Scribd
Upload
248 views34 pages

9 - Correlation, Energy Spectral Density, and Power Spectral Density

This document discusses signal energy, power, correlation, autocorrelation, and spectral density. It defines signal energy and power, and explains that a signal can be an energy signal, a power signal, or neither. Correlation and autocorrelation are introduced as measures of similarity between signals. The properties of autocorrelation for energy and power signals are covered. Finally, energy spectral density and power spectral density are defined as measures related to the Fourier transform of autocorrelation.

Uploaded by

Christine Allen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
248 views34 pages

9 - Correlation, Energy Spectral Density, and Power Spectral Density

This document discusses signal energy, power, correlation, autocorrelation, and spectral density. It defines signal energy and power, and explains that a signal can be an energy signal, a power signal, or neither. Correlation and autocorrelation are introduced as measures of similarity between signals. The properties of autocorrelation for energy and power signals are covered. Finally, energy spectral density and power spectral density are defined as measures related to the Fourier transform of autocorrelation.

Uploaded by

Christine Allen
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 34

Signals & Systems

Lecture 9 – Correlation, Energy Spectral


Density, and Power Spectral Density

Alp Ertürk
[email protected]
Signal Energy & Power
• Signal energy & power:

𝐸= 𝑥(𝑡) 2 𝑑𝑡
−∞

𝑇
1
𝑃 = lim 𝑥(𝑡) 2 𝑑𝑡
𝑇→∞ 2𝑇
−𝑇
Signal Energy & Power
• A signal is called an energy signal if 𝐸 < ∞

• A signal is called a power signal if 0 < 𝑃 < ∞

• A signal can be an energy signal, a power signal or neither.

• But a signal cannot be both an energy signal and a power


signal
Correlation
• The correlation function between two signals x(t) and y(t) is
defined as:
∞

𝜙𝑥𝑦 𝑡 = 𝑥 𝜏 𝑦 ∗ 𝑡 + 𝜏 𝑑𝜏
−∞

• This is also referred as cross-correlation

• The function 𝜙𝑥𝑥 𝑡 is referred as autocorrelation


Correlation
• The correlation function between two energy signals:

𝑅𝑥𝑦 𝜏 = 𝑥 𝑡 𝑦 ∗ 𝑡 + 𝜏 𝑑𝑡
−∞

is equal to the area under the product of these two signals, as a


function of how much y is shifted with respect to x
Correlation
• Cross-correlation is a measure of similarity of two signals as a
function of the time shift of one relative to the other

• Cross-correlation is often used for searching a long signal for a


shorter feature / signal

• It can also be used to compare the similarity of two signals


Correlation
Correlation

• v(t) contains u(t) with an unknown delay and an added noise


Correlation
• Cross-correlation of u(t) and v(t) gives w(t). The lag between
the signals is found as t = 450
Correlation
• The signal y(t) below is the signal v(t) with even more noise.
The cross-correlation of u(t) and y(t) is found as z(t) below
Correlation
• Cross-correlation is similar to convolution
Correlation
• If h(t) is real:
∞ ∞

𝑅𝑥ℎ 𝜏 = 𝑥 𝑡 ℎ 𝑡 + 𝜏 𝑑𝑡 = 𝑥 𝜆 − 𝜏 ℎ 𝜆 𝑑𝜆
−∞ −∞

∞ ∞

𝑅ℎ𝑥 𝜏 = ℎ 𝑡 𝑥 𝑡 + 𝜏 𝑑𝑡 = ℎ 𝜆 − 𝜏 𝑥 𝜆 𝑑𝜆
−∞ −∞

= ℎ 𝑢 𝑥 𝑢 + 𝜏 𝑑𝜆
−∞

𝑅ℎ𝑥 𝜏 = 𝑅𝑥ℎ −𝜏
Correlation of Energy Signals
• The correlation function between two energy signals, when x and y are both
real: ∞

𝑅𝑥𝑦 𝜏 = 𝑥 𝑡 𝑦 𝑡 + 𝜏 𝑑𝑡
−∞

• This is very similar to the convolution of two real energy signals:


∞

𝑥 𝑡 ∗𝑦 𝑡 = 𝑥 𝜏 𝑦 𝑡 − 𝜏 𝑑𝜏
−∞

• Therefore it is possible to use convolution to find the correlation:


∞ ∞

𝑅𝑥𝑦 𝜏 = 𝑥 −𝜏 ∗ 𝑦 𝜏 = 𝑥 −𝜆 𝑦 𝜏 − 𝜆 𝑑𝜏 = 𝑥 𝑢 𝑦 𝜏 + 𝑢 𝑑𝑢
−∞ −∞
Correlation of Energy Signals
• Therefore it is possible to use convolution to find the correlation:

𝑅𝑥𝑦 𝜏 = 𝑥 −𝜏 ∗ 𝑦 𝜏

• In Fourier Domain:

𝐹
𝑅𝑥𝑦 𝜏 𝑋 ∗ 𝑓 𝑌(𝑓)
Correlation of Power Signals
• The correlation function between two power signals is the
average value of the product of 𝑥 and 𝑦 ∗ as a function of how
much 𝑦 ∗ is shifted relative to x

1
𝑅𝑥𝑦 𝜏 = lim 𝑥 𝑡 𝑦 ∗ 𝑡 + 𝜏 𝑑𝑡
𝑇→∞ 𝑇
𝑇

• If both signals are periodic and their fundamental periods have


a finite least common period:

1
𝑅𝑥𝑦 𝜏 = 𝑥 𝑡 𝑦 ∗ 𝑡 + 𝜏 𝑑𝑡
𝑇
𝑇
Correlation of Power Signals
• If both signals are periodic and their fundamental periods have
a finite least common period:

1
𝑅𝑥𝑦 𝜏 = 𝑥 𝑡 𝑦 ∗ 𝑡 + 𝜏 𝑑𝑡
𝑇
𝑇

• For real periodic signals, this equation becomes:

1
𝑅𝑥𝑦 𝜏 = 𝑥 𝑡 𝑦 𝑡 + 𝜏 𝑑𝑡
𝑇
𝑇
Autocorrelation
• Autocorrelation for energy signals:

𝑅𝑥𝑥 𝜏 = 𝑥 𝑡 𝑥 ∗ 𝑡 + 𝜏 𝑑𝑡
−∞

• At a shift of 0, we get:
∞

𝑅𝑥𝑥 0 = 𝑥(𝑡) 2 𝑑𝑡
−∞
Autocorrelation
• Autocorrelation for power signals:

1
𝑅𝑥𝑥 𝜏 = lim 𝑥 𝑡 𝑥 ∗ 𝑡 + 𝜏 𝑑𝑡
𝑇→∞ 𝑇
𝑇

• Average signal power:

1
𝑅𝑥𝑥 0 = lim 𝑥(𝑡) 2 𝑑𝑡 = 𝑃𝑥
𝑇→∞ 𝑇
𝑇
Properties of Autocorrelation
• For real signals, autocorrelatio is an even function

𝑅𝑥𝑥 𝜏 = 𝑅𝑥𝑥 −𝜏

• Autocorrelation magnitude can never be larger than it is at


zero shift:
𝑅𝑥𝑥 0 ≥ 𝑅𝑥𝑥 𝜏

• If a signal is time shifted, its autocorrelation does not change

• The autocorrelation of a sum of sinusoids of different


frequencies is the sum of the autocorrelation of the individual
sinusoids
Autocorrelation examples
Autocorrelation examples
• Autocorrelation for a cosine and a sine:
Autocorrelation examples
• Autocorrelation for a cosine burst and a sine burst
Autocorrelation examples
• Autocorrelation for a cosine burst and a sine burst
Autocorrelation examples
• Three random power signals and their autocorrelations:
Autocorrelation examples
• Four different random signals with identical autocorrelations:
Autocorrelation examples
• Four different random signals with identical autocorrelations:
Autocorrelation examples
• Four different random signals with identical autocorrelations:
Matched Filters
• Matched filter is a useful technique for detecting the
presence of a signal of a certain shape in the presence of
noise

• Matched filters use correlation to detect the signal

• Often used to detect 1’s and 0’s in a binary data stream


Matched Filters
• The optimal filter to detect a noisy signal is one whose
impulse response is proportional to the time inverse of the
signal
Matched Filters
• Even in the presence of a large additive noise signal, the
matched filter below indicates the presence of a 1 and the
presence of a 0
Energy Spectral Density
• Total signal energy in an energy signal:
∞ ∞

𝐸𝑥 = 𝑥(𝑡) 2 𝑑𝑡 = 𝑋(𝑓) 2 𝑑𝑓
−∞ −∞

• The quantity 𝑋(𝑓) 2 is called the energy spectral density


(ESD) of the signal x, and conventionally given the symbol, Ψ

Ψ 𝑓 = 𝑋(𝑓) 2
Energy Spectral Density
• Energy spectral density (ESD):

2
Ψ 𝑓 = 𝑋(𝑓)

• For an energy signal, ESD and autocorrelation form a Fourier


transform pair.

𝐹
𝑅𝑥 𝑡 Ψ𝑥 𝑓

• If x is a real-valued signal, ESD is even, non-negative and real


Energy Spectral Density
• The ESD of the response of an LTI system is related to the
ESD of the system by:

Ψ𝑦 𝐹 = 𝐻(𝐹) 2 Ψ𝑥 𝐹 = 𝐻(𝐹)𝐻 ∗ (𝐹)Ψ𝑥 𝐹


Power Spectral Density
• Power spectral density (PSD) applies to power signals in the
same way that energy spectral density applies to energy
signals.

• The PSD of a signal x is conventionally indicated by 𝐺𝑥 𝑓

• For an power signal, PSD and autocorrelation form a Fourier


transform pair.

• For an LTI system:

𝐺𝑦 𝐹 = 𝐻(𝐹) 2 𝐺𝑥 𝐹 = 𝐻(𝐹)𝐻 ∗ (𝐹)𝐺𝑥 𝐹

You might also like