Chapter 2
2.0 Introduction
• This chapter introduces the different elementary discrete-time signals that are important in the
treatment of signal processing. These are used as basis functions or building blocks to describe
more complex signals.
• This chapter also emphasizes the characterization of discrete-time systems in general and the
class of linear time-invariant (LTI) systems in particular.
• The motivation for studying LTI systems is twofold: first, there are a large collection of
mathematical techniques that can be applied to the analysis of LTI systems; second, many
practical systems are either LTI systems or can be approximated by LTI systems.
Discrete-Time Signal
2
1.5
1
x(n)
0.5
-0.5
-1
-4 -3 -2 -1 0 1 2 3 4 5
n
Note: A discrete-time signal is NOT DEFINED at instants between two successive samples.
• A discrete-time signal is defined for every integer value n for -∞ < n < +∞.
• x(n) refers to the nth sample of the signal.
1, for n = 1,3
xn = 4, for n = 2
0, elsewhere
x (n ) = . . . 0 0 1 4 1 0 0. . .
↑
x (n ) = 0 1 4 1 0 0 . . .
↑
x (n ) = 3 − 1 − 2 5 0 4 − 1
↑
x (n ) = [0 1 4 1]
1 for n = 0
δn =
0 for n ≠ 0
2.1.6
In words, the unit sample sequence is a signal that is zero everywhere, except at n = 0 where its
value is unity. It is also called unit impulse. The graphical representation of δ(n) is shown below.
1.5
0.5
0
-10 -5 0 5 10
1 for n ≥ 0
un =
0 for n < 0
2.1.7
1.5
0.5
0
-10 -5 0 5 10
n for n ≥ 0
rn =
0 for n < 0
2.1.8
15
10
0
-10 -5 0 5 10
xn = a 2.1.9
If a is a real number, then x(n) is also a real number. The figures below illustrate x(n) for various
values of the parameter a.
Figure 2.5. Real exponential signals for various values of the parameter a
a = re
and x(n) as
Figure below shows the real and imaginary plots of the complex exponential signal. Notice that
the real part is a damped cosine wave while the imaginary part is a damped sine wave.
real part
1
0.5
-0.5
0 5 10 15 20 25 30
imaginary part
1
0.5
-0.5
0 5 10 15 20 25 30
Complex exponential signals can also be represented graphically using the amplitude function
The figure below shows the plot of complex exponential function in terms of magnitude and
phase functions.
magnitude
1
0.5
0
0 5 10 15 20 25 30
phase
5
-5
0 5 10 15 20 25 30
Figure 2.7. Graph of amplitude and phase function of a complex-valued exponential signal
E = 0 |xn|1
2.1.15
3 42
3 4N
o If there is no value of N that satisfies the above equation, the signal is considered
aperiodic.
o The energy of a periodic signal x(n) over a single period is finite, if x(n) takes on finite
values over the period. However, for the whole duration of the signal (from negative to
positive infinity) its energy is infinite.
o For the whole duration of the signal, the average power of the periodic signal is finite
and is equal to the average power over a single period, given by
N4:
1
P= 0|xn|1
N
2.1.23
3;
• Example: Determine whether the following signals are energy or power signals and determine
their energy and power.
a) Unit sample sequence
Even
0
-5 0 5
Odd
5
-5
-5 0 5
o Any signal can be expressed as the sum of two signal components, one even and one
odd.
1
xC n = Dxn + x−nE
2
2.1.26
1
xF n = Dxn − x−nE
2
2.1.27
• Example: Resolve the following signals into its odd and even components and plot the resulting
sequence. Verify your answers.
1 + > for − 3 ≤ n ≤ −1
a) Unit step sequence
b) xn = G
1 for 0 ≤ n ≤ 3
0 elsewhere
c) x (n) = L 3 − 2 1 − 1 4 3 2 L
↑
Figure 2.9. Graphical representation of a signal, and its delayed and advanced versions
o Time folding – the signal x(n) is folded about n = 0 when the time variable n is replaced
by –n.
o Time-scaling – the signal x(n) is downsampled when the time variable n is replaced by
an, where a is an integer. It is upsampled when the time variable n is replaced by n/a,
where a is an integer.
o Downsampling means resampling the sampled signals, that is, decreasing the sampling
rate by a. Upsampling means inserting a – 1 samples in between samples.
o The sum of two signals x1(n) and x2(n) is a signal y(n), whose value at any instant is equal
to the sum of the values of these two signals at that instant, that is
n
1+ for − 3 ≤ n ≤ −1
xn = K 3
1 for 0 ≤ n ≤ 3
0 elsewhere
• Example: A discrete-time signal x(n) is shown below. Sketch and label carefully each of the
following signals:
a) x(n – 2)
b) x(4 – n)
c) 3x(n + 2)
d) x(n) u(2 – n)
e) x(n – 1) δ(n – 3)
f) x(n2)
g) even part of x(n)
h) odd part of x(n)
|n|, −3 ≤ N ≤ 3
xn = M
0, OPℎRSTUVR
Determine the output of the system defined by the following input – output relationship.
a) yn = xn
b) yn = xn − 1
c) yn = xn + 1
d) yn = > Dxn + 1 + xn + xn − 1E
:
• An adder – a system that performs the addition of two signals sequences to form another
sequence, which is the sum of the two inputs. The symbol for an adder is illustrated below
• A constant multiplier – a system that multiplies the input signal by a scale factor. The symbol for
the constant multiplier is shown below.
• Signal multiplier – a system that multiplies two input sequences to produce the product
sequence. A graphical representation of the signal multiplier appears below.
• Unit delay element – a system that simply delays the signal passing through it by one sample.
The figure below illustrates such system.
• Unit advance element – a system that moves the input ahead by one sample. A unit advance
system is graphically illustrated as the figure below.
• Example: Sketch the block diagram representation of the discrete-time system described by the
input-output relation
1 1 1
yn = yn − 1 + xn + xn − 1
4 2 2
yn = axn
are both static or memoryless. Note that there is no need to store any of the past inputs
or ouputs in order to compute the present output.
o If the output of the system at any instant is computed using the past or future sample of
the input and the past output, the system is said to be dynamic or to have memory. The
systems described by the following input-output relations
yn = 0 xn − k
Y3;
yn = 0 xn − k
Y3;
The system is said to be time-invariant if y(n,k) = y(n – k); otherwise it is said to be time
varying.
yn = nxn
is time-varying system.
b) yn = nxn
c) yn = x−n
d) yn = xn cos ω; n
for any arbitrary input sequences x1(n) and x2(n) and any arbitrary constants a1 and a2.
o Linear systems exhibit multiplicative or scaling property and additive property. This is
the consequence of the definition of the superposition principle (Eq. 2.2.26)
o The linearity condition stated by Eq. 2.2.26 can also be extended to any weighted linear
combination of signals.
• Example: Determine if the following systems described by the following input-output equations
a) yn = nxn
are linear or nonlinear.
b) yn = xn1
c) yn = x 1 n
d) yn = Axn + B
e) yn = e_
yn = y 1 n − 1 + xn
• We now turn our attention to the analysis of the important class of linear, time-invariant (LTI)
systems.
• In particular, we shall demonstrate that such systems are characterized in the time domain
simply by their response to a unit sample sequence.
• We shall also demonstrate that any arbitrary input signal can be decomposed and represented
as a weighted sum of unit sample sequences.
• The general form of the expression that relates the unit sample response of the system and the
arbitrary input signal to the output signal, called the convolution sum or convolution formula, is
also derived.
• Two basic methods for analyzing the behavior or response of a linear system to a given input
signal.
o Direct solution of the input-output relationship of the system, whose general form (for
the LTI discrete-time systems) is
N M
Y3: Y3;
called the difference equation and represents one way to characterize the behavior of a
discrete-time LTI systems.
o The second method of analyzing the behavior of a linear system to a given input signal is
first decompose or resolve the input signal into a sum of elementary signals. The
elementary signals are selected so that the response of the system to each signal
component is easily determined. Then, using the linearity property of the system, the
responses of the system to the elementary signals are added to obtain the total
response of the system to the given input signal.
o The first method is discussed in the next section, the second method is discussed in this
section.
• The choice of the elementary signals appears to be arbitrary, as long as the response can be
determined conveniently.
• Resolution of the input signal to a weighted sum of unit sample (impulse) sequence proves to be
mathematically convenient and completely general solution to the response of the system. But
if the input signal is periodic with period N, it can be more mathematically convenient for us to
resolve these signals into harmonically related exponentials
• Suppose we have an arbitrary signal x(n) that we wish to resolve into a sum of unit sample
sequence. We select the elementary signals xk(n) to be
Note that the signal δ(n – k) is zero everywhere except at n = k, where its value is unity.
• If we multiply the input signal x(n) with δ(n – k), the result of this multiplication is another
sequence that is zero everywhere except at n = k, where its value is x(k). Thus if we repeat this
process at all possible values of k, the equation below holds true:
2
Y3 42
• Example: Resolve the sequence x (n) = 2 4 0 3 into a sum of weighted impulse
↑
sequence.
• We denote the response of the system to the input unit sample sequence as h(n) and is called
the impulse response of the system. A relaxed LTI discrete-time system is characterized, in time-
domain, by its impulse response.
• The response of the LTI system as a function of the input signal x(n) and its impulse response
h(n) is given as
2
yn = 0 xnhn − k
2.3.17
Y3 42
Y3 42
h(n) = 1 2 1 − 1
↑
x (n ) = 1 2 3 1
• Example: Determine the output y(n) of a relaxed linear time-invariant system with impulse
response
xn = un
and
• A causal system is one whose output at time n depends only on present and past inputs but
does not depend on future inputs. For LTI systems, the causality condition also puts a restriction
on the impulse response of the system.
• A causal system has a causal impulse response, that is h(n) = 0 for n < 0. It is a necessary and
sufficient condition for causality.
• The convolution sum formula can now be modified to reflect this condition. Thus
• If the input signal x(n) is also causal, that is x(n) = 0 for n < 0, the above formula can be further
simplified, that is
• An arbitrary relaxed system is BIBO stable if and only if its output sequence y(n) is bounded for
every bounded x(n).
• In terms of its impulse response h(n), an LTI system is stable if and only if its impulse response is
absolutely summable, that is, it decays over time, or
• This implies that any excitation at the input of the system, which is of finite duration, produces
an output that is “transient” in nature, that is its amplitude decays with time and dies out
eventually, when the system is stable.
• Example: Determine the range of values of the parameter a for which the LTI system with
impulse response
h(n) = an u(n)
is stable.
• Example: Determine the range of values of a and b for which the LTI system whose impulse
response
a , n ≥ 0
hn = M
b , n<0
is stable.
• With the knowledge on the impulse response as the time-domain characterization of linear,
time-invariant systems, a classification on this subclass of systems can now be discussed
• LTI systems can be further subdivided into class of system depending on the duration of their
impulse response:
o Finite impulse response (FIR) systems whose impulse response is of finite length.
o Infinite impulse response (IIR) systems whose impulse response has infinite duration.
• For causal, finite impulse response system, it can be said that its impulse response is zero for
some finite time interval, that is
yn = 0 hkxn − k
Y3;
• This means that the output at any time n is the weighted linear combination of the input signal
samples from x(0) to x(n – M + 1). In effect, this system acts as a “window” that views only the
most recent M input signals in forming the output. It neglects or “forgets” all prior input
samples. Thus we can say that FIR systems have finite memory.
• IIR systems, on the other hand have infinite memory. Its output, through the convolution
2
formula is
yn = 0 hkxn − k
Y3;
• We have treated linear and time-invariant systems that are characterized by their unit sample
response h(n). In turn, h(n) allows us to determine the output y(n) of the system for any given
2
input sequence x(n) by means of convolution summation
yn = 0 hkxn − k
2.4.1
Y342
• This convolution formula in Eq. 2.4.1 suggests a means for the realization of the system. In case
of FIR systems, such a realization involves additions, multiplications and a finite number of
memory locations. Thus, FIR systems are readily implemented directly by convolution
summation.
• If the system is IIR, however it is impractical to implement it through the convolution
summation, since it requires an infinite number of memory locations, multiplications, and
additions.
• Thus IIR systems are more practically realized through the difference equation, which gives
computationally more efficient way of implementing IIR systems. This section describes the
method of analyzing systems using the difference equation.
• Recursive systems is a system whose output y(n) at time n depends on any number of past
output values y(n-1), y(n-2), . . . etc. If the output y(n) does not depend on the previous outputs,
the system is called nonrecursive.
• Recursive algorithm oftentimes provide more computationally efficient means of implementing
algorithms.
• Example: Analyze the system which computes the cumulative average of a signal x(n) in the
interval 0 ≤ k ≤ n.
2
1
yn = 0 xk
n+1
Y3;
1 xn
yn = dyn − 1 + e
2 yn − 1
• The difference between recursive and nonrecursive systems is illustrated in the diagram below.
Note the presence of the feedback loop and the delay element for recursive system.
• The first part of the response, which contains y(-1) is the result of the initial condition y(-1) of
the system. The second part of the response is the response of the system to the input signal
x(n).
• If the system is initially relaxed, i.e. y(-1) = 0, its corresponding output is called its zero-state
response or forced response. It is denoted by yzs(n).
• If the initial conditions of the system is nonzero, and the input x(n) is zero for all values of n, the
system still has output, and is called the zero-input response or natural response. It is denoted
by yzi(n).
N M
general form for such an equation is
The part yh(n) is called the homogenous or complementary solution, whereas yp(n) is called the
particular solution.
0 aY yn − k = 0
2.4.14
Y3;
yg n = m 2.4.15
N
o Substitute this to Eq. 2.4.14 to form the polynomial equation
0 aY m4Y = 0
Y3;
or
complex-valued. Since the coefficients a1, a2, …, aN are real, complex valued roots occur
are complex-conjugate pair. Some of the N roots may be identical, in which case we
have multiple order roots.
o If the roots are distinct, the most general solution to the homogenous difference
equation in Eq. 2.4.14 is
o These coefficients are determined from the initial conditions specified for the system.
o Since the input x(n) = 0, Eq. 2.4.17 can be used to obtain the zero-input response of the
system.
• Example: Determine the homogenous solution of the system described by the first order
• Example: Determine the zero-input response of the system described by the homogenous
second-order difference equation
• Example: Determine the homogenous solution and the zero-input response of the system
described by the difference equation
• Example: Determine the homogenous solution and the zero-input response of the system
described by the difference equation
o The resultant y(n) contains the constant parameters embodied in the homogenous
solution. These constants can be determined to satisfy the initial conditions.
when x(n) is a unit step sequence and y(-1) is the initial condition.
to the input (a) x(n) = 4u(n) and (b) x(n) = 2nu(n). The initial conditions of this system are y(-1) =
y(-2) = 1. Also determine the zero-state and zero-input response of this system.
to the input x(n) = 4nu(n). Also determine the zero-state and zero-input response of this system.
• The impulse response of the system h(n) was previously defined as the response of the system
to a unit sample excitation.
• In the case of a recursive system, h(n) is simply equal to the zero-state response of the system
when the input x(n) = δ(n) and the system is initially relaxed.
• We have established the fact that the total response of the system to any excitation function
consists of the sum of the two solutions of the difference equation: the homogenous and the
particular solution to the excitation function.
• In the case where the excitation is an impulse, the particular solution is zero, since x(n) = 0 for n
> 0.
• The response of the system to an impulse consists only of the solution to the homogenous
equation, with the constant parameters evaluated to satisfy the initial conditions dictated by the
impulse.
• Example: Determine the impulse response h(n) for the system described by the second order
Reference:
John G. Proakis, Dimitris G. Manolakis. Digital Signal Processing: Principles, Algorithms, and
Applications: Third Edition. Prentice – Hall, New Jersey. 1996