Basic Deﬁnitions and The Spectral Estimation Problem
Lecture 1
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L1–1
Informal Deﬁnition of Spectral Estimation
Given: A ﬁnite record of a signal. Determine: The distribution of signal power over frequency.
signal t ω t=1, 2, ... ω ω+∆ω π spectral density
! = (angular) frequency in radians/(sampling interval) f = !=2 = frequency in cycles/(sampling interval)
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L1–2
Applications Temporal Spectral Analysis Vibration monitoring and fault detection Hidden periodicity ﬁnding Speech processing and audio devices Medical diagnosis Seismology and ground movement study Control systems design Radar, Sonar
Spatial Spectral Analysis Source location using sensor arrays
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997 Slide L1–3
Deterministic Signals
fy(t)g1 .1
jy(t)j2 < 1
X
Then:
Y (! ) =
1
t=. Moses.i!t
exists and is called the DiscreteTime Fourier Transform (DTFT)
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Prentice Hall. 1997
Slide L1–4
.1
y(t)e.1 = t=
If:
discretetime deterministic data sequence
X
1
t=.
Energy Spectral Density
Parseval's Equality:
jy(t)j2 = 21 .1
X Z
1
where
4 S (!) = jY (!)j2 = Energy Spectral Density
We can write
S (!) =
where
X
1
k=. Moses.1
y(t)y (t . 1997
. Prentice Hall. S (!)d! t=. Stoica and R.1
X
(k)e. k)
Slide L1–5
Lecture notes to accompany Introduction to Spectral Analysis by P.i!k
(k) =
1
t=.
Prentice Hall. 1997
Slide L1–6
. Moses.Random Signals
Random Signal
random signal probabilistic statements about future variations t current observation time
Here:
X
1
t=. Stoica and R.1
n
jy(t)j2 = 1
o
But:
E jy(t)j2 < 1
E f g = Expectation over the ensemble of realizations E jy(t)j2
n o
= Average power in y
(t)
PSD = (Average) power spectral density
Lecture notes to accompany Introduction to Spectral Analysis by P.
(!)d!
Z
(!)d! =
inﬁnitesimal signal power in the band
! d! 2
Lecture notes to accompany Introduction to Spectral Analysis by P. (!)ei!k d!
n o
(Inverse DTFT)
Interpretation:
so
r(0) = E jy(t)j2 = 21 . k)g r(k) = r (. 1997
Slide L1–7
.i!k
(k) is the autocovariance sequence (ACS) r(k) = E fy(t)y (t .1
r(k)e.k) r(0) jr(k)j
Z
Note that
r(k) = 21 . Moses. Prentice Hall.First Deﬁnition of PSD
(! ) =
where r
X
1
k=. Stoica and R.
1997
Slide L1–8
.i!t (!) = Nlim E N !1 t=1
X > :
8 > <
2
9 > = >
Note that
1 jY (!)j2 (!) = Nlim E N N !1
YN (!) =
X
where
N
is the ﬁnite DTFT of
fy(t)g. Prentice Hall.Second Deﬁnition of PSD
1 N y(t)e.
t=1
y(t)e.i!t
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses.
Moses. Then: (! ) = (.! )
P3: If y
Lecture notes to accompany Introduction to Spectral Analysis by P.1=2 1=2]
(!) 0 (t) is real. Prentice Hall.Properties of the PSD
P1:
(!) = (! + 2 ) for all !. Stoica and R.! ) Otherwise: (! ) 6= (.
Thus. we can restrict attention to
!2 . 1997
Slide L1–9
.
P2:
] () f 2 .
Moses. 1997
Slide L1–10
. 1)
H (q ) y (t) 2 y (! ) = jH (! )j
e(
e(t) e (! )
Then
!)
y(t) =
X
1
k=0
hk e(t . k) hk e.i!k
H (! ) =
X
1
y (!) = jH (!)j2 e(!)
k=0
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R.Transfer of PSD Through Linear Systems
System Function: where q .1
H (q) =
X
1
k=0
hk q.k
= unit delay operator: q.1y(t) = y(t . Prentice Hall.
Parametric: – Assumes a parameterized functional form of the PSD
Lecture notes to accompany Introduction to Spectral Analysis by P. ]g
Find an estimate of
Two Main Approaches :
Nonparametric: – Derived from the PSD deﬁnitions. 1997
Slide L1–11
. Stoica and R. Moses.The Spectral Estimation Problem
The Problem: From a sample
fy(1) : : : y(N )g
(!): f ^(!) ! 2 . Prentice Hall.
Moses. Stoica and R. Prentice Hall. 1997
Slide L2–1
.Periodogram and Correlogram Methods
Lecture 2
Lecture notes to accompany Introduction to Spectral Analysis by P.
1997
Slide L2–2
.i!t N t=1
Natural estimator
N
2
Used by Schuster ( 1900) to determine “hidden periodicities” (hence the name). Stoica and R.i!t (!) = Nlim E N !1 t=1
X > :
8 > <
2
9 > = >
Given : Drop “
fy(t)gN t=1
lim
” and “E
N !1
f g” to get
X
^p(!) = 1 y(t)e.Periodogram Recall 2nd deﬁnition of (! ):
1 N y(t)e. Prentice Hall.
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses.
Correlogram
Recall 1st deﬁnition of (! ):
(! ) =
P
X
1
Truncate the “ ” and replace “r
k=.(N .1 X
r(k)e. Moses. 1997
Slide L2–3
.i!k
^c(!) =
k=.1 N .i!k
(k)” by “^(k)”: r
r ^(k)e. Stoica and R. Prentice Hall.1)
Lecture notes to accompany Introduction to Spectral Analysis by P.
Covariance Estimators (or Sample Covariances)
Standard unbiased estimate:
1 N y(t)y (t .k) k < 0 r
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. k) k 0 ^ r(k) = N . k t=k+1
X
Standard biased estimate:
1 N y(t)y (t . Prentice Hall. 1997
Slide L2–4
. Stoica and R. k) k 0 r(k) = N ^ t=k+1
X
For both estimators:
r ^(k) = ^ (.
Prentice Hall.1)
r ^(k)e.i!k
^p(!) = ^c(!)
Consequence: Both p ! and
^( )
^c(!) can be analyzed simultaneously.1 X
= ^c(!)
k=. 1997
.
Slide L2–5
Lecture notes to accompany Introduction to Spectral Analysis by P.
N
^p(!) = 1 y(t)e. Stoica and R.(N . Moses.Relationship Between ^p (! ) and ^c(! )
If: the biased ACS estimator r Then:
^(k) is used in ^c(!).i!t N t=1
X
2
=
N .
Statistical Performance of ^p(! ) and ^c(! ) Summary: Both are asymptotically (for large N ) unbiased:
E ^p(!) ! (!) as N ! 1
n o
Both have “large” variance. jk =0 there are “so many” that when summed in ! . p!
f^( ) . r(k) may be large for large jkj ^
Even if the errors r k r k Nj. the PSD error is large. Prentice Hall. ( )g
^ ( ) .1 are small. ( )]
Lecture notes to accompany Introduction to Spectral Analysis by P. Thus. 1997
Slide L2–6
. Moses. Stoica and R.
Intuitive explanation:
r(k) . even for large N .
^p(!) and ^c(!) have poor performance.
n o
0
k 1 . window
E ^p(!) = 21 . jNj
jkj N .Bias Analysis of the Periodogram
E ^p(!) = E ^c(!) = E f^(k)g e. or triangular. Prentice Hall.1 jkj r(k)e.i!k = 1.1) N .
Slide L2–7
Lecture notes to accompany Introduction to Spectral Analysis by P. 1 jkj N
Bartlett.i!k k=.i!k r k=. Moses.(N .1) 1 = wB (k)r(k)e. ) d
Z
Ideally:
WB (!) = Dirac impulse (!). Stoica and R. N k=.(N . ( )WB (! .1 X
wB (k) =
8 < :
=
Thus. 1997
.1
n o n o X ! X
N .
(!) may differ quite a bit from (!). 1997
.
Slide L2–8
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. for N = 25
−10
−20
dB
−30
−40
−50
−60
−3
−2
−1 0 1 ANGULAR FREQUENCY
2
3
Main lobe 3dB width For “small” N .Bartlett Window WB (! )
1 sin(!N=2) 2 WB (!) = N sin(!=2)
" #
0
WB (!)=WB (0). Stoica and R. Moses. WB
1=N .
^ φ(ω)≅W (ω) B
Sidelobe Level: leakage
φ(ω)≅δ(ω) leakage ω ω
Lecture notes to accompany Introduction to Spectral Analysis by P.
φ(ω) smearing ω ω
(!) separated in f by less than 1=N are not
^ φ(ω)
<1/Ν
Thus:
Periodogram resolution limit =
1=N . Prentice Hall.Smearing and Leakage
Main Lobe Width: smearing or smoothing Details in resolvable. 1997
Slide L2–9
. Stoica and R. Moses.
!1 ^( )
( )! ( )
Lecture notes to accompany Introduction to Spectral Analysis by P. severe bias As N . 1997
Slide L2–10
. Prentice Hall. WB ! !.Periodogram Bias Properties
Summary of Periodogram Bias Properties: For “small” N . Moses. so ! is asymptotically unbiased. Stoica and R.
Lecture notes to accompany Introduction to Spectral Analysis by P.Periodogram Variance As N
!1 E ^p(!1) . Stoica and R. dev = φ(ω) too 
Resolvability properties depend on both bias and variance. (!2) 2(!1) !1 = !2 = 0 !1 6= !2
nh ih (
io
Inconsistent estimate Erratic behavior
^ φ(ω) ^ φ(ω)
asymptotic mean = φ(ω) ω + 1 st. (!1) ^p(!2) . 1997
Slide L2–11
. Prentice Hall. Moses.
i 2 . N Let ! = N
Then YN
( 2 k) is the Discrete Fourier Transform (DFT): N
X
Y (k) =
N
t=1
y(t)W tk
k = 0 ::: N . Moses. 1
Direct computation of O N 2 ﬂops
( )
. Prentice Hall. Stoica and R.Discrete Fourier Transform (DFT)
Finite DTFT: YN
(!) =
X
N
t=1
y(t)e.i!t
2 k and W = e. 1997
Slide L2–12
.1 fY (k)gN=0 from fy(t)gN : t=1 k
Lecture notes to accompany Introduction to Spectral Analysis by P.
Radix–2 Fast Fourier Transform (FFT)
N = 2m N=2 N Y (k) = y(t)W tk + y(t)W tk t=1 t=N=2+1 N=2 2 = y(t) + y(t + N=2)W Nk ]W tk t=1 Nk +1 for even k with W 2 = . Moses. 1997
. 2 = 2p:
Y (2p) =
For k
X
~ N
= 1 3 5 : : : N . 1 = 2p + 1:
X
t=1
~ ~ y(t) + y(t + N )]W tp
Y (2p + 1) =
Each is a N
~ N
~ = N=2point DFT computation. ~
4 = 0 2 4 : : : N .i2 =N . y(t + N )]W tgW tp
Lecture notes to accompany Introduction to Spectral Analysis by P.1 for odd k
Assume:
X X X
Let N For k
~ ~ = N=2 and W = W 2 = e. Prentice Hall.
Slide L2–13
t=1
~ ~ f y(t) . Stoica and R.
Prentice Hall. Moses. 1997
Slide L2–14
. 2k + 2c ck = 2 k. Stoica and R.1 k2k ) ck = 2
ck = 1 N log2 N 2
Thus.FFT Computation Count
Let ck Then
= number of ﬂops for N = 2k point FFT.
Lecture notes to accompany Introduction to Spectral Analysis by P.
Moses. Stoica and R.1 N k=0
sampled. 1997
Slide L2–15
. Prentice Hall.Zero Padding Append the given data by zeros prior to computing DFT (or FFT):
fy(1) : : : y(N ) 0 : : : 0g

N
{z
}
Goals: Apply a radix2 FFT (so N Finer sampling of
= power of 2)
^(!): 2 k N . N=8 ω
2 k N .1 ! N k=0
^ φ(ω)
continuous curve
Lecture notes to accompany Introduction to Spectral Analysis by P.
Stoica and R. Prentice Hall.Improved PeriodogramBased Methods
Lecture 3
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. 1997
Slide L3–1
.
BlackmanTukey Method
Basic Idea: Weighted correlogram.1)
w(k)^(k)e.1 X k=. 1997
Slide L3–2
.i!k r
fw(k)g =
Lag Window
w(k)
1
M
M
k
Lecture notes to accompany Introduction to Spectral Analysis by P.
^( )
jj
^BT (!) =
M . with small weight applied to covariances r k with “large” k . Stoica and R.(M . Prentice Hall. Moses.
con't
^BT (!) = 1 ^p( )W (! . Stoica and R. Moses. 1997
.BlackmanTukey Method. Prentice Hall.
Z
W (!) = =
Conclusion: Effect:
DTFT
fw(k)g
Spectral Window
^BT (!) = “locally” smoothed periodogram
Variance decreases substantially Bias increases slightly By proper choice of M : MSE
= var + bias2 ! 0 as N ! 1
Slide L3–3
Lecture notes to accompany Introduction to Spectral Analysis by P. )d 2 .
W (!)d! = equiv bandwidth e= W (0)
Ne e = 1
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. 1997 Slide L3–4
.1) Ne = k=. )d 2 .(M (0) w
Z
w (k )
= equiv time width
1 2 . Stoica and R. w(k) is a psd sequence) ^BT (!) 0
(which is desirable)
Then:
TimeBandwidth Product
M . 0
Z  {z }
If W
(!) 0 (.1 X
.Window Design Considerations Nonnegativeness:
^BT (!) = 1 ^p( ) W (! . Prentice Hall.
con't
e = 1=Ne = 0(1=M )
is the BT resolution threshold. Prentice Hall. Moses.
Once M is given.
The energy in the main lobe and in the sidelobes cannot be reduced simultaneously.
e) is essentially
)
Choose window shape to compromise between smearing (main lobe width) and leakage (sidelobe level).
)
Choose M as a tradeoff between variance and bias.Window Design. bias decreases and variance increases. As M increases. Ne (and hence ﬁxed. Stoica and R. 1997
Slide L3–5
. once M is given.
Lecture notes to accompany Introduction to Spectral Analysis by P.
Moses. M
0
= 25
−10
−20
dB
−30
−40
−50
−60
−3
−2
−1 0 1 ANGULAR FREQUENCY
2
3
Rectangular Window.Window Examples Triangular Window. Stoica and R. M
0 −10
= 25
−20
dB
−30
−40
−50
−60
−3
−2
−1 0 1 ANGULAR FREQUENCY
2
3
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L3–6
. Prentice Hall.
Bartlett Method Basic Idea:
1 Μ 2Μ ... Ν
^1(ω) ^2(ω) φ φ
... average
^L(ω) φ
^B(ω) φ
Mathematically:
yj (t) = y((j ; 1)M + t) t = 1 : : : M = the j th subsequence 4 (j = 1 : : : L = N=M ])
^j (!) = 1 yj (t)e;i!t M t=1
X
M
2
1 L ^ (!) ^B (!) = L j =1 j
X
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L3–7
Comparison of Bartlett and BlackmanTukey Estimates
1 L ^B (!) = L j =1
X
8 > < > :
M ;1 X
=
'
Thus:
k=;(M ;1) M ;1 X k=;(M ;1)
k=;(M ;1) 8 M ;1 < 1 X L X
:
r ^j (k)e;i!k
9 =
9 > = >
L j =1
r ^j (k) e;i!k
r ^(k)e;i!k
^B (!) ' ^BT (!)
Since B ! implicitly uses method has
with a rectangular lag window wR k
()
^( )
fwR(k)g, the Bartlett
High resolution (little smearing) Large leakage and relatively large variance
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997 Slide L3–8
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 mainlobesidelobe tradeoff capability
1
2 . . . subseq subseq #1 #2 subseq #S
N
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 nonrectangular lag window.
^BT (!) with a
Slide L3–9
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
J
X
Lecture notes to accompany Introduction to Spectral Analysis by P. for N
1.
f ^p(!j )g are (nearly) uncorrelated random variables for 2 j N . 1997
Slide L3–10
. Prentice Hall. Moses.Daniell Method
By a previous result.
k +J 1 ^D(!k ) = ^p (!j ) 2J + 1 j=k. Stoica and R.1 ! =
N j =0 Idea: “Local averaging” of (2J + 1) samples in the
j
(2 + 1)
frequency domain should reduce the variance by about J .
^D(!) ' 1 ^p(!)d! 2 . Then. Prentice Hall.Daniell Method. 1997
Slide L3–11
. for N 1. con't
As J increases: Bias increases (more smoothing) Variance decreases (more averaging)
= 2J=N . Stoica and R. Hence: ^D (! ) ' ^BT (! ) with a rectangular spectral
Let
Z
window.
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses.
Moses.
Daniell Periodogram
– Approximate interpretation: spectral 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. more general for Welch). Welch periodograms
– Approximate interpretation: BT ! with a suitable lag window (rectangular for Bartlett.
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. Stoica and R.
^ ()
– Implemented by averaging subsample periodograms.
BT periodogram
– Local smoothing/averaging of spectral window. 1997
.
^BT (!) with a rectangular
Slide L3–12
– Implemented by local averaging of periodogram values.
^p(!) by a suitably selected ^(k) using a lag
– Implemented by truncating and weighting r window in c !
^( )
Bartlett.
Prentice Hall. Stoica and R. Moses.Parametric Methods for Rational Spectra
Lecture 4
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L4–1
.
! can approximate arbitrarily well any continuous PSD.θ) Estimate parameters in φ(ω.θ)
^ θ
Estimate PSD
^ ^ =φ(ω. provided m and n are chosen sufﬁciently large. Moses.Basic Idea of Parametric Spectral Estimation
Observed Data Assumed functional form of φ(ω.i!k (!) = P jkj n k e. 1997 Slide L4–2
( )
.θ) φ(ω)
possibly revise assumption on φ(ω)
Rational Spectra
jkj m k e. Stoica and R. however: choice of m and n is not simple some PSDs are not continuous
Lecture notes to accompany Introduction to Spectral Analysis by P.i! .i!k (!) is a rational function in e. Note. Prentice Hall.
P
By Weierstrass theorem.
a rational factored as
(!) can be
B (!) 2 2 (!) = A(!) A(z) = 1 + a1z.1 + + bmz. A(! ) = A(z )jz =ei!
Signal Modeling Interpretation:
e(t) e (! ) =
2

B (q) A(q)
white noise
B (!) 2 2 A(!) ltered white noise
y(t)y (! ) =
ARMA: AR: MA:
A(q)y(t) = B (q)e(t) A(q)y(t) = e(t) y(t) = B (q)e(t)
Slide L4–3
Lecture notes to accompany Introduction to Spectral Analysis by P.. MA. 1997
. e. Prentice Hall. Moses.m and.AR.n B (z) = 1 + b1z. and ARMA Models
By Spectral Factorization theorem. Stoica and R.1 + + anz.g.
k) and take E f g to give:
X
r (k ) +
X
2 m bj hj .k (h = 1) where H (q ) = A(q ) 0 k k=0
X X
i=1
air(k .k = j =0 = 0 for k > m B (q) = 1 h q. k)g
Lecture notes to accompany Introduction to Spectral Analysis by P. j )
(b0 = 1)
Multiply by y
(t . j )y (t . Moses.ARMA Covariance Structure
ARMA signal model:
y(t)+
X
n
i=1 n
aiy(t . Stoica and R. 1997
Slide L4–4
. i) =
X
m
j =0
bj e(t . i) =
m
j =0
bj E fe(t . Prentice Hall.
1) : : : r(. an 0
3 7 7 7 5 2 6 6 6 4
Writing covariance equation in matrix form for k : : : n:
=1 r(0) r(. Stoica and R. . Moses.
2 1 a1 = 0 .1) . . r(n) : : : r(0) 1 = 2 R 0
2 6 6 6 4 7 7 7 5 " # "
32 6 6 6 4
3 7 7 7 5
#
These are the Yule–Walker (YW) Equations.n) . . r(1) r(0) . . . 1997
Slide L4–5
. . .AR Signals: YuleWalker Equations
AR: m
= 0. Prentice Hall.
Lecture notes to accompany Introduction to Spectral Analysis by P. r (.
AR Spectral Estimation: YW Method
YuleWalker Method: Replace r
2 6 6 6 4
(k) by r(k) and solve for f^ig and ^2: ^ a r ^(0) ^(. 1997
Slide L4–6
. .1 = 0 a . .1) . . r . r ^(1) r(0) ^ ^. Prentice Hall. Moses. . .1) : : : ^(. r ^(n) : : : r(0) ^ ^n a 0
32 7 7 7 5 6 6 6 4 3 7 7 7 5 2 6 6 6 4
3 7 7 7 5
Then the PSD estimate is
^2 ^(!) = ^ 2 jA(!)j
Lecture notes to accompany Introduction to Spectral Analysis by P. . Stoica and R.n) 1 r r ^2 . ^(.
i) = y(t) + 'T (t)
= a1 : : : an]T to minimize
f( ) =
X
N
This gives
2
y=6 4
y(n + 1) y(n + 2) y(N )
.(Y Y ). Prentice Hall. n)
.
y(N .
Find
i=1
aiy(t .1(Y y) where
3 7 5 2 6
t=n+1
je(t)j2
y(1) y(2) y(N . . 2)
Lecture notes to accompany Introduction to Spectral Analysis by P. 1) : : : y (t . . .
^ = .AR Spectral Estimation: LS Method
Least Squares Method:
e(t) = y(t) +
X
n
4 = y(t) + y(t) ^ where '(t) = y (t .
3 7 5
Y
=4
y(n) y(n . Moses. n)]T . . Stoica and R. 1) y(N . . 1997
Slide L4–7
. 1) y(n + 1) y(n)
. .
Moses. .
{z
.Levinson–Durbin Algorithm
Fast.
Direct Solution: For one given value of n: For k
Rn+1
1
0
= 0 . Prentice Hall.
.
n
..
32 7 7 7 5 } 6 6 6 4
1
n
3 7 7 7 5
2 6 6 6 4
^ k = either r(k) or r(k).1 1 .
..n .. 0
2 n
3 7 7 7 5
O(n3) ﬂops
= 1 : : : n: O(n4) ﬂops
Levinson–Durbin Algorithm: Exploits the Toeplitz form of Rn+1 to obtain the solutions for k : : : n in O n2 ﬂops!
=1
( )
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. . . .. orderrecursive solution to YW equations
2 6 6 6 4 
0 . 1997
Slide L4–8
.1
.0 ..
con't
Relevant Properties of R:
Rx = y $ Rx = y. 1997
Slide L4–9
. Moses. Prentice Hall.
n
3 7 5
1
Rn+2 4
1 n = 0
3 5
2 4
Rn+1
n+1
r ~n
n+1 r ~n
0
32 54
1 n n = 0 0 n
3 5 2 4
2
3 5
where
r n = n+1 + ~n n
Lecture notes to accompany Introduction to Spectral Analysis by P. where x = xn : : : x1]T ~ ~ ~
Nested structure
2
Rn+2 =
Thus. . Stoica and R.
2
4
Rn+1
n+1
r ~n
n+1 r ~n
0
3 5
2
r ~n =
6 4
.LevinsonDurbin Alg.
1997
. Stoica and R. n = n )
n+1 =
0 + kn 2+1 = n + kn n = n(1 . Prentice Hall.LevinsonDurbin Alg.
2 kn = . Moses. jknj2) 2 2 n
"
n
#
"
~n 1
#
Computation count: k ﬂops for the step k
2 )
n2 ﬂops
!k+1 2 to determine f k
k gn=1 k
Slide L4–10
This is O
(n2) times faster than the direct solution. con't
2
Rn+2 4
82 < 4 +2 :
1 n n = 0 0 n
3 5 2 4
2
3 5
2
Rn+2 4
39 = 5 2
0 ~n = 0n 2 1 n
3 5 2 4 3 5 2
3 5
Combining these gives:
Rn
0 1 n + kn ~n 0 1
3 5 2 4
=
4
n + kn n
2
0 n + kn
2 n
=
4
n
2 +1
3 5
0 0
Thus.
Lecture notes to accompany Introduction to Spectral Analysis by P.
Prentice Hall. m)
Thus.
r(k) = 0 for jkj > m
and
2 2 = m r(k)e. Moses. 1997
Slide L4–11
. Stoica and R.MA Signals
MA: n
=0 y(t) = B (q)e(t) = e(t) + b1e(t .m
X
Lecture notes to accompany Introduction to Spectral Analysis by P.i!k (!) = jB(!)j k=. 1) +
+ bme(t .
Lecture notes to accompany Introduction to Spectral Analysis by P. Estimate
fbkg and 2 and insert them in (!) = jB(!)j2 2
nonlinear estimation problem
^(!) is guaranteed to be 0
2. ^(!) is not guaranteed to be 0 =0
Both methods are special cases of ARMA methods described below. Stoica and R. Moses. 1997 Slide L4–12
. Insert sample covariances
(! ) =
This is length
X
m
f^(k)g in: r
r(k)e.MA Spectrum Estimation Two main ways to Estimate (! ): 1.m
^BT (!) with a rectangular lag window of 2m + 1.i!k
k=. Prentice Hall. with AR model order n .
j )
Lecture notes to accompany Introduction to Spectral Analysis by P. k)] g = E f A(q)y(t)] A(q)y(t . 1997
Slide L4–13
. k)] g =
X
n
X
n
j =0 p=0
aj ap r(k + p .ARMA Signals
ARMA models can represent spectra with both peaks (AR part) and valleys (MA part). Prentice Hall.m k e. Moses.
A(q)y(t) = B (q)e(t) B (!) 2 = m=.i!k k (!) = 2 A(!) jA(!)j2
P
where
k
= E f B(q)e(t)] B(q)e(t . Stoica and R.
Estimate
fai r(k)g in (!) =
P
m . 1997 Slide L4–14
. can use an approximate linear twostage least squares method
^(!) is guaranteed to be 0
2.i!k k=. Stoica and R. Prentice Hall.ARMA Spectrum Estimation
Two Methods:
2g in (!) = 2 B(!) 2 1. Estimate fai bj A(!)
nonlinear estimation problem. Moses.
^(!) is not guaranteed to be 0
Lecture notes to accompany Introduction to Spectral Analysis by P.m k e
jA(!)j2
linear estimation problem (the Modiﬁed YuleWalker method).
Prentice Hall. 1) + + K y(t . for some large K e(t) ' y(t) + 1y(t . 2) + = AR(1) with j kj ! 0 as k ! 1 Step 1: Approximate. Stoica and R. K )
1a) Estimate the coefﬁcients modelling techniques. 1) + e
and its variance
+ ^K y(t . K t=K +1
X
Lecture notes to accompany Introduction to Spectral Analysis by P.
f kgK=1 by using AR k
1b) Estimate the noise sequence
^(t) = y(t) + ^1y(t . 1) + 2y(t .TwoStage LeastSquares Method Assumption: The ARMA model is invertible:
A e(t) = B(q) y(t) (q ) = y(t) + 1y(t . K )
^2 =
N 1 j^(t)j2 e N . 1997
Slide L4–15
. Moses.
^(t) ' . Stoica and R. Prentice Hall. con't
fe(t)g by ^(t) in the ARMA equation. 1) : : : ^(t . e A(q)y(t) ' B (q)^(t) e and obtain estimates of fai bj g by applying least squares
Step 2: Replace techniques. y(t . m)] e e = a1 : : : an b1 : : : bm]T
Lecture notes to accompany Introduction to Spectral Analysis by P.y(t . Moses. Note that the ai and bj coefﬁcients enter linearly in the above equation:
y(t) . 1997
Slide L4–16
. n) e ^(t . 1) : : : .TwoStage LeastSquares Method.
5 . . If M > n overdetermined squares solution for ai . . the accuracy of ai is often better for M > n
Lecture notes to accompany Introduction to Spectral Analysis by P. least
f^ g
Note: For narrowband ARMA signals. .
: : : r(m .. .
#
2
=
r(m + M . 1997
Slide L4–17
.
=
f^ g
Y W system of equations. Prentice Hall. Moses. . n + 1) r(m . fast Levinsontype algorithms exist for obtaining ai .6 4
Replace
fr(k)g by f^(k)g and solve for faig. Stoica and R. . .
= m + 1 ::: m + M
a1 an
.. i) = 0 k > m
3 7 5 "
r(m) r(m + 1)
.Modiﬁed YuleWalker Method
ARMA Covariance Equation:
r(k) +
2 6 4
X
n
In matrix form for k
i=1
air(k . 1) : : : r(m . n + 2)
. r f^ g
. r(m + M )
r(m + 1) r(m + 2)
3 7
If M n. . n + M )
.
Moses. Prentice Hall.Summary of Parametric Methods for Rational Spectra
Method AR: YW or LS low lowmedium mediumhigh mediumhigh medium lowmedium No No Yes
Computational Burden low Accuracy medium
Guarantee ^(! ) 0 ? Yes
MA: BT
ARMA: MYW
ARMA: 2Stage LS
Use for Spectra with (narrow) peaks but no valley Broadband spectra possibly with valleys but no peaks Spectra with both peaks and (not too deep) valleys As above
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L4–18
. Stoica and R.
Parametric Methods for Line Spectra — Part 1
Lecture 5
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L5–1
. Moses. Prentice Hall. Stoica and R.
Prentice Hall. Stoica and R. !1
!2
!3

An “Ideal” line spectrum
Lecture notes to accompany Introduction to Spectral Analysis by P. sonar geophysical seismology ARMA model is a poor approximation Better approximation by Discrete/Line Spectrum Models
2 6(2 1) 2
6
(!)
6
(2 2) 2
6
(2 2) 3
!
Slide L5–2
.Line Spectra Many applications have signals with (near) sinusoidal components. Examples: communications radar. 1997
. Moses.
1997
.
]
(prevents model ambiguities)
f'k g = independent rv's. ]
(realistic and mathematically convenient)
A3:
e(t) = circular white noise with variance 2 E fe(t)e (s)g = 2 t s E fe(t)e(s)g = 0
(can be achieved by “slow” sampling)
Slide L5–3
Lecture notes to accompany Introduction to Spectral Analysis by P. uniformly distributed on . Stoica and R. Moses. power
f g
f g
y(t) = x(t) + e(t) t = 1 2 : : : n x(t) = k ei(!k t+ k ) k=1 xk (t)
X  {z }
Assumptions: A1: A2:
k>0
!k 2 . superimposed in white noise of k 2. Prentice Hall.Line Spectral Signal Model Signal Model: Sinusoidal components of frequencies !k and powers 2 .
k) = 2 ei!pk p j p
n o
Hence. for p 6= j =2 . k)g = n=1 2ei!pk + 2 k 0 p p
P
and
(! ) = 2
X
n
p=1
2 (! . Prentice Hall.i'j o = E nei'p o E ne. e
E xp(t)xj (t . Moses.Covariance Function and PSD
Note that:
E E
n
i'p e. Stoica and R. 1997
.
r(k) = E fy(t)y (t .i'j o e
Z
n
1 i' d' 2 = 0. !p) + 2 p
Slide L5–4
Lecture notes to accompany Introduction to Spectral Analysis by P. for p = j e i'p e.i'j o = 1.
1997
Slide L5–5
. Stoica and R.
^ ei!k t + residuals k
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. Prentice Hall.Parameter Estimation Estimate either:
f!k
k 'k gn=1 k
2
(Signal Model)
f!k 2gn=1 2 k k
Once
(PSD Model)
^ Major Estimation Problem: f!k g
f!kg are determined: ^
X
f ^2g can be obtained by a least squares method from k n 2ei!pk + residuals ^ ^ r(k) = p
p=1
OR: Both k and 'k can be derived by a least squares method from
f^ g
f^ g
X
y(t) =
with
n
k=
k=1 k ei'k .
Moses.Nonlinear Least Squares (NLS) Method
min' g f!
Let:
X
N
k k k t=1 
y(t) . . Stoica and R. . 1997
Slide L5–6
. Prentice Hall. B = eiN!1 eiN!n
k
2 6 4
3 7 5
Lecture notes to accompany Introduction to Spectral Analysis by P.
X
n
k=1 {z F (! ')
k ei(!k t+'k )
2
}
= kei'k = 1 : : : n]T Y = y(1) : : : y(N )]T ei!1 ei!n . .
B k2 = . (B B). Moses. B ) (Y . Y B(B B).1B Y !=^ !
and
! = arg max Y B (B B ). (B B).1B Y ] +Y Y .1B Y ] B B] .1B Y ^ !
Lecture notes to accompany Introduction to Spectral Analysis by P.Nonlinear Least Squares (NLS) Method. Prentice Hall. B ) = kY .1B Y
This gives:
^ = (B B). con't
Then:
F = (Y . Stoica and R. 1997
Slide L5–7
.
it is difﬁcult to avoid convergence to local minima. Prentice Hall. Moses. Stoica and R.
Lecture notes to accompany Introduction to Spectral Analysis by P.5.NLS Properties
Excellent Accuracy:
6 2 (for N var (^k ) = 3 2 ! N k Example: N = 300 SNRk = 2= 2 = 30 dB k
q
1)
Then
var
(^k ) 10. !
Difﬁcult Implementation: The NLS cost function F is multimodal. 1997
Slide L5–8
.
Unwindowed Periodogram as an Approximate NLS Method
For a single (complex) sinusoid, the maximum of the unwindowed periodogram is the NLS frequency estimate: Assume: Then:
n=1
X
B B=N
N t=1
BY=
y(t)e;i!t = Y (!)
(ﬁnite DTFT)
1 Y B (B B );1B Y = N jY (!)j2 = ^p(!) = (Unwindowed Periodogram)
So, with no approximation,
! = arg max ^p(!) ^ !
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L5–9
Unwindowed Periodogram as an Approximate NLS Method, con't
Assume: Then:
n>1
f!kgn=1 ' the ^ k
provided that
locations of the peaks of p !
^( )
n
largest
inf j!k ; !pj > 2 =N
which is the periodogram resolution limit. If better resolution desired then use a High/Super Resolution method.
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L5–10
HighOrder YuleWalker Method Recall:
y(t) = x(t) + e(t) =
k ei(!k t+'k ) +e(t)  {z } k=1 xk (t)
X
n
“Degenerate” ARMA equation for y (t):
n
(1 ; ei!k q;1)xk (t) = k ei(!kt+'k) ; ei!k ei !k(t;1)+'k] = 0
o
Let
B (q) = 1 +
4 bk q;k = A(q)A(q) k=1 A(q) = (1 ; ei!1 q;1) (1 ; ei!n q;1) A(q) = arbitrary
X
L
Then B
(q)x(t) 0 ) B (q)y(t) = B (q)e(t)
Slide L5–11
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Moses.HighOrder YuleWalker Method.
f!kgn=1. 1997
Slide L5–12
. along with L . n ^ k
Lecture notes to accompany Introduction to Spectral Analysis by P. con't
Estimation Procedure: Estimate
f^igL=1 using an ARMA MYW technique b i
^( )
Roots of B q give “spurious” roots. Stoica and R. Prentice Hall.
. 1997
Slide L5–13
. . Moses. . i) = 0 k > L
In matrix form for k
2 6 6 6 4 
= L + 1 ::: L + M
3 7 7 7 5 2 6 6 6 4 3 7 7 7 5 }
: : : r(1) r(L + 1) : : : r(2) b = . 1) : : : r(M ) r(L + M ) 4 4 = =
{z }  {z
r(L) r(L + 1)
This is a highorder (if L > n) and overdetermined (if M > L) system of YW equations.HighOrder and Overdetermined YW Equations
ARMA covariance:
r (k ) +
X
L
i=1
bir(k . Prentice Hall. r(L + 2) . . Stoica and R. .
Lecture notes to accompany Introduction to Spectral Analysis by P. r(L + M .
con't Fact: SVD of rank :
( )=n =U V
U = (M n) with U U = In V = (n L) with V V = In
= (n n). y = .
The MinimumNorm solution is
b = .HighOrder and Overdetermined YW Equations. Moses. Prentice Hall.
(U V )b = .1U Important property: The additional (L . diagonal and nonsingular
Thus.
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. if the MinimumNorm solution b is used.V . 1997 Slide L5–14
. n) spurious zeros of B (q ) are located strictly inside the unit circle.
Practical Solution
Let
^= ^ ^ ^ ^
but made from
f^(k)g instead of fr(k)g. V be deﬁned similarly to U .HOYW Equations. Stoica and R.V ^ . this approach may give spurious frequency estimates when L > n. V from the
Let U . SVD of . Moses. r
. 1997
Slide L5–15
. Prentice Hall. .1U ^ ^ ^ b ^(q) that
Then !k n=1 are found from the n zeroes of B k are closest to the unit circle.
f^ g
When the SNR is low. Compute
^ = .
Lecture notes to accompany Introduction to Spectral Analysis by P. this is the price paid for increased accuracy when L > n.
Prentice Hall. 1997
Slide L6–1
. Stoica and R.Parametric Methods for Line Spectra — Part 2
Lecture 6
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses.
i(m. Moses. 1)
(for m
3 7 7 7 5
n)
2
46 y(t) = 6 ~
6 4
y(t . m + 1)]T e
4 R = E fy(t)~ (t)g = APA + 2I ~ y
Then
with
2
P = E fx(t)~ (t)g = ~ x
6 4
2 1
0
..
= Ax(t) + ~(t) ~ e
where
x(t) = x1(t) : : : xn(t)]T ~ ~(t) = e(t) : : : e(t . .. Stoica and R. Prentice Hall.1)! ]T A = a(!1) : : : a(!n)] (m n)
rank
Note: Deﬁne
(A) = n
y(t) y(t .i! : : : e.The Covariance Matrix Equation Let:
a(!) = 1 e. m + 1)
.
0
2 n
3 7 5
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L6–2
.
m
7 5
S G
#
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L6–3
.ng: orthonormal eigenvectors associated with f n+1 : : : m g S = s1 : : : sn] (m n) G = g1 : : : gm. Moses..n] (m (m .Eigendecomposition of R and Its Properties
R = APA + 2I (m > n)
Let:
1
2 :::
m: eigenvalues of R
fs1 : : : sng: orthonormal eigenvectors associated with f 1 : : : ng fg1 : : : gm. n))
Thus..
2
R = S G]
6 4
1
3
"
. Stoica and R. Prentice Hall.
. Moses. 1997
. 2 . Therefore. =0 A G=0
Slide L6–4
Lecture notes to accompany Introduction to Spectral Analysis by P. con't As rank
(APA ) = n: k > 2 k = 1 ::: n k = 2 k = n + 1 ::: m 0 1. Prentice Hall. Stoica and R.Eigendecomposition of R and Its Properties. 2
2 6 4 3 7 5 2 6 4
Note:
RS = APA S + 2S = S
1
0
.
0
n
3 7 5
with C (since rank S ... = = nonsingular 0 n.1) = AC S = A(PA S
( ) = rank(A) = n). since S G
j j 6= 0
4 ..
Prentice Hall. A G= G=0 a (!n) ) fa(!k )gn=1 ? R(G) k
2 6 4 3 7 5
Thus.
f!kgn=1 are the unique solutions of k a (!)GG a(!) = 0. Stoica and R.
Let:
1 N y(t)~ (t) ^ ~ y R = N t=m ^^ S G = S G made from the ^ eigenvectors of R
X
Lecture notes to accompany Introduction to Spectral Analysis by P.MUSIC Method
a (!1) . 1997
Slide L6–5
. . Moses.
(m.1 : : : z.1)GG a(z) = 0
that are closest to the unit circle. a (!)G ^
Root MUSIC Method:
]
f!kgn=1 = the angular positions of the n roots of: ^ k
^^ aT (z. Moses. 1997 Slide L6–6
. Here.1)]T
Note: Both variants of MUSIC may produce spurious frequency estimates.Spectral and Root MUSIC Methods Spectral MUSIC Method:
f!kgn=1 = the locations of the n highest peaks of the ^ k
“pseudospectrum” function:
1 ^G a(!) ! 2 .
a(z) = 1 z. Prentice Hall.
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R.
Stoica and R. If:
=n+1
m=n+1
Then:
^ g G = ^1. 1997
Slide L6–7
. ) f!kgn=1 can be found from the roots of ^ k aT (z. Moses. Prentice Hall.Pisarenko Method
Pisarenko is a special case of MUSIC with m (the minimum possible value).1)^1 = 0 g
no problem with spurious frequency estimates computationally simple (much) less accurate than MUSIC with m
n+1
Lecture notes to accompany Introduction to Spectral Analysis by P.
Stoica and R. Prentice Hall. Moses. with ﬁrst element equal ^ ^ to one. g
#
Lecture notes to accompany Introduction to Spectral Analysis by P. but only one vector in G (as in Pisarenko). 1997
Slide L6–8
. and reduce risk of false frequency estimates.
R( )
Let
"
1 = the vector in R(G). Uses m n (as in MUSIC).MinNorm Method
Goals: Reduce computational burden. that has minimum Euclidean norm.
Stoica and R. Prentice Hall. 1997
Slide L6–9
.1) 1 ^ g
"
#
that are closest to the unit circle. con't
Spectral MinNorm
f!gn=1 = ^k
the locations of the n highest peaks in the “pseudospectrum”
2 1 = a (!) 1 ^ g
" #
Root MinNorm
f!gn=1 = ^k
the angular positions of the n roots of the polynomial
aT (z.MinNorm Method. Moses.
Lecture notes to accompany Introduction to Spectral Analysis by P.
g
" #
MinNorm solution: As:
^ = . Stoica and R. at least. k k2) = (S S ) ) (S S ). k k2) ) ^ = . for N 1. 1997 Slide L6–10
.1 gm
" # " #
MinNorm Method: Determining g ^
Then:
1 2 R(G) ) S 1 = 0 ^ ^ ^ ^ g g ) S ^= .S =(1 . k k2) g
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall.1 g
+ S S .1 = =(1 .) ^^ I=S S=
Multiplying the above equation by gives:
(1 .Let S
^= S g1 . (S S ).S(S S ).1 exists iff = k k2 6= 1 (This holds. Moses.
D= 0 e.1]A
Then A2
Also.1 0]S S2 = 0 Im. where e.1]S
2 6 4 
3 7 5
Recall S
= AC with jC j 6= 0. Prentice Hall..1S1S2
Lecture notes to accompany Introduction to Spectral Analysis by P.. let
= A1D. 1997 Slide L6–11
. Moses. Stoica and R. determined as
is uniquely
= (S1S1).1 0]A A2 = 0 Im.1DC
{z
}
So has the same eigenvalues as D .ESPRIT Method
Let
A1 = Im. Then S2 = A2C = A1DC = S1 C .i!1 0 .i!n S1 = Im.
then S1 and S2 . ﬁnd S . Prentice Hall. Moses. Stoica and R.
^
^
^
^
The frequency estimates are found by:
where
f^kgn=1 are the eigenvalues of k ^ = (S1S1). arg(^k ) ^ k
ESPRIT Advantages: computationally simple no extraneous frequency estimates (unlike in MUSIC or Min–Norm) accurate frequency estimates
Lecture notes to accompany Introduction to Spectral Analysis by P.ESPRIT Implementation From the eigendecomposition of R. 1997
Slide L6–12
.1S1S2 ^^ ^^
f!kgn=1 = .
1997
.Summary of Frequency Estimation Methods
Method Periodogram Nonlinear LS YuleWalker Pisarenko MUSIC MinNorm ESPRIT
Computational Accuracy / Risk for False Burden Resolution Freq Estimates small mediumhigh medium very high very high very high medium high medium small low none high high medium medium high small medium very high none
Recommendation:
Use Periodogram for mediumresolution applications
Use ESPRIT for highresolution applications
Slide L6–13
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. Prentice Hall. Stoica and R.
1997
Slide L7–1
. Moses. Stoica and R. Prentice Hall.Filter Bank Methods
Lecture 7
Lecture notes to accompany Introduction to Spectral Analysis by P.
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall.
(!) by a
2. 1997 Slide L7–2
. by assuming that ! is nearly constant over the bands
f ( )g
( )
!. Nonparametric Approach: Implicitly smooth ! !=. Stoica and R.
!+ ] N >1
1
2 is more general than 1. Moses. Parametric Approach: Parameterize ﬁnitedimensional model. (
=2 2 =1
N > 1 leads to the variability / resolution compromise
associated with all nonparametric methods.Basic Ideas
Two main PSD estimation approaches: 1. but 2 requires
to ensure that the number of estimated values = = ) is < N .
as well as (b) and (c). 2 ! + 2 ]
Note that assumptions (a) and (b).
!c
@Q Q !
.
Z
6 ~ (! ) @ !
(b)
Z
!c
^FB (!) ' 1 2
(a)
. Prentice Hall. Stoica and R.Filter Bank Approach to Spectral Estimation
H 6 (!)
1
!c
!
Filtered Signal
y (t)
. 1997
.with varying !
Power Calculation
Bandpass Filter c and xed bandwidth

Power in the band

Division by lter bandwidth
^(!c )

6
y
y
(! )
.
c ( )d = (') (!)
(a) consistent power calculation
(b) Ideal passband ﬁlter with bandwidth
(c)
( ) constant on 2 ! .
Slide L7–3
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses.
jH ( )j2 ( )d = ' 21
!+ !. are conﬂicting.
i!k = 1 eiN (~.t) t=1
X X
= N
where
(
X
1
k=0
hk y(N . k)
2
! .!) .Filter Bank Interpretation of the Periodogram
1 N y(t)e. Prentice Hall. 1997
. 1 H (!) = hk e ! N ei(~. 1 ~ hk = N otherwise 0
1
center frequency of H 3dB bandwidth of H
(!) = ! ~
(!) ' 1=N
Slide L7–4
Lecture notes to accompany Introduction to Spectral Analysis by P.i!t 2 ~ ^p(~) = ! 4 N t=1 2 N 1 ~ = N y(t)ei!(N .!) . 1 k=0
X
1 ei!k k = 0 : : : N . Moses. Stoica and R.
^( )
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. 1997
Slide L7–5
. Stoica and R.Filter Bank Interpretation of the Periodogram. and: narrow ﬁlter passband. con't
jH (!)j as a function of (~ . !
0 −5 −10
−15 dB
−20
−25
−30
−35
−40
−3
−2
−1 0 1 ANGULAR FREQUENCY
2
3
Conclusion: The periodogram p ! is a ﬁlter bank PSD estimator with bandpass ﬁlter as given above. power calculation from only 1 sample of ﬁlter output. Prentice Hall. for N = 50. !).
Stoica and R. Each ﬁlter covers a small band centered on ! .
~
provides several samples for power calculation. 1997
Slide L7–6
. Prentice Hall. This “multiwindow approach” is similar to the Daniell method. 2. This approach leads to Bartlett and Welch methods. more data points for the power calculation stage. and bandpass ﬁlter each subsample.Possible Improvements to the Filter Bank Approach 1.
Both approaches compromise bias for variance. and in fact are quite related to each other: splitting the data sample can be interpreted as a special form of windowing or ﬁltering. Split the available sample.
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. Use several bandpass ﬁlters on the whole sample.
i! : : : e. m) h y(t) ~ E jyF (t)j2 = h Rh R = E fy(t)~ (t)g ~ y
{z } 7 5 {z } n o
y(t)
3
H (! ) =
where a
X
m
(! ) = 1
k=0 e. 1997
Slide L7–7
. Stoica and R.im! ]T
hk e.Capon Method
Idea: Datadependent bandpass ﬁlter design. Moses. y(t . k)
2 6 4 
.i!k = h a(!)
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall.
yF (t) =
X
m
k=0

hk y(t . = h0 h1 : : : hm] .
Moses.
n
The Bandwidth
1 ' m+1 = (ﬁlter 1 length)
Conclusion Estimate PSD as:
+1 ^(!) = m^.1a
o
The power at the ﬁlter output is:
E jyF (t)j2 = h0Rh0 = 1=a (!)R.1a(!) which should be the power of y (t) in a passband centered on ! .Capon Method.m t=m+1
X
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L7–8
. Stoica and R. Prentice Hall. con't Capon Filter Design Problem:
min(h Rh) h
Solution:
subject to h
a(!) = 1
h0 = R.1 a (!)R a(!)
with
N 1 ^ y(t)~ (t) ~ y R = N .1a=a R.
Prentice Hall. but it splits the N data point sample into (N m) subsequences of length m with maximum overlap.
Lecture notes to accompany Introduction to Spectral Analysis by P.
Capon uses one bandpass ﬁlter only.
.Capon Properties
m is the user parameter that controls the compromise
between bias and variance: – as m increases. Moses. 1997
Slide L7–9
. bias decreases and variance increases. Stoica and R.
s)e. 1997
Slide L7–10
.m m + 1 1 m m ^(t .Relation between Capon and BlackmanTukey Methods
Consider
^BT (!) with Bartlett window: m m + 1 .s) = m+1 r t=0 s=0 ^ a (!)Ra(!) R = ^(i .i!k ^ k=. Stoica and R. Moses. jkj ^BT (!) = r(k)e.i!(t. j )] ^ r = m+1
X X X
Then we have
^ a (!)Ra(!) ^BT (!) = m+1 m+1 ^C (!) = ^ a (!)R. Prentice Hall.1a(!)
Lecture notes to accompany Introduction to Spectral Analysis by P.
C ! generally has worse resolution and bias properties than the AR method. Prentice Hall.
AR Then
^C (!) =
1 1 m 1= ^AR(!) m + 1 k=0 k
X
Consequences: Due to the average over k.
^( )
^( )
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. Due to the loworder AR terms in the average. Stoica and R.Relation between Capon and AR Methods Let
2 ^k ^k (!) = ^ 2 jAk(!)j be the kth order AR PSD estimate of y (t). 1997
Slide L7–11
. C ! generally has less statistical variability than the AR PSD estimator.
Stoica and R. 1997
Slide L8–1
.Spatial Methods — Part 1
Lecture 8
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. Moses.
electromagnetic.@ . . seismometers Applications: Radar. @.. @. @. communications.
v Source n
@. @.The Spatial Spectral Estimation Problem
Source 1
vSource 2
B B B B N B cccc ccc c c c c c
v
@ . @. sonar. . mechanical Receiving sensors: Hydrophones. Emitted energy: Acoustic.R . underwater surveillance Basic Approach: Determine energy distribution over space (thus the name “spatial spectral analysis”)
Lecture notes to accompany Introduction to Spectral Analysis by P. seismology. . antennas. Moses. Stoica and R. ..
Sensor 1
@..@ . Prentice Hall.
Sensor 2 Sensor
m
Problem: Detect and locate n radiating sources by using an array of m passive sensors. 1997 Slide L8–2
.
Stoica and R.Simplifying Assumptions
Farﬁeld sources in the same plane as the array of sensors Nondispersive wave propagation
Hence: The waves are planar and the only location parameter is direction of arrival (DOA) (or angle of arrival. Moses. 1997
Slide L8–3
. the array is calibrated. Prentice Hall.)
Lecture notes to accompany Introduction to Spectral Analysis by P. 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.
)
hk (t) = ek (t) =
Note:
t 2 R (continuoustime signals). Prentice Hall. Stoica and R. etc.g.
Basic Problem: Estimate the time delays known but x t unknown.Array Model — Single Emitter Case
x(t) =
k=
the signal waveform as measured at a reference point (e.
()
f kg with hk(t)
This is a timedelay estimation problem in the unknown input case. Then the output of sensor k is yk (t) = hk (t) x(t . thermal noise in sensor electronics... background noise. k) + ek (t) ( = convolution operator). 1997 Slide L8–4
. Moses.
Lecture notes to accompany Introduction to Spectral Analysis by P.g. at the “ﬁrst” sensor) the delay between the reference point and the kth sensor the impulse response (weighting function) of sensor k “noise” at the kth sensor (e.
Moses. k ) ' (t) cos !ct + '(t) . the kth sensor output is
yk (t) = jHk (!c)j (t) cos !ct + '(t) . Stoica and R. !c k + arg Hk (!c)] + ek(t)
Slide L8–5
Lecture notes to accompany Introduction to Spectral Analysis by P. k ) ' '(t) Time delay is now ' to a phase shift !c k : x(t . Then:
x(t) = (t) cos !ct + '(t)] where (t) '(t) vary “slowly enough” so that (t . k ) ' jHk (!c)j (t)cos !ct + '(t) . !c k + argfHk (!c)g]
where Hk function
(!) = Ffhk(t)g is the kth sensor's transfer
Hence. k) ' (t) '(t . !c k ] hk (t) x(t . 1997
. Prentice Hall.Narrowband Assumption
Assume: The emitted signals are narrowband with known carrier frequency !c .
i !ct+ (t)] 2 Demodulate z (t) (translate to baseband): 2z(t)e.i 2!ct+ (t)] g
n o  {z }  {z }
Lowpass ﬁlter
2z(t)e.
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall.i!c k + ek (t) where s(t) is the complex envelope of x(t).i!ct to obtain (t)ei (t)
lowpass
highpass
Hence. Moses. 1997 Slide L8–6
or
.Complex Signal Representation The noisefree output has the form:
z(t) = (t) cos !ct + (t)] = = (t) ei !ct+ (t)] + e.i!ct ~ = yk(t) cos(!ct) .!ct = (t)f ei (t) + e. by lowpass ﬁltering and sampling the signal
yk (t)=2 = yk(t)e. Stoica and R.i! + ek (t)

s(t)
{z
} 
Hk (!c)
{z
k
c
}
c k
yk (t) = s(t)Hk (!c) e. 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.
Vector Representation for a Narrowband Source
Let
= m =
a( ) =
2 6 4
the emitter DOA the number of sensors
2 6 4
H1(!c) e;i!c
. .
1
3 7 5
Hm(!c) e;i!c m y1(t) e1(t) . . y(t) = e(t) = . . ym(t) em(t)
3 7 5 2 6 4
3 7 5
Then
y(t) = a( )s(t) + e(t)
NOTE: enters a via both For omnidirectional sensors the on .
()
f kg and fHk(!c)g. fHk(!c)g do not depend
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L8–7
Analog Processing Block Diagram
Analog processing for each receiving array element
SENSOR DEMODULATION
A/D CONVERSION


Internal Noise External Noise XXXXX ? XX ll z l  Cabling and , Filtering : ,, Incoming Transducer Signal x(t ; k )
Re yk (t)] Lowpass Re yk (t)]  !! ~ Filter A/D cos(!ct) 6 y k (t) Oscillator sin(!ct) ? ~  Im yk (t)] Lowpass Im yk (t)]  !! Filter A/D

Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997
Slide L8–8
Multiple Emitter Case Given n emitters with received signals: DOAs:
fsk(t)gn=1 k
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 .
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. Moses, Prentice Hall, 1997 Slide L8–9
Moses. Stoica and R. Prentice Hall. 1997
Slide L8–10
. 1) c
Lecture notes to accompany Introduction to Spectral Analysis by P.Uniform Linear Arrays
Source h
JJ J J
J J
JJ J ^ J
J
JJ
JJ+ JJ
JJ
2
] J d sin J ^ JJ J JJ 3 J J
v
4
1
v ]? v
d 
v
ppp
m
vSensor
ULA Geometry Sensor #1 = time delay reference Time Delay for sensor k:
where c = wave propagation speed
d sin k = (k .
i!s : : : e. Stoica and R.
!s = spatial frequency The function !s 7! a( ) is onetoone for j!sj $ dj sin j 1 d =2
As
=2
d
=2 is a spatial Shannon sampling theorem. Prentice Hall. Moses.i(m.Spatial Frequency
Let:
sin 4 !s = !c d sin = 2 dc=f = 2 d sin c c = c=fc = signal wavelength a( ) = 1 e.
Slide L8–11
d = spatial sampling period
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
.1)!s ]T By direct analogy with the vector a(! ) made from
uniform samples of a sinusoidal time series.
Stoica and R. 1997
Slide L9–1
.Spatial Methods — Part 2
Lecture 9
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. Moses.
Stoica and R. Prentice Hall. Moses.
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L9–2
.Spatial Filtering
Spatial ﬁltering useful for DOA discrimination (similar to frequency discrimination of timeseries ﬁltering) Nonparametric DOA estimation
There is a strong analogy between temporal ﬁltering and spatial ﬁltering.
m + 1)]T
If u
m.1 X
(t) = ei!t then yF (t) = h a(!)] u(t)
 {z }
ﬁlter transfer function
a(!) = 1 e. k) = h y(t) k=0 h = ho : : : hm. Stoica and R.i! : : : e. Moses.Analogy between Temporal and Spatial Filtering
Temporal FIR Filter:
yF (t) = hk u(t .i(m.
Lecture notes to accompany Introduction to Spectral Analysis by P.1] y(t) = u(t) : : : u(t .1)! ]T We can select h to enhance or attenuate signals with different frequencies ! . Prentice Hall. 1997
Slide L9–3
.
i!c 2 : : : e. Prentice Hall. Stoica and R.Analogy between Temporal and Spatial Filtering Spatial Filter:
fyk(t)gm=1 = k
the “spatial samples” obtained with a sensor array. 1997 Slide L9–4
.i!c m ]T
yF (t) = h a( )] s(t)
 {z }
The corresponding ﬁlter output is
ﬁlter transfer function
We can select h to enhance or attenuate signals coming from different DOAs.
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses.
Spatial FIR Filter output:
yF (t) =
X
m
k=1
hk yk (t) = h y(t)
Narrowband Wavefront: The array's (noisefree) response to a narrowband ( sinusoidal) wavefront with complex envelope s t is:
() y(t) = a( )s(t) a( ) = 1 e.
Analogy between Temporal and Spatial Filtering
u(t) = ei!t
z
1
s q. s qq
h

?
1
?

m.i! C @ h a(!)]u(t) R @ C . B .B . 1997 Slide L9–5
. Moses.. Prentice Hall.i! Be 2 B !! B B B B ..
?
qq
1
!
(Spatial sampling)
(b) Spatial ﬁlter
Lecture notes to accompany Introduction to Spectral Analysis by P.? B P C B C B B B .i! m  e {z a( ) aa m!
point
( )
qq
( )
1 C C C C C Cs(t) C C C C C A }

? @ ? @ @
h

1
@ R @ 
P
h a( )]s(t) 
hm. Cu(t) C B . ! hm. B B @ .  e {z } ? a(!)
1
?
0
qq
(
1)
1
(Temporal sampling)
(a) Temporal ﬁlter
narrowband source with DOA=
Z
z
Z
Z
Z ~ aa 1reference !!
h
0
0 1 B aa 2. Stoica and R.1
q .1
? 
1 .i m. C C B C B C B C B C B A @ .@ 0 B 1 C h @@ C B B e.
where 0
j ( )j
= ( )
= 25
0
30
−30
Th et a (d eg )
60
−60
90 0 2 4 6 Magnitude 8 10
−90
Lecture notes to accompany Introduction to Spectral Analysis by P. h a 0 . Prentice Hall. Here. con't
Example: The response magnitude h a of a spatial ﬁlter (or beamformer) for a 10element ULA.Spatial Filtering. 1997
Slide L9–6
. Stoica and R. Moses.
Spatial Filtering Uses
Spatial Filters can be used To pass the signal of interest only. Prentice Hall. by sweeping the ﬁlter through the DOA range of interest (“goniometer”). Moses. hence ﬁltering out interferences located outside the ﬁlter's beam (but possibly having the same temporal characteristics as the signal). Stoica and R. To locate an emitter in the ﬁeld of view. 1997
Slide L9–7
.
Lecture notes to accompany Introduction to Spectral Analysis by P.
1997
.Nonparametric Spatial Methods
A Filter Bank Approach to DOA estimation. and evaluate the powers of the ﬁltered signals:
E jyF (t)j2 = E jh ( )y(t)j2 = h ( )Rh( ) with R = E fy (t)y (t)g.
n o n
o
The (dominant) peaks of h DOAs of the sources. Prentice Hall. 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. Moses. Stoica and R.
( )Rh( ) give the
Slide L9–8
Lecture notes to accompany Introduction to Spectral Analysis by P.
Beamforming Method
Assume the array output is spatially white:
Then:
R = E fy(t)y (t)g = I E jyF (t)j2 = h h
n o
Hence: In direct analogy with the temporally white assumption for ﬁlter bank methods. 1997
Slide L9–9
. Stoica and R. y t can be considered as impinging on the array from all DOAs.
()
Filter Design:
min (h h) h
Solution:
subject to
h a( ) = 1
h = a( )=a ( )a( ) = a( )=m E jyF (t)j2 = a ( )Ra( )=m2
n o
Lecture notes to accompany Introduction to Spectral Analysis by P. Moses. Prentice Hall.
Resolution Threshold:
inf j k .
1
= 1. pj > =
wavelength array length array beamwidth
Inconsistency problem: Beamforming DOA estimates are consistent if n inconsistent if n > .Implementation of Beamforming
1 N y(t)y (t) ^ R=N t=1
X
The beamforming DOA estimates are:
f^kg =
the locations of the n largest peaks of a Ra . Prentice Hall. 1997
. Moses. Stoica and R.
( )^ ( )
This is the direct spatial analog of the BlackmanTukey periodogram. but
Slide L9–10
Lecture notes to accompany Introduction to Spectral Analysis by P.
1997 Slide L9–11
.1a( )=a ( )R.1a( ):
Performance: Slightly superior to Beamforming. Stoica and R.1a( ) E jyF (t)j2 = 1=a ( )R.
Lecture notes to accompany Introduction to Spectral Analysis by P.Capon Method Filter design:
min(h Rh) subject to h a( ) = 1 h
Solution:
h = R. Moses. Prentice Hall.1a( )
n o
Implementation:
f^kg =
the locations of the n largest peaks of
^ 1=a ( )R. 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).
MUSIC. almost without modiﬁcation. for DOA estimation. Moses. Prentice Hall. MINNORM and ESPRIT methods of frequency estimation can be used.
Then:
R = E fy(t)y (t)g = APA + 2I
Thus: The NLS.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. Stoica and R. 1997 Slide L9–12
.
Lecture notes to accompany Introduction to Spectral Analysis by P. YW.
1A ]Rg ^ Thus. A(A A).1A ]y(t) = N t=1 ^ = trf I . 1997 Slide L9–13
=1
.Nonlinear Least Squares Method
1 N ky(t) . this is precisely the form of the NLS method of frequency estimation.1A y(t) t = 1 : : : N s
X
Then
1 N k I .1A ]Rg
X
f kg
For N .
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. Moses.1A ]y(t)k2 f ( ^) = N s t=1 1 N y (t) I . A(A A). A(A A). As(t)k2 min(t)g N f k g fs t=1
X 
Minimizing f over s gives
f ( s)
{z
}
^(t) = (A A). f ^k g = arg max trf A(A A). Stoica and R.
Prentice Hall.
Lecture notes to accompany Introduction to Spectral Analysis by P. 1997
Slide L9–14
. Stoica and R.Nonlinear Least Squares Method
Properties of NLS: Performance: high Computational complexity: high Main drawback: need for multidimensional search. Moses.
Stoica and R.) = n.n 0 V2 g L n M . = U1 U2 ] 0 n n 0 V1 g n . Moses. and the SVD of . Prentice Hall. 1997
Slide L9–15
.= y
(M L)
Also assume:
M > n L > n ( ) m = M + L > 2n)
rank
~ (A) = rank(A) = n
" # " #
(.n ~ ~ Properties: A V2 = 0 V1 2 R(A)
Then: rank
{z} {z}
Lecture notes to accompany Introduction to Spectral Analysis by P.YuleWalker Method
e y(t) = y(t) = A s(t) + ~(t) ~ y(t) ~ e(t) A
" # " # "
#
Assume: Then:
E fe(t)~ (t)g = 0 e
4 E fy(t)~ (t)g = AP A ~ . is .
=N y t=1
X
Properties: Computational complexity: medium Performance: satisfactory if m
2n
Main advantages: – weak assumption on e t – the subarray A need not be calibrated
f ( )g
Lecture notes to accompany Introduction to Spectral Analysis by P. Prentice Hall. 1997
Slide L9–16
.YWMUSIC DOA Estimator
f^kg =
where
the n largest peaks of
1=~ ( )V2V2 ~( ) a ^^a
~( ). Stoica and R. using 1 N y(t)~ (t) ^ . (L 1). Moses. is the “array transfer vector” for y(t) a ~
at DOA
^ V2 is deﬁned similarly to V2.
that is. 1997
Slide L9–17
.MUSIC and MinNorm Methods
Both MUSIC and MinNorm methods for frequency estimation apply with only minor modiﬁcations to the DOA estimation problem. Prentice Hall. Spectral forms of MUSIC and MinNorm can be used for arbitrary arrays Root forms can be used only with ULAs MUSIC and MinNorm break down if the source signals are coherent. Moses.
Lecture notes to accompany Introduction to Spectral Analysis by P. Stoica and R. if rank
(P ) = rank(E fs(t)s (t)g) < n
Modiﬁcations that apply in the coherent case exist.
. 1997 Slide L9–18
( ) = d sin( )=c
. Stoica and R.
Lecture notes to accompany Introduction to Spectral Analysis by P.ESPRIT Method Assumption: The array is made from two identical subarrays separated by a known displacement vector. Prentice Hall. Let
m = # sensors in each subarray A1 = Im 0]A (transfer matrix of subarray 1) A2 = 0 Im]A (transfer matrix of subarray 2)
Then
A2 = A1D. Moses.i!c ( n) ( ) = the time delay from subarray 1 to
2 6 4
3 7 5
subarray 2 for a signal with DOA = : where d is the subarray separation and is measured from the perpendicular to the subarray displacement vector.i!c ( 1) 0 . D= 0 e. where e..
so m m and
.1 A1 = Im. con't ESPRIT Scenario
source θ
known
subarray 1
displac ement v
ector
subarray 2
Properties: Requires special array geometry Computationally efﬁcient
No risk of spurious DOA estimates
Does not require array calibration Note: For a ULA.1 A2 = 0 Im.1]A
Lecture notes to accompany Introduction to Spectral Analysis by P.ESPRIT Method.1 0]A
= . Stoica and R. Moses. the two subarrays are often the ﬁrst m and last m array elements. 1997
Slide L9–19
.1
. Prentice Hall.