Академический Документы
Профессиональный Документы
Культура Документы
Ioan Tabus
1 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
Overview
The slides follow Chapters 1 and 2 from P. Stoica, R. Moses, ”Spectral analysis of signals ”, available on-line at
http://user.it.uu.se/~ps/SAS-new.pdf
2 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
I Goal: Given a finite number of values y1 , y2 , . . . , yN of a stationary signal, estimate the power over
narrow frequency bands.
I Applications:
I the power over frequency represents a signature of the signal, by which it can be identified,
classified, compared.
I music decomposition over the semitones of the scale for automatic music transcription
I speech decomposition over bark or mel scales, for speech recognition
I monitoring the functioning of industrial machines or building and bridge structure
I characterization of physiological waves (EEG,EKG)
I estimate the direction of targets in underwater surveillance
I First we introduce the definitions of spectral density for deterministic signals, then for random signals, and
then we state some properties of the spectral density.
I The non-parametric and parametric estimation methods are discussed in next lectures.
3 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
Deterministic signals
I Given a discrete-time signal {yt }∞
t=−∞ having finite energy
∞
2
X
yt < ∞
t=−∞
∞
−jωt
X
Y (ω) = y (t)e
t=−∞
Here Y (ω) has a complex value and is the transform coefficient at frequency ω. The absolute value |Y (ω)|
is the amplitude and arg(Y (ω)) is the phase corresponding to transform coefficient Y (ω) at frequency ω.
I The Parseval inequality holds for the DTFT:
∞ Z π
X 2 1 2
yt = |Y (ω)| dω
t=−∞ 2π −π
The lefthand side (l.h.s) is the total energy of the signal, the r.h.s is the integral of the squared amplitude
over frequency. Hence the term |Y (ω)|2 has the significance of a density of energy over the frequency axis.
I Energy spectral density definition
2
S(ω) = |Y (ω)|
and
P∞from Parseval identity the distribution of energy sums up to the total energy of the signal
2 1 Rπ
t=−∞ yt = 2π −π S(ω)dω.
4 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
Deterministic signals
I Property: Define for all k the ”autocorrelation” of the deterministic signal (note the missing division by the
length of the signal, which is ∞)
∞
X
ρ(k) = yt yt−k = ρ(−k)
t=−∞
Then
∞
−jωk
X
S(ω) = ρ(k)e
k=−∞
I Proof:
∞ ∞
2 ∗ −jωt +jωτ
X X
S(ω) = |Y (ω)| = Y (ω)Y (ω) = y (t)e y (τ )e
t=−∞ τ −∞
∞ ∞ ∞ ∞
X X jω(τ −t) (k=t−τ ) X X −jωk
= y (t)y (τ )e = y (t)y (t − k)e
t=−∞ τ =−∞ t=−∞ k=−∞
∞ ∞ ∞
−jωk −jωk
X X X
= e y (t)y (t − k) = ρ(k)e
k=−∞ t=−∞ k=−∞
I The two equivalent definitions of S(ω), the first as S(ω) = |Y (ω)|2 and second
P∞ −jωk
S(ω) = k=−∞ ρ(k)e will be used for defining power spectral density for random signals.
5 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
6 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
∞
−jωk
X
P(ω) = r (k)e
k=−∞
Z π
1 jωk
r (k) = P(ω)e dω
2π −π
Z π
2 1
r (0) = E [yt ] = P(ω)dω
2π −π
This relation reinforces the power density interpretation: autocorrelation at lag 0 is the power of the signal,
which is equal to the integral of the power density over all frequencies.
I One can interpret P(ω)dω as the infinitesimal power in the band (ω − dω
2
, ω + dω
2
)
7 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
I The second definition generalizes to random variables the definition S(ω) = |Y (ω)|2 used for
deterministic signals of finite energy.
I Define the finite Discrete Fourier transform of a sequence y1 , y1 , . . . , yN as
N
−jωt
X
YN (ω) = yt e
t=1
1
2
P(ω) = lim E |YN (ω)|
N→∞ N
2
1 N
X −jωt
P(ω) = lim E yt e
N→∞ N
t=1
8 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
N X
X N N−1
X
f (t − s) = (N − |τ |)f (τ )
t=1 s=1 τ =−N+1
N X
N N−1
−jω(t−s) −jωτ
X X
= r (t − s)e = (N − |τ |)r (τ )e
t=1 s=1 τ =−N+1
where the last equality comes from the preliminary result on double sums. Now the second definition gives
2
1 X N N−1
1
−jωt −jωτ
X
lim E y t e = N→∞
lim (N − |τ |)r (τ )e
N→∞ N N
t=1 τ =−N+1
∞ N−1 ∞
X −jωτ 1 X −jωτ
X −jωτ
= r (τ )e − lim |τ |r (τ )e = r (τ )e = P(ω)
N→∞ N τ =−N+1
τ =−∞ τ =−∞
1 PN−1
since the term limN→∞ N |τ |r (τ )e −jωτ is zero if r (k) decays rapidly.
τ =−N+1
9 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
P(ω) ≥ 0 ∀ω
n o
which can be seen from the second definition P(ω) = limN→∞ E 1
N
|YN (ω)|2 where the squared
value is always nonnegative.
I For real valued signals yt the function P(ω) is even, i.e.,
Proof:
∞ ∞ ∞
−jωk −jωk jωk
X X X
P(ω) = r (k)e = r (0) + r (k)(e +e ) = r (0) + 2 r (k)cos(ωk)
k=−∞ k=1 k=1
∞
X
= r (0) + 2 r (k)cos(−ωk) = P(−ω)
k=1
10 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
2
Py (ω) = |H(ω)| Pe (ω) ∀ω
where H(ω) = H(z) z=e jω .
e y
t t
H(q)
P (ω)
e Py(ω)=|H(ω)|2Pe(ω)
11 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
∞
X
yt = H(q)et = hk et−k
k=−∞
∞
X ∞
X ∞
X ∞
X
ry (k) = Eyt yt−k = E hm et−m hs et−k−s = hm hs E [et−m et−k−s ]
m=−∞ s=−∞ m=−∞ s=−∞
∞
X ∞
X
= hm hs re (k + s − m)
m=−∞ s=−∞
∞ ∞ ∞ ∞
−jωk −jωk
X X X X
Py (ω) = ry (k)e = e hm hs re (k + s − m)
k=−∞ k=−∞ m=−∞ s=−∞
∞ ∞ ∞
−jωm jωs −jω(k+s−m)
X X X
= hm e hm e r (k + s − m)e
m=−∞ s=−∞ k+s−m=−∞
∗ 2
= H(ω)H (ω)Pe (ω) = |H(ω)| Pe (ω)
12 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
z1,2 = 0.9e ±jπ/3 , z3,4 = 0.7e ±j3π/4 , z5,6 = 0.99e ±jπ/6 , and the white noise et has zero mean and
standard deviation σe = 1.
I The power spectral density of the white noise is Pe (ω) = σe2 = 1.
I The power spectral density of the random process yt obtained by passing white noise et through the linear
1 is
system with transfer function A(z)
1 2
Py (ω) = σe
|A(e jω )|2
I short realizations of the white noise and of the random process yt are shown below
One realization of the white noise et
4
2
et
−2
−4
0 50 100 150 200 250 300 350 400
t ime t
One realization of AR process
20
10
yt
−10
−20
0 50 100 150 200 250 300 350 400
t ime t
13 / 14
Deterministic signals
Power spectral density -definitions
Power spectral density -properties
10 3
5 2
logP ( ω)
yt
0 1
−5 0
−10 −1
−15 −2
0 50 100 150 200 250 300 350 400 0 0.5 1 1.5 2 2.5 3 3.5
t im e t fr e que nc y ω
Estimated spectrum from one realization Estimated overlapped with true spectrum
4 4
3 3
logP̂ ( ω) , log P̂ ( ω)
2 2
logP̂ ( ω)
1 1
0 0
−1 −1
−2 −2
−3 −3
−4 −4
0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5
fr e que nc y ω fr e que nc y ω
14 / 14