Chapter 3

Analysis and Transmission of Signals

 Introduction
• Engineers view signals in terms of frequency
spectra
• Audio signals having bandwidth of 20 kHz
• Loudspeakers responding to 20 kHz audio
signal
• We will study spectra representation of
aperiodic signals
 Aperiodic Representation
• Aperiodic signal g(t) can be represented as a continuous sum
(integral) of everlasting exponential
• We have to construct a new periodic gT0(t) by repeating signal
g(t) every T0 seconds.
• gT0(t) can be represented as exponential Fourier series
• As T0 -> ∞ :

• Exponential Fourier series for gT0(t)

 Aperiodic Representation
• Integrating gT0 over (-T0/2, T0/2) is the same as integrating g(t)

• If we define a function G(f)as a function ω

 
G( f )  g(t)e

 jt
dt  

g(t)e  j 2ft dt
1
Dn  G (nf 0 )
T0

• Shows that Fourier Coefficient Dn are (1/T0 times) the samples of

G(f) uniformly spaced at intervals of f 0 Hz

• (1/T0)(G(f)) is the envelope for the coefficient Dn

 Aperiodic Representation
• If T0 is doubled, the envelop is halved and magnitude
gets smaller as T0 is doubled more.
• Shape remains the same

• gT0(t) becomes 
G(nf 0 ) jn 2f0t
g T0 (t)  
n T0
e

• Because T0 ->∞, f0 = 1/T0 becomes infinitesimal (f0->0).


Hence ∆f = 1/T0 g T (t)   [ G ( n  f )f ]e ( j 2 nf )t
0
n  

• gT0(t) can be expressed as a sum of everlasting exponentials

of 0,f ,2f ,3f .....( fourierseries)
• The amount of the component of frequency n∆f is
G(n∆f)∆f
• As T0 -> ∞, ∆f ->0 and gT0(t) ->g(t):


lim  G(nf )e
( j 2nf )t
g(t)  lim gT0(t)  f
T0  f 0 

• This is the area under G(f)ej2πft called Fourier integral:


g(t)   G( f )e j2ft
dt

 Fourier Tranform
• Direct Fourier transform of g(t)
– G(f)   g(t)e  j 2ft dt
 
• Inverse Fourier Transform of G(f)


g(t)   G( f )e j2ftdt


• Symbolically: G ( f )  F [ g (t)]
g (t)  F 1
[G( f )]
• We can plot G(f) as a function of f and amplitude and
angle (phase) spectra exist
 Conjugate Symmetry Property
• If g(t) is a real function of t, G(f) and G(-f) are complex
conjugate
G(-f) = G*(f)
• This means:
|G(-f)| = |G(f)|
θg(-f) = -θg(f)
 Find the Fourier Transform of e-atu(t)
• hi

• Expressing a+jω in polar form

• The amplitude spectrum |G(f)| and the phase
spectrum θg(f) is shown:
 Existence of Fourier Transform
• Fourier transform exist for a function g(t) if:

• The physical existence of a signal is a sufficient condition

for the existence of its transform
• The fourier transform is linear if:
 Transform of Useful Functions
• Unit Rectangular Function

• The unit rectangular pulse rect(x) in (a) is expanded by a

factor τ rect(x/τ) in (b)
 Unit Triangular function
• Triangular pulse ∆(x) of unit height and unitwidth,
centered at the origin:
 Sinc function
• This is the function sin x/x (sine over argument)
• Plays an important role in signal processing
sinc(x) = sin x/x
• It is an even function of x
• Sinc (x) = 0 when sin x = 0 except at x = 0, i.e sinc (x) = 0
for t =   ,2  ,3 
• Sinc(0) = 1
• Sinc (x) is an oscillating function with decreasing
amplitude.
• It has a unit peak at x = 0 and zero crossings at integer
multiples of π
• Product of sin x (T0 = 2π) and 1/x
 Find the Fourier transform of g(t) = Π(t/τ)
• Solution  t
G( f )   ( )e j2ft dt
 
• Since Π(t/τ)=1 for |t| < τ/2, and since it is zero
for |t|> τ/2 
G ( f )   2 e  j 2 ft dt

2
1 2 sin( f  )
 (e  jf  e jf ) 
j2  f 2 f
sin( f  )
   sin c( f  )
f 
 t  
     sin c 
    sin c( f  )

   2 
• Since sinc(x) = 0 when x = ±nπ
• Sinc(ωt/2) = 0 when ωt/2 = ±nπ when f = ±n/τ (n = 1, 2 ,
3, ……)
• G(f) is real, hence spectra is just a single plot of G(f)
• See plot in the text
• The spectrum peaks at f = 0 and decays at higher
frequencies.
• Π(t/τ) is a low-pass signal with most of the signal energy
in lower frequency components
• Signal bandwidth – difference between highest
(significant) frequency and lowest (significant) frequency
in the signal spectrum
• A rough estimate of bandwidth of a rectangular pulse is
2π/τ rad/s or 1/τ Hz
Find the Fourier transform of the unit impulse signal (δ(t))
• Using sampling property of impulse function

F[ (t)]   (t)e  j2ft dt  e j 2f .0  1

 Find the inverse Fourier transform of δ(2πf) = (1/2π)δ(f)
• Using sampling property of impulse function (pg. 33)

 1 
F [ (2f )]   (2f )e
2 
1 j2ft
df   (2f )e j2ft
d (2f)


1  j 0.t 1
 e 
2 2
1
  (2f )
2
1  ( f )

• The spectrum of a constant signal g(t) = 1 is an impulse

δ(f) = 2πδ(2πf)
• g(t) = 1 is a dc signal that has frequency f = 0(dc)
 Find the inverse Fourier transform of δ(f-f0)
• From sampling property of the impulse function

F [ ( f  f 0)]   ( f f )e df e j2f0t
1 j2ft
0

e j 2f0t  ( f  f )
0

• This shows that the spectrum of an everlasting exponential

ej2πf0t is a single impulse at f = f0
• The spectrum is made up of a single component at
frequency f = f0

e 2 f 0 t   ( f  f 0 )
Find the Fourier transform of the everlasting sinusoid cos 2πf0t
• Using Euler formula
1
c o s 2  f0 t  ( e j 2f 0t  e  j 2f 0t )
2
1
c o s 2 f 0t  [ ( f  f 0 )   ( f  f 0 ) ]
2

• The spectrum of cos 2πf0t consists of two impulses at f0

and –f0 in the f-domain or two impulses ±ω0 = ±2πf0 in
the ω –domain
• An everlasting sinusoid can be synthesized by two
everlasting exponentials ejω0t and ejω0t
• The fourier spectrum consists of only two components of
frequencies ω0 and -ω0
Find the Fourier transform of the sign function sgn(t) –(signum t)
its value is +1 or -1 depending on whether t is positive or
negative
 Some properties of Fourier Transform
• Properties of Fourier transform, implications and
Applications will be studied
• Time-Frequency Duality- similar to photograph and its
negative

• If g(t) <=>G(f)
• Time-shifting property
• Dual of the time-shifting property
• There is role reversal, and we quarantee that any result
will have a dual
 Duality Property
• If the fourier transform of g(t) is G(f) then the
fourier transform of G(t) with f replaced by t is
g(-f)
• g(-f) is the original time domain signal with t
replaced by -f
 Apply duality property to the pair of figures below
• Figure

• G(t) is the same G(f), and g(-f) is the same as g(t) with t
replaced by –f

• Substituting τ = 2πα
 Time-Scaling Property
• If
• For any real constant a

• Proof

• If a < 0

• A time compression of a signal results in a

spectral expansion and vice versa
 Time scaling property
• Compression in time by a factor a means that the signal
is varying more rapidly by the same factor
• To synthesize, frequencies of the signal must be
increased by a (frequency spectrum expanded by a)
• A signal cos 4πf0t is the same as the signal cos 2πf0t time
compressed by a factor of 2
 Reciprocity of signal and its bandwidth
• Time-scaling property implies that if g(t) is
wider, its spectrum is narrower and vice versa
• Doubling the signal duration halves it
bandwidth and vice versa
• Bandwidth of a signal is inversely proportional
to the signal duration or (width in seconds)
• E.g bandwidth of previous figures.
 Example
• Show that
• Use this result and the fact that
 Time-Shifting Property
• Delaying a signal by to second does not change
its amplitude spectrum, however the phase
spectrum is changed by -2πft0
• If

• Proof:

If t-t0 = x
 Time-shifting Property
• Time delay in a signal causes linear phase shift
in its spectrum
• To achieve the same delay, higher frequency
sinusoids must undergo proportionately larger
phase shift
 Example
• Find the Fourier transform

• Delay causes a linear phase spectrum

 Frequency-Shifting Property
•Multiplication of a signal by a factor ejπfot shifts
the spectrum of the signal by f=f0

•If
Then
•Proof:

• Changing f to –f0
• Because ejπfot is not a real function in practise, frequency
shifting in practice is achieved by multiplying g(t) by a
sinusoid.

• Multiplication of g(t) by a sinusoid of frequency f0 shifts

the spectrum G(f) by ±f0.
• Mulitplication of a sinusoid cos 2πf0t by g(t) amounts to
modulating the sinusoid amplitude. (amplitude
modulation)
• cos 2πf0t is the carrier, g(t) is the modulating signal and
cos 2πf0t g(t) is the modulated signal (chapt 4 and 5)
• To sketch a signal g(t) cos 2πf0t

• g(t) cos 2πf0t touches g(t) when the sinusoid cos 2πf0t is
at its peaks and touches when cos 2πf0t is at its negative
peaks
• g(t) and –g(t) therefore acts as envelopes for the signal
g(t) cos 2πf0t
• g(t) and –g(t) are mirror images of each other about the
horizontal axis.
 Shifting the phase spectrum of a
modulated signal
• By using cos (2πf0t+θ) instead of cos 2πf0t
• If signal g(t) is multiplied by cos (2πf0t+θ),

• If
 Example
• Find and sketch the Fourier transform of the modulated
signal g(t)cos2πf0t in which g(t) is a rectangular pulse
Π(t/T) as shown below:
• From pair , G(f)

• Spectrum is shown below

 Application of Modulation
• Modulation is used to shift signal spectra in this
scenarios:
– Interference will occur if signals occupying the same
frequency band are transmitted over the same medium
– Example, if radio stations broadcast audio signals, receiver
will not be able to separate them
– Radio stations could be assigned different carrier
frequency.
– Radio station transmit modulated signal, thus shifting the
signal spectrum to allocated band.
– Both modulation and demodulation utilizes spectra
shifting
– Transmitting several signals simultaneously over a channel
by using different frequency bands is called Frequency
division multiplexing
 Application of Modulation
• Audio freqencies are so low (large wavelength) that it will
require impractical large antennas for radiation.
• Shifting the spectrum to a higher frequency (smaller
wavelength) by modulation will solve the problem
 Bandpass Signal
• If gc(t) and gs(t) are low-pass signal, each with a
bandwidth B Hz or 2πB rad/s
– gc(t)cos2 πf0t and gc(t)sin2 πf0t are both bandpass signals
occupying the same band each with bandwidth 2B Hz.
– Linear combination of both signal is a bandpass signal
occupying the same band as that of either signal (2B Hz)
– General bandpass signal
gbp(t) = gc(t)cos2 πf0t + gc(t)sin2 πf0t
Magnitude spectra of individual signal is symmetrical about
±f0 but not of their sum

• General bandpass gbp(t) could be expressed as:

 Bandpass Signal

• gc(t) and gs(t) are low-pass, E(t) and Ψ(t) are also low-
pass signals
 Example
• Find the fourier transform of a general periodic signal g(t)
of period T0 and hence determine the Fourier transform
of the periodic impulse train δT0(t)
• A periodic signal g(t) can be expressed as an exponential
Fourier series as
• Dn = 1/T0

• Spectrum of the impulse train is an impulse train

(frequency domain)
 Convolution Theorem
• Convolution of two function g(t) and w(t)

• If

• Then time convolution

• Frequency convolution

• Time convolution of two signal in the time domain

becomes multiplication in the frequency domain
• Multiplication of two signal in the time domain becomes
convolution of the signals in the frequency domain
 Convolution Theorem
• Prove

• Using time shifting property

• Bandwidth of the product of two signals

– If g1(t) and g2(t) have bandwidths B1 and B2, bandwidth of g1(t)
g2(t) is B1 + B2
– Bandwidth of g2(t) is 2B Hz and gn(t) is nB Hz
 Example
• Use the time convolution property to show
g(t) G(f)
Then

Knowing
•
Example
Use the time differentiation property to find the fourier
transform of the triangular pulse ∆(t/τ)
• Involves successive differentiation as shown in b and c
• Second derivative consists of sequence of impulses
• Derivative of a signal at jump discontinuity is an impulse
of strength equal to the amount of jump

• Time differentiation property

• Time shifting property
Properties of Fourier Tranform Operations
 Signal transmission through a linear system
• Linear time-invariant (LTI) system can be characterized in
the time domain or frequency domain
• The LTI system model can be used to characterize
communication channels
• A stable LTI system can be characterized in the time
domain by its impulse response h(t)
• Impulse response is the system response to a unit
impulse input
• The system response to a bounded signal x(t) follows:
y(t) = h(t)*x(t)
• Taking fourier transform Y(f) = H(f).X(f) (convolution
theorem
Analysis and Transmission of signal
• Fourier transform of the impulse response
h(t), given as H(f) is called transfer function or
frequency response of LTI system.
• H(f) is complex
H(f) = |H(f)|ejθh(f)
|H(f)| is the amplitude response
Θh(f) is the phase response of the LTI system
 Signal Distortion during Transmission
• During transmission, signal x(t) is changed to signal y(t)
• If X(f) and Y(f) are spectra of the input and output
respectively, H(f) is the spectra response
Y(f) = H(f).X(f)
• Output spectrum is given by the input spectrum
multiplied by the spectral response
• Equation above expressed in polar form

• Amplitude and phase relationship

• During transmission, input signal amplitude |X(f)| is

changed to |X(f)|x|H(f)|
• Input signal phase spectrum θx(f) is changed to θx(f) +
θh(f)
• An input signal is modified by a factor of |H(f)| and
shifted in phase by θh(f)
• Plot of |H(f)| and θh(f) as a function of fshows how
the system is modifies the amplitude and phase of
various sinusoidal input
• H(f) is the frequency response of the system
• During transmission some components are amplified
and some are attenuated.
• Output will be different from input in general
 Distortionless Transmission
• In applications such as message transmission over
communication channel, the output waveform is
required to be a replica of the input waveform
• To achieve this, distortion due to amplification or
communication channel must be minimized
• Distortionless transmission is thus desired
• Transmission is said to be distortionless if the input and
the output have identical wave shapes within a
multiplicative constant.
• A delayed output that retains the input waveform is
distortionless
• Given input x(t) and output y(t), a distortionless
transmission satisfies
 Distortionless Transmission
• Fourier transform of previous equation:

• Since Y(f) = X(f)H(f) then

• From previous equation

• For distortionless transmission, amplitude response

|H(f)| must be a constant and phase response θh(f)
must be linear function of f going through the origin
• The slope of θh(f) with respect to ω = 2πf is –td (delay of
output w.r.t input
 All-Pass vs. Distortionless System
• All-pass has constant gain for all frequencies (|H(f) = k)
without linear phase requirement
• Distortionless system is always and all-pass system but
converse is not true
• Transmitting recorded music signal that contains high
frequency and low frequency component.
• An all-pass signal could cause extra delay on the high
frequency component, which makes the music out of
sync even if the signal components have the same gain
and all components present
• Difference in transmission delay is due to non-linear
phase H(f) in the all-pass filter
• For distortionless system (td(f) have to be constant
 Distortion in Audio and Video Signal
• Human ear can readily perceive amplitude distortion but
relatively insensitive to phase distortion
• To notice phase distortion, variation in delay (i.e in the
slope of θh) should be comparable to signal duration
• In audio, each spoken syllable can be considered an
individual signal.
• Audio systems may have nonlinear phases, yet no
noticeable signal distortion because maximum variation in
the slope of θh is a small fraction of millisecond
• Audio manufacturers only provide |H(f)|
• In video, human eye is sensitive to the phase distortion but
relatively insensitive to amplitude distortion
• In TV – smeared pictures
• In digital communication – pulse dispersion (spreading out)
 Example
• If g(t) and y(t) are the input and the output respectively
of a simple RC low-pass filter. Determine the transfer
function H(f) and sketch |H(f)|,θh(f) and td(f). For
distortionless transmission through this filter, what is the
requirement on the bandwidth of g(t) if amplitude
response variation within 2% and time delay variation
within 5% are tolerable? What is the transmission delay?
Find the output y(t).
• Applying voltage division rule

• Where
 Example
• Hence

• Time delay

• The phase linearity results in a constant time delay

characteristic. The filter can transmit low-frequency
signals with negligible distortion
 Example
• In our case, amplitude response within 2% and 5% is
tolerable
• Let f0 be the highest bandwidth of a signal that can be
transmitted within these specification,
• To compute f0, |H(f)| = 1 and td(0) = 1/a second
• Hence |H(f0)
 Ideal versus Practical filters
• Ideal filters allow distortionless transmission of a certain
band of frequencies and suppress all remaining
frequencies
• Ideal filter shown below allow all components below f=B
Hz to pass without without distortion and suppresses all
components above f = B
• If g(t) is input signal and y(t) is output then

• Ideal high pass and bandpass characteristic

 Ideal versus Practical filters
• Signal g(t) is transmitted without distortion but with
delay td
• For this filter

• Unit impulse response of this filter ( using pair 18)

• For a physically realizable system h(t) must be causal

• In frequency domain
 Ideal versus Practical filters
• The impulse response of previous filter is not realizable.
To make it realizable (causal), the tail could be cut off

• Ideally a delay of td = ∞ is needed for ideal filter

• The half-power bandwidth of a filter is defined as the
bandwidth over which the amplitude response |H(f)|
remains constant within a 3 dB or (ratio of 0.707)
• Half-power bandwidth of a lowpass filter is called the
cutoff frequency
 Digital Filter
• Analog signal can be processed by digital means (A/D
conversion)
• Involves sampling, quantizing and coding
• Resulting digital signal can be processed by computer by
algorithm e.g low-pass, high-pass filter algorithm

• Advantages of digital processing of analog signal

– Computers can be timed shared, cost is usually lower, accuracy
is dependent only on computer word length, quantizing interval
and sampling rate(alias)
Signal Distortion over a Communication Channel
• Nature of signal distortion will be studied.
• Linear Distortion
– Distortion from linear time invariant channel due to non-
ideal characteristics of magnitude distortion, phase
distortion or both
– We can study the effect of the nonidealities on signal g(t)
– Assuming the pulse exist within interval (a, b) and is zero
outside
– The components of the fourier spectrum of the pulse have a
perfect and delicate balance of magnitudes and phases that
adds up precisely to the pulse g(t) over interval [a, b]
– The balanced is left undisturbed if the signal is passed
through a distortionless channel because a distortionless
channel multiplies each component by the same factor and
delays by the same time
Signal Distortion over a Communication Channel
– If the amplitude response of the channel is not ideal (i.e
|H(f)| is not equal to a constant), the balance will be
disturbed, and the sum of all components cannot be zero
outside the interval (a,b) –pulse will spread out.
– Dispersion also occurs if phase characteristic is not ideal
– Linear channel distortion (dispersion in time) is damaging
to digital communication systems
– Leads to intersymbol interference (ISI) – digital symbol
when transmitted over a dispersive channels tends to
spread wider than its allotted time
– This makes adjacent symbols to interfere with one another
thereby increasing the probability of detection error at the
receiver
 Example
• A low-pass filter transfer function H(f) is given by

• A pulse g(t) band-limited to B Hz is applied at the input of this

filter. Find the output y(t).
• This filter has ideal phase and nonideal magnitude
characteristics. Because g(t)  G(f), y(t)  Y(f) and

• Because g(t) is band-limited by to B Hz

 Distortion caused by Channel Nonlinearities
• Considering a memoryless nonlinear channel in which input
g and the output y are related by some memoryless
nonlinear equation
• Expanding the right-hand side of this equation in a
Maclaurin series
• Bandwidth of y(t) is greater than KB Hz hence the output
spectrum spreads well beyond the input spectrum
• The output signal contains new frequency components not
present in the input signal
• In broadcast communication, signals are amplified at high
power levels where high efficiency amplifiers are desirable.
However they (amplifiers) are nonlinear and they cause
distortion. This is a problem in AM signals
• Non-linear distortion does not affect FM signal (chapter 5)
 Distortion caused by channel nonlinearity
• If a signal is transmitted over a nonlinear channel, the
nonlinearity not only distorts the signal, but also causes
interference with other signals in the channel because of
spectral dispersion
• Linear distortion causes interference between signals
between the same channel
• Spectral distortion due to nonlinear distortion causes
interference among signals using different frequency
channels
 Example
• The input x(t) and the output y(t) of a certain nonlinear
channel are related as

• Find the output signal y(t) and its spectrum Y(f) if the
input signal is x(t) = 2000sinc(2000πt). Verify that the
bandwidth of the output signal is twice that of the input
signal. This is due to signal squaring. Can the signal x(t)
be recovered (without distortion) from the output y(t).
 Example
• 0.316.2000sinc2(2000πt) is the unwanted (distortion)
term in the received signal.

• The bandwidth of the received signal is twice that of the

input signal x(t) because of the squaring
• The received signal contains the input signal x(t) plus an
unwanted signal 632 sinc2(2000πt).
• Spectra of the of the desired signal and the distortion
overlap, it is impossible to recover the signal x(t) from
the received signal y(t) without distortion
• Passing through a low pass filter gives (d) – still some
residual distortion
 Distortion Caused by Multipath
• Multipath transmission occurs when transmitted signal
arrives at the receiver by two or more paths of different
delays due to impedance irregularities (mismatching)
• In radio links, the signal can be received by direct path
between the transmitting and receiving antennas and
also by reflections from other objects e.g hills or building

• Transmission can be represented as several channel in

parallel each with relative attenuation and different delay
• The transfer functions of the two paths are given by
 Distortion from Multipath
• The overall transfer function of the channel is H(f)

• Both magnitude and phase characteristics of H(f) are

periodic in f with period of 1/∆t and can cause linear
distortion
• If the gains from two paths are very close, then the
signals received from the two paths may have opposite
phase and will almost cancel each other
• E.g if f = n/(2∆t)(n = odd), cos 2πf ∆t = -1, H(f) = 0 when α = 1.
these frequencies are called multipath null frequencies.
• At frequencies f = n/(2∆t)(n = even), the two signal interfere
constructively to enhance gain.
• This result in frequency selective fading of transmitted signals
• In practise channel characteristic vary over time due to
random changes in the propagation medium.
• E.g reflection properties vary with meteorological conditions
• Effective channel transfer function varies semiperiodically and
randomly causing random attenuation of the signal and this is
called fading
• Slow fading can be reduced by using Automatic gain control
(AGC)
 Fading Channel
• Fading could be strongly frequency dependent
where different frequencies are affected
unequally (Frequency slective fading)
 Signal Energy and Energy Spectral Density
• Energy Eg of a signal g(t) is defined as the area under
|g(t)|2
• We can determine the energy signal from its Fourier
transform G(f) through Parseval’s

theorem
g(t)  

G ( f )e j 2ft dt

• Hence we can determine energy from both the time

domain and the frequency domain
 Example
• Verify the parseval’s theorem for the signal g(t) = e-atu(t)
(a > 0)

• Determining Eg from the signal spectrum

• Using parseval’s theorem

• This verifies the Parseval theorem

 Energy Spectral Density
• The previous example shows that the energy of a signal
g(t) is the result of energies contributed by all the
spectral components of the signal g(t).
• The contribution of a spectral component of frequency f
is proportional to |G(f)|
• Assuming g(t) is input to a bandpass filter whose H(f) is

• Energy spectral density (ESD) Ψg(t) (per unit bandwidth

in Hz) is defined as

• The ESD of the signal g(t) = e-atu(t)

 Essential Bandwidth of a signal
• Spectra of most signal extend to infinity
• However, because energy of most practical signal is finite,
signal spectrum must approach 0 as f -> ∞
• Most of the signal energy is contained within certain band of
B Hz and the energy content beyond B is negligible
• The bandwidth B is called essential bandwidth of the signal
• Criterion for selecting B depends on error tolerance in a
particular application.
• You can select B to contain 95% of the signal energy
• Suppression of all spectral components of g(t) beyond the
essential bandwidth results in which is an approximation
of g(t)
• If 95% criterion is used for essential bandwidth, energy of
the error (g(t) - ) is 5% of Eg
 Example
• Estimate the essential bandwidth W (in rad/s) of the
signal g(t) = e-atu(t) if the essential band is required to
contain 95% of the signal

• Signal energy Eg is the area under ESD which 1/2a. If W

rad/s is the essential bandwidth which contains 95% of
the total energy Eg, 95% of the area in the figure

• or
• In Hz
 Energy of Modulated Signal
• Modulation shifts the signal spectrum G(f) to the left and
right by f0. Similar thing happen in ESD
• Let g(t) be a baseband signal band-limited to B Hz. The
amplitude modulated signal is

• And the spectrum of φ(t) is

• The ESD of the modulated signal φ(t) is |Φ(f)|2 is

• F0 >= B, then G(f+f0) and G(f-f0) are non overlapping

• Energy of modulated Eφ = ½(Eg)

Time autocorrelation Function and Energy Spectral Density

• Autocorrelation is an even function

• Relationship between autocorrelation of a signal and its

ESD. They form a fourier transform pair

• Proof in textbook
 ESD of the input and the Output
• If x(t) and y(t) are the input and the corresponding
output of an LTI system then

• The output signal ESD is |H(f)|2 times the input signal

ESD
 Signal Power and Power Spectral Density (PSD)
• Power is the area under PSD
• PSD is