frequency estimates.
Spatial Methods — Part 1
Lecture 8

The Spatial Spectral Estimation Problem
vSource 2
Source 1 B
v B
@ ,,, B
v Source n
@
,,, B
,, BN ccc
,,@,@
R cc
c
, cc
c

cc
c
@@;;
@@;;
@@;;
Sensor 1
Sensor m
Sensor 2
Problem: Detect and locate n radiating sources by using

an array of m passive sensors.
Emitted energy: Acoustic, electromagnetic, mechanical
Receiving sensors: Hydrophones, antennas, seismometers
Applications: Radar, sonar, communications, seismology,
underwater surveillance
Basic Approach: Determine energy distribution over
space (thus the name “spatial spectral analysis”)

Simplifying Assumptions
Far-field sources in the same plane as the array of

sensors
Non-dispersive wave propagation
Hence: The waves are planar and the only location

parameter is direction of arrival (DOA)
(or angle of arrival, AOA).
The number of sources n is known. (We do not treat

the detection problem)
The sensors are linear dynamic elements with known

transfer characteristics and known locations
(That is, the array is calibrated.)

Array Model — Single Emitter Case
x(t) = the signal waveform as measured at a reference

point (e.g., at the “first” sensor)
k = the delay between the reference point and the

kth sensor
hk (t) = the impulse response (weighting function) of
sensor k
ek(t) = “noise” at the kth sensor (e.g., thermal noise in

sensor electronics; background noise, etc.)
Note: t 2 R (continuous-time signals).

Then the output of sensor k is
yk (t) = hk (t) x(t ; k) + ek (t)
( = convolution operator).
Basic Problem: Estimate the time delays fk g with hk (t)

known but x(t) unknown.
This is a time-delay estimation problem in the unknown

input case.
Narrowband Assumption
Assume: The emitted signals are narrowband with known

carrier frequency !c .
Then: x(t) = (t) cos[!ct + '(t)]

where (t); '(t) vary “slowly enough” so that
(t ; k ) ' (t); '(t ; k ) ' '(t)

Time delay is now ' to a phase shift !ck :
x(t ; k ) ' (t) cos[!ct + '(t) ; !ck ]

hk (t) x(t ; k )
' jHk (!c)j(t)cos[!ct + '(t) ; !ck + argfHk (!c)g]
where Hk (!) = Ffhk(t)g is the kth sensor's transfer

function
Hence, the kth sensor output is

yk (t) = jHk (!c)j(t)
cos[!ct + '(t) ; !ck + arg Hk (!c)] + ek(t)
Complex Signal Representation
The noise-free output has the form:
z(t) = (t) cos [!ct + (t)] =

= (2t) ei[!ct+ (t)] + e;i[!ct+ (t)]
n o
Demodulate z (t) (translate to baseband):
2z(t)e;!ct = (t)f ei (t) + e;i[2!ct+ (t)] g

| {z } | {z }
lowpass highpass
Lowpass filter 2z(t)e;i!ct to obtain (t)ei (t)
Hence, by low-pass filtering and sampling the signal
y~k (t)=2 = yk(t)e;i!ct

= yk(t) cos(!ct) ; iyk(t) sin(!ct)
we get the complex representation: (for t 2 Z )
yk (t) = (t) ei'(t) jHk (!c)j ei arg[H (! )] e;i! + ek (t)
| {z } | {z
k c
}
c k
s(t) Hk (!c)
or
yk (t) = s(t)Hk (!c) e;i!ck + ek (t)
where s(t) is the complex envelope of x(t).

Vector Representation for a Narrowband Source
Let
= the emitter DOA

m = the number of sensors
H1(!c) e;i!c1
2 3
a() = 6
4
.. 7
5
Hm(!c) e;i!cm
y1(t) e1(t)
2 3 2 3
y(t) = .. 6
4 e(t) = .. 7
5
6
4
7
5
ym(t) em(t)
Then
y(t) = a()s(t) + e(t)
NOTE: enters a via both () fkg and fHk(!c)g.

For omnidirectional sensors the fHk(!c)g do not depend
on .

Multiple Emitter Case
Given n emitters with
received signals: fsk(t)gnk=1

DOAs: k
Linear sensors )
y(t) = a(1)s1(t) + + a(n)sn(t) + e(t)
Let
A = [a(1) : : : a(n)]; (m n)
s(t) = [s1(t) : : : sn(t)]T ; (n 1)
Then, the array equation is:
y(t) = As(t) + e(t)
Use the planar wave assumption to find the dependence of

k on .
Uniform Linear Arrays
hSource
JJJ
J
JJ
JJJ^
J
J
JJ
-
JJ+

J] d sin
JJ
J^
JJ

JJ
v ]? v J v v vSensor
J ppp
1 2 JJ 3 4 m
d - J
ULA Geometry
Sensor #1 = time delay reference
Time Delay for sensor k:

d
k = (k ; 1) csin
where c = wave propagation speed

Spatial Frequency
Let:
!s = 4 ! d sin = 2 d sin = 2 d sin

c c c=fc
= c=fc = signal wavelength
a() = [1; e;i!s : : : e;i(m;1)!s ]T
By direct analogy with the vector a(! ) made from
uniform samples of a sinusoidal time series,
!s = spatial frequency
The function !s 7! a( ) is one-to-one for
j!sj $ dj=sin j 1 d =2

2
As
d = spatial sampling period
d =2 is a spatial Shannon sampling theorem.

Spatial Filtering
Spatial filtering useful for
DOA discrimination (similar to frequency

discrimination of time-series filtering)
Nonparametric DOA estimation
There is a strong analogy between temporal filtering and

spatial filtering.

Analogy between Temporal and Spatial Filtering
z s
u(t) = ei!t
-
0 1 - @

h

?
0
BB 1 C

@
q;
?
C h
s BB e;i! C
1
1 @1
P h- a ! [ ( )]u(t)
C

? @
R
@
- BB C
C - -
qq ? BB
BB ... C
BB
C
C
C
C
u(t)
qq

BB C
C
?
@ C
A
m;1 q ;1 hm;
e;i m; !

( 1)
| {z }
1

- ?
a(!) -
(Temporal sampling)
(a) Temporal filter
narrowband source with DOA=

z
Z
Z
ZZ

h

~
aa 1-
reference point -

0
?
!!
0 1 1 @

@
B C h @
aa 2- B C
1
B C
P ha [ ( )]s(t)
? @
R
@
!! B
BB e; i! 2
( )
C
C
- - -
BB C
Cs(t)
qq

qq BB .. C
B@
. C
C
C
A

;
| e {z }
i! m
( )
hm;
1
a()

?
aa m- -
!!
(Spatial sampling)
(b) Spatial filter

Spatial Filtering, con't
Example: The response magnitude h a of a spatial j ( )j

filter (or beamformer) for a 10-element ULA. Here,
= ( )
h a 0 , where 0 = 25
0
30 −30
Th
et
a
(d
eg
)
60 −60
90 −90
0 2 4 6 8 10
Magnitude

Spatial Filtering Uses
Spatial Filters can be used
To pass the signal of interest only, hence filtering out

interferences located outside the filter's beam (but
possibly having the same temporal characteristics as
the signal).
To locate an emitter in the field of view, by sweeping

the filter through the DOA range of interest
(“goniometer”).

Nonparametric Spatial Methods
A Filter Bank Approach to DOA estimation.
Basic Ideas
Design a filter h() such that for each

– It passes undistorted the signal with DOA =
– It attenuates all DOAs 6=
Sweep the filter through the DOA range of interest,

and evaluate the powers of the filtered signals:
E jyF (t)j2 = E jh()y(t)j2

n o n o
= h()Rh()
with R = E fy (t)y (t)g.
The (dominant) peaks of h()Rh() give the

DOAs of the sources.

Beamforming Method
Assume the array output is spatially white:
R = E fy(t)y(t)g = I
E jyF (t)j2 = hh
n o
Then:
Hence: In direct analogy with the temporally white

assumption for filter bank methods, y t can be ()
considered as impinging on the array from all DOAs.
Filter Design:
min ( h h) subject to ha() = 1

h
Solution:
h = a()=a()a() = a()=m
E jyF (t)j2 = a()Ra()=m2
n o

Implementation of Beamforming
1 N
R^ = N y(t)y(t)
X
t=1
The beamforming DOA estimates are:
f^kg = the locations of the n largest peaks of

( )^ ( )
a Ra .
This is the direct spatial analog of the Blackman-Tukey
periodogram.
Resolution Threshold:
inf jk ; pj > wavelength

array length
= array beamwidth
Inconsistency problem:
Beamforming DOA estimates are consistent if n = 1, but
inconsistent if n > . 1
Capon Method
Filter design:
min( hRh) subject to ha() = 1

h
Solution:
h = R;1a()=a()R;1a()
E jyF (t)j2 = 1=a()R;1a()
n o
Implementation:
f^kg = the locations of the n largest peaks of

1=a()R^;1a():
Performance: Slightly superior to Beamforming.
Both Beamforming and Capon are nonparametric

approaches. They do not make assumptions on the
covariance properties of the data (and hence do not
depend on them).
Parametric Methods
Assumptions:
The array is described by the equation:

y(t) = As(t) + e(t)
The noise is spatially white and has the same power in
all sensors:
E fe(t)e(t)g = 2I
The signal covariance matrix
P = E fs(t)s(t)g
is nonsingular.
Then:
R = E fy(t)y(t)g = APA + 2I

Thus: The NLS, YW, MUSIC, MIN-NORM and ESPRIT
methods of frequency estimation can be used, almost
without modification, for DOA estimation.

Lecture PPT Stocia

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture PPT Stocia

Uploaded by

Copyright:

Available Formats

Periodogram

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–1

Recall 2nd definition of (! ):

(!) = Nlim E 1 y (t )e ;i!t

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–2

Recall 1st definition of (! ):

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–3

Standard unbiased estimate:

Standard biased estimate:

For both estimators:

^r(k) = ^r(;k); k < 0

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–4

If: the biased ACS estimator r ^(k) is used in ^c(!),

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–5

 Both are asymptotically (for large N ) unbiased:

 Both have “large” variance, even for large N .

Thus, p ^ (!) and ^c(!) have poor performance.

 r^(k) ; r(k) may be large for large jkj

 Even if the errors f^r(k) ; r(k)gNjkj;=01 are small,

E p(!) = 2 ; ( )WB (! ;  ) d

Ideally: WB (!) = Dirac impulse (!).

WB (!)=WB (0), for N = 25

Main lobe 3dB width  1=N .

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–8

Main Lobe Width: smearing or smoothing

Details in  (!) separated in f by less than 1=N are not

Thus: Periodogram resolution limit = 1=N .

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–9

Summary of Periodogram Bias Properties:

 For “small” N , severe bias

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–10

- 1 st. dev = φ(ω) too

Resolvability properties depend on both bias and variance.

Lecture notes to accompany Introduction to Spectral Analysis Slide L2–11

Lecture notes to accompany Introduction to Spectral Analysis Slide L3–1

Basic Idea: Weighted correlogram, with small weight

fw(k)g = Lag Window

Lecture notes to accompany Introduction to Spectral Analysis Slide L3–2

^BT (!) = 2 ; ^p( )W (! ;  )d

W (!) = DTFT fw(k)g

Conclusion: BT ^ (!) = “locally” smoothed periodogram

 Variance decreases substantially

MSE = var + bias2 ! 0 as N ! 1

^BT (!) = 2 ; ^p( ) W (! ;  )d

If W (!)  0 (, w(k) is a psd sequence)

 As M increases, bias decreases and variance

) Choose M as a tradeoff between variance and

 Once M is given, Ne (and hence e) is essentially

) Choose window shape to compromise between

The energy in the main lobe and in the sidelobes

Lecture notes to accompany Introduction to Spectral Analysis Slide L3–5

Lecture notes to accompany Introduction to Spectral Analysis Slide L3–6

^φ1(ω) ^φ2(ω) ^φL(ω)

^B (!) = L ^rj (k)e;i!k

 High resolution (little smearing)

Similar to Bartlett method, but

 allow overlap of subsequences (gives more

 use data window for each periodogram; gives

The data in each subsequence are weighted by a temporal

Welch is approximately equal to BT ^ (!) with a

By a previous result, for N  1,

Lecture notes to accompany Introduction to Spectral Analysis Slide L3–10

Recall 2nd definition of (! ):

(!) = Nlim E 1 y (t )e ;i!t

Recall 1st definition of (! ):

If: the biased ACS estimator r ^(k) is used in ^c(!),

Both are asymptotically (for large N ) unbiased:

Both have “large” variance, even for large N .

Thus, p ^ (!) and ^c(!) have poor performance.

r^(k) ; r(k) may be large for large jkj

Even if the errors f^r(k) ; r(k)gNjkj;=01 are small,

E p(!) = 2 ; ( )WB (! ; ) d

Ideally: WB (!) = Dirac impulse (!).

Main lobe 3dB width 1=N .

Details in (!) separated in f by less than 1=N are not

For “small” N , severe bias

^BT (!) = 2 ; ^p( )W (! ; )d

Conclusion: BT ^ (!) = “locally” smoothed periodogram

Variance decreases substantially

^BT (!) = 2 ; ^p( ) W (! ; )d

If W (!) 0 (, w(k) is a psd sequence)

As M increases, bias decreases and variance

Once M is given, Ne (and hence e) is essentially

^B (!) = L ^rj (k)e;i!k

High resolution (little smearing)

allow overlap of subsequences (gives more

use data window for each periodogram; gives

Welch is approximately equal to BT ^ (!) with a

By a previous result, for N 1,

Bias increases (more smoothing)

Let = 2J=N . Then, for N 1,

choice of m and n is not simple

By Spectral Factorization theorem, a rational (!) can be

Two main ways to Estimate (! ):

1. Estimate fbkg and 2 and insert them in

^(!) is not guaranteed to be 0

(!) = 2 A(!) = jA(!)j2

nonlinear estimation problem; can use an approximate

2. Estimate fai; r(k)g in (!) = jA(!)j2

^(!) is not guaranteed to be 0

A1: k > 0 !k 2 [;; ]

A3: e(t) = circular white noise with variance 2

i' ;i' i'

E xp(t)xj (t ; k) = 2p ei!pk p;j

f!k ; k; 'k gnk=1; 2 (Signal Model)

f!k ; 2k gnk=1; 2 (PSD Model)

Both f^kg and f'^kg can be derived by a least

R = APA + 2I (m > n)