Mehran University of Engineering and

Technology, Jamshoro

Signals and
FOURIER SERIES Systems (S&S)
- Lecture#13-18
 TENTATIVE TEACHING PLAN
S. No. Topic No. of Lectures
1 Continuous-Time signals: Classification of signals, basic operations on signals, elementary signals, and 3
signal representation and models.
2 Time Domain Analysis of Elementary Signals: Sinusoidal and complex exponential signals, singularity 3
function signals, signal energy and signal power. +Tutorial#01
3 Continuous-Time systems: System characteristics and classification, linear time-invariant (LTI) systems, 6
impulse response of LTI systems, convolution and its properties, characterization of systems using
differential equations, zero-state response and zero-input responses, and properties of LTI systems.
+Tutorial#02+Class Test#01
4 Fourier Series: Orthonormal signal, signal representation by generalized Fourier series and frequency 6
domain analysis, properties of Fourier series, periodic input systems and its transfer functions.
5 Laplace Domain Representation: Laplace transform (Unilateral and Bilateral), properties of Laplace 6
transform, inverse Laplace transform, LTI system representation in Laplace domain using transfer
functions, pole-zero plot and system stability. +Tutorial#3
6 Frequency Domain Representation and Analysis: Transfer function, Fourier transform, properties of Fourier 9
transform, system frequency response, energy and power density spectrum, auto-correlation function,
phase and group delay. +Tutorial#04+Class Test#02
7 Continuous-Time Filters: Distortion less transmission, filtering, ideal and practical filters, filter 6
classification using transfer function (LP, HP, BP and BS), passive filter circuits, frequency transformation,
and analog filter design (Butterworth, Chebyshev I and II and Elliptic). +Tutorial#05
8 Sampling: Sampling theorem, Ideal and practical sampling effects, time and frequency domain 2
representation of sampling theorem.
9 Discrete-Time Systems: DT Signals, DT signals as vector, linear transformation of these vectors, DT signals 4
as difference equation, and unit pulse response of DT signals. +Tutorial#06+Class Test#03
  Total Number of Lectures 45 2
 TOPICS TO BE COVERED

 Basics
 Motivation of Fourier Series
 Fourier Series

 Coefficients of Fourier Series

 Basis Signal of Fourier Series

 Generalized Fourier Series

 Standard Fourier Series

3
Basics

4
 MOTIVATION OF FOURIER SERIES

  Convolution is derived by decomposing the signal into the

sum of a series of delta functions

5
 MOTIVATION OF FOURIER SERIES

 Can we decompose the signal into the sum of other functions

such that the calculation can be simplified?
 Yes.
 We can decompose periodic signal as the sum of a sequence of
complex exponential signals or basis signals. Fourier Series

 This allows us to decompose the signal into its component-signals to

study the individual characteristics of the signal.

6
   Why complex exponential signal? (what makes complex
exponential signal so special?)

 Complex exponential function has a one-to-one relationship with

sinusoidal functions

 Each complex exponential signal has a unique frequency - frequency

decomposition
 Complex exponential signals are periodic

In a one-to-one relationship, one record in a table is associated with oneand only one record in another table. For example, in a
7
school database, each student has only onestudent ID, and each student ID is assigned to only one person.
 GENERALIZED FOURIER SERIES

Superposition holds for the response of a linear system to an input signal. This permits
us to find the system response to an input signal by decomposing the input signal into a
Sum of signal components, and adding the component responses.

Even if we are not concerned about system response, we may want to decompose a signal
Into a sum of signal components to show individual important characteristics of a signal
Like average value and slope of a signal.

8
For more complicated signals, we can not easily decompose a signal into sum
of its Components.

However, we may be able to approximate the signal, over a time interval by a

linear combination of set of signals, that is by a series.

9
 GENERALIZED FOURIER SERIES

  For any complex signal we find the approximation of the

signal by a formula over a time interval t 1 < t < t 2 , by linear
combination of set of signals. That is:
N
xˆ (t )  X
n 1
n (t )
n

 Where (coefficients of Fourier series) are given as,

 - Basis Signal

10
  1. The very first step is to select basis signals to construct the
approximation of the signal over the approximation interval .
 We can choose these signals to highlight specific signal
characteristics or to analyze simple components for use in system
analysis.

 2. The second step is to choose the weighting constant to

make approximate as closely as possible, over the
xˆ (t )
approximation interval.

 The choice of the basis signals and corresponding weighting

constants should generate a closer approximation as more
terms are added to the sum.

11
 APPROXIMATION CRITERIA

 There are different approximation criteria used to measure the

closeness of ˆ (t )x(t).
xto
 Some of them are:

1. min  max xˆ (t )  x(t )  Error signal



t2


2. min   x (t )  x (t ) dt 
ˆ Area under error signal

 t1 

3. 
2
t

Area under square of the
min   x
2
ˆ (t )  x (t ) dt 

 t1 
 error signal
 APPROXIMATION CRITERIA

 Equation 1 : minimize the maximum

 Equation 2 : minimize the area under the error or produce the

smallest area under the magnitude.

 Equation 3 : minimize the energy of the error signal over the

approximation interval
 It has the advantage of weighting larger errors more heavily and is
easier to use.
 WHAT CRITERION DO WE USE?

  We will use equation 3 as the criterion of approximation.

 We will choose the ’s for a given set of basis signals to

minimize the integral square error, given by:
t2

 xˆ (t )  x (t )  dt
2
N 
t1

t2 2
 N

N  t  
n 1
X 
n n (t )  x (t )  dt

1

 Keep in mind, xˆ (t ) does not have to approximate x(t) outside the

interval .
Coefficients of Fourier
Series

15
 FINALITY OF COEFFICIENTS

 If the coefficient of the first term changes when the second

term is added to the approximation and this happens because
of the basis signals that are selected.

 To avoid recalculation, we need to select such basis signals

so that coefficients do not change as more terms are added
to the approximation.

 This fixed nature of coefficients of basis signals is known as

the ‘finality of the coefficients’.
N
xˆ (t )  X
n 1
n (t )
n
EXAMPLE 3.13 (CARLSON)

2
 𝑥 ( 𝑡 )=𝑡 +1

 ∅ 1 ( 𝑡 ) =1

 ∅ 2 ( 𝑡 ) =𝑡

17
18
SOLUTION – FOR N=1

19
FOR N=2

20
21
  The coefficient X1 changes from to

 This doesn’t demonstrate finality of coefficients

 Need to re-compute the already computed coefficients

 Orthogonal basis signals can assure finality of coefficients

22
EXAMPLE 3.14 (REPEAT EXAMPLE 3.13
FOR THE BELOW GIVEN BASIS SIGNALS)

 ∅ 1 ( 𝑡 ) =1

  2 ( 𝑡 ) =𝑡 −1
∅

23
FOR N=1

24
 FOR N=2

 E 2 = 8/45

25
 HOW MANY FOURIER SERIES
COEFFICIENTS ARE SUFFICIENT?
  The number of Fourier series coefficients depends on the
accuracy that we want to achieve.
 Typically, the number N is chosen such that the residue of the
approximation , for some target error level .

26
Basis Signals of Fourier
Series

27
 MUTUAL ORTHOGONAL BASIS SIGNALS

 In previous example 3.14, one set of the basis signals

produce finality of coefficients and one did not (in example
3.13).

 The property of a set of basis signals that produces the

finality of coefficients is called ‘mutual orthogonality’ of the
basis signals over the approximation interval.

 The general definition of mutually orthogonal basis signals is

given on next slide.
  “The real basis signals, φ i (t) are mutually orthogonal over the
time interval if and only if:”

n n  m
t2

t n (t )m (t )dt   0 n  m
1
 Note that λ n >0 since λ n is the integral of [φ n (t)] 2 which is +ve.

Verify orthogonality property for basis signals

of example 3.13 and 3.14.
Generalized Fourier Series

30
 FOURIER SERIES CAN BE DEFINED AS

  “A generalized Fourier series is a weighted sum of orthogonal

basis signals that approximates a signal over the interval by
minimizing the integral square error, over the interval.”

 To find the generalized Fourier Series,

 we first select a set of basis signals that are orthogonal on the
expansion interval.
 Values of the generalized Fourier Series Coefficients are then found
to minimize ε N .

31
FORMULAS FOR X N AND E N
Example 3.15 (Carlson)…………Homework
 PARSEVAL’S THEOREM

  An infinite basis signal set is called complete if ε N  0

as N  infinity for all signals for which the series exists.
 When ε N  0, it means the integral square error energy approaches
zero, but that does not mean that approaches for all or any t as
Ninfinity.
 Using the equation for ε N
 PARSEVAL’S THEOREM

  The L.H.S. is the signal energy present within the expansion

interval while the R.H.S. is the energy within the expansion
interval contributed by the term of the generalized Fourier
series.
 This same equation shows that the signal energy in the
expansion interval equals the sum of the energies contained
in each of the individual series terms.

 This result is called Parseval’s Theorem.

36
Standard Fourier Series

37
 STANDARD FOURIER SERIES

  Fourier Series: A periodic signal is represented as a weighted

sum of an infinite number of mutually orthogonal basis
signals

 In this case, the basis signals are periodic signals (complex

exponentials/ sine,cosine) that continue from –∞ to +∞ in
time

Exponential
Fourier Series

38
 BASIS SIGNALS

  The basis signals are mutually orthogonal

n nm
t2

t n (t )m (t )dt   0 nm

1

39
 COEFFICIENTS

  These are complex numbers associated with each of the basis

signals as weights

 If x(t) is real, X n is also real and is the average value of the

signal x(t) over the approximation interval.
 The weighted sum of periodic basis signals yield a periodic
signal. Therefore, approximation of any signal from to
continues from in time.
 While approximating a periodic signal, only one complete
cycle of the signal is approximated

40
TRIGONOMETRIC FOURIER SERIES

ao=

41
 EXAMPLE 5.2 (CARLSON)

  Find the complex exponential Fourier Series expansion of over

the interval . Plot x(t) and the truncated series for N=4, i.e.
ˆ4 (t ) series coefficients
 Solution: The complex exponential xFourier
are:

t1 T1
1
 Where T1=20 and
Xn 
T1
f1=1/20.
Therefore:
t1
x(t )e  j 2 nf1t dt

15 15
1 1  10    t 
  
20 5
 t /10  j nt /10
Xn  e e dt   exp   1  j n
20  1  j n    10   5
0.5
Xn  (1.649e j n /2  0.223e  j 3 /2 )
(1  j n)
42
The complex exponential Fourier series becomes:

0.5
xˆ (t )  
n  (1  j n)
(1.649e j n /2
 0.223e  j 3 /2 j nt /10
)e

After evaluating the Fourier series coefficients for

0≤n≤4, we get:
xˆ4 (t )  (0.057e j1.491 )e  j 4 t /10  (0.075e j 3.036 )e  j 3 t /10  (0.112e  j1.729 )e  j 2 t /10
(0.216e j 0.308 )e  j t /10  0.713  (0.216e j 0.308 )e j t /10  (0.112e j1.729 )e j 2 t /10
(0.075e j 3.036 )e j 3 t /10  (0.057e  j1.491 )e j 4 t /10

The plot is shown in next figure.

43
Figure
44
 In figure, it can be seen that the truncated Fourier series
x 4 (t) provides a reasonable approximation to x(t) over most
of the expansion interval.

 The effect of ripples in the interval is known as the Gibbs

Phenomenon.

 Example 5.4: Homework

45
 FREQUENCY DOMAIN REPRESENTATION
OF CONTINUOUS-TIME SIGNALS
  Previously signals have been represented as functions of
time, that portray signal characteristics by showing variation
of signal as function of time (time domain representation).

 Signals characteristics can also be specified by representing

signals as a function of frequency.

 Consider a signal .
 This signal is completely characterized by its amplitude=2,
frequency = 1Hz, and phase π/4.

46
 These quantities portray the signal amplitude and phase as a
function of frequency.

 These functions of frequency are referred to as frequency

domain representation of signal v(t) or simply the Spectrum of
the signal.

 Signal spectrum is ‘signal amplitude and signal phase

represented as a function of frequency’.

 These two functions are called the amplitude spectrum and

phase spectrum.

47
 Amplitude
  spectrum A(f) : gives width of frequency interval
contained in a signal known as signal bandwidth.

 Phase spectrum Ph(f): gives time shift or time delay

information about signal.

 To define amplitude and phase spectra consider the cosine

signal

 Amplitude Ax>0
 Frequency fx>0.
 Plots of amplitude and phase as function of frequency are
spectrum of cosine signal.

48
 Spectrum of cosine is line spectrum or discrete spectrum.
Since discrete frequency functions are plotted as vertical
lines.

 Single-sided spectrum: plotted only for positive frequencies.

 Double-sided spectrum: plotted for +ve and –ve frequencies.

(double sided amplitude and phase spectra are even and odd
functions of frequency respectively).

Example: x(t) = 3 cos ( 2pi(10) t - pi/3 ).

Plot amplitude and phase spectrum of x(t).
49
  For Double-sided spectrum:

Exercise Problems 5.1 and 5.2………Homework

50
END

51