Chapter 2: Discrete Time

Systems Convolution

Sem. II, 2010 E.C Electrical and Computer Engineering By: Waltengus
2.1 Discrete Time Signals

 A discrete-time signal is an indexed sequence of real or

complex numbers.
 A discrete-time signal is a function of an integer-valued
variable, n, that is denoted by x(n).
 Although the independent variable n need not necessarily

Fig. 2-1. The graphical representation of a discrete-time signal x(n)

 Most information-bearing signals of practical interest are complicated
functions of time, there are three simple, yet important, discrete-time
signals that are frequently used in the representation and description of
more complicated signals.
 These are Unit sample, Unit step and Exponential.
Unit sample, denoted by δ(n), is defined by
2.1.1 Signal Duration
 Discrete-time signals may be conveniently classified in terms of their
duration or extent.
 For example, a discrete- time sequence is said to be a finite-length
sequence if it is equal to zero for all values of n outside a finite interval
[N1, N2].
 Signals that are not finite in length, such as the unit step and the
complex exponential, are said to be infinite-length sequences.
 Infinite-length sequences may further be classified as either being right-
sided, left-sided, or two-sided.
 A right-sided sequence is any infinite-length sequence that is equal to
zero for all values of n < no for some integer no. The unit step is an example
of a right-sided sequence.
 Similarly, an infinite-length sequence x(n) is said to be left-sided if, for
some integer no, x(n) = 0 for all n > no. An example of a left-sided sequence
is

Which is a time-reversed and delayed unit step. An infinite-length signal

that is neither right-sided nor left-sided, such as the complex
exponential, is referred to as a two sided sequence.
2.1.2 Periodic and Aperiodic Sequences
 A discrete-time signal may always be classified as either being
periodic or aperiodic.
 A signal x(n) is said to be periodic if, for some positive real
integer N,

 For all n. This is equivalent to saying that the sequence repeats itself
every N samples.
 If a signal is periodic with period N, it is also periodic with period 2N,
period 3N, and all other integer multiples of N.
 The fundamental period, which we will denote by N, is the smallest
positive integer for which Eq. above is satisfied. If Eq. above is not
satisfied for any integer N, x(n) is said to be an aperiodic signal.
Example 2.1 The signals

Where gcd(N1, N2) means the greatest common divisor of N1 and N2.
 The same is true for the product; that is,

will be periodic with a period N given by Eq. (above). However, the

fundamental period may be smaller. Given any sequence x(n), a periodic
signal may always be formed by replicating x(n) as follows:

Where N is a positive integer. In this case, y(n) will be periodic with period N.

2.1.3 Symmetric Sequences

A discrete-time signal will often possess some form of symmetry that may
be exploited in solving problems. Two symmetries of interest are as follows:
 A complex signal is said to be conjugate symmetric if, for all n

 Any complex signal may always be decomposed into a sum of a

conjugate symmetric signal and a conjugate anti symmetric signal.

2.1.4 Signal Manipulations

 Discrete-time signals and systems we will be concerned with the
manipulation of signals. These manipulations are generally compositions
of a few basic signal transformations.
 These transformations may be classified either as those that are
transformations of the independent variable n or those that are
transformations of the amplitude of x(n) (i.e., the dependent variable).
 The most common transformations include shifting, reversal, and
scaling, which are defined below.

Shifting : This is the transformation defined by f(n) = n - no. If y(n) =

x(n - no), x(n) is shifted to the right by no samples if no is positive
(this is referred to as a delay), and it is shifted to the left by no
samples if no is negative (referred to as an advance).

Reversal: This transformation is given by f(n) = -n and simply

involves "flipping" the signal x(n) with respect to the index n.
Time Scaling : This transformation is defined by f(n) = Mn or f(n) = n/ N
where M and N are positive integers.
 In the case of f(n) = Mn, the sequence x(Mn) is formed by taking every
Mth sample of x(n) (this operation is known as down-sampling. With
f(n) = n/N the sequence y(n) =x(f(n)) is defined as follows:

This operation is known as up-sampling.

2.2 Discrete Time Systems
 A discrete-time system is a mathematical operator or mapping that
transforms one signal (the input) into another signal (the output) by
means of a fixed set of rules or operations.
 The notation T[.] is used to represent a general system as shown in Fig.
2.2, in which an input signal x(n) is transformed into an output signal
y(n) through the transformation T[.].
 The input-output properties of a system may be specified in any one of
a number of different ways.
 The relationship between the input and output, for example, may be
expressed in terms of a concise mathematical rule or function such as
Fig. 2.2. The representation of a discrete-time system as a trans- formation
T[.] that maps an input signal x(n) into an output signal y(n).
 Discrete-time systems may be classified in terms of the
properties that they possess.
 The most common properties of interest include
memoryless, linearity, shift-invariance, causality, stability,
and invertibility.

2.2.1 Memoryless Systems

 A system is said to be memoryless if the output at any time
n = no depends only on the input at time n = no.
 In other words, a system is memory less if, for any no, we
are able to determine the value of y(no) given only the value
of x(no).
Example 2.2 The system is memoryless because y(no) depends only on the
value of x(n) at time no.

The system on the other hand, is not memoryless because the output at time
no depends on the value of the input both at time no and at time no - 1.
y(n) = x(n) + x(n - I)

2.2.2 Linear Systems

A system that is both additive and homogeneous is said to be linear.
Thus, A system is said to be linear if
Linearity greatly simplifies the evaluation of the response of a system to a
given input.
2.2.3 Time (Shift) Invariance Systems
 If a system has the property that a shift (delay) in the input by no
results in a shift in the output by no,
 Let y(n) be the response of a system to an arbitrary input x(n).
 The system is said to be time-invariant if, for any delay no, the
response to x(n- no) produces the out put value y(n - no).
 A system that is not shift-invariant is said to be shift varying.
A system will be shift-invariant if its properties or
characteristics do not change with time.
 To test for shift-invariance one needs to compare y(n - no) to
T[x(n - no)].
 If they are the same for any input x(n) and for all shifts no, the
system is shift-invariant.
Example 2.3 The system defined by
2.2.4 Linear time-Invariant Systems
 A system that is both linear and shift-invariant is referred to as a
linear shift-invariant (LSI) system.
 If h(n) is the response of an LSI system to the unit sample δ(n), its
response to δ(n - k) will be h(n - k).

The above Equation which is known as the convolution sum, is written as

 Where * indicates the convolution operator. The sequence h(n),
referred to as the unit sample response, provides a complete
characterization of an LSI system.

2.2.5 Causality
 A system property that is important for real-time applications is
causality.
 A system is said to be causal if, for any no, the response of the system at
time no depends only on the input up to time n = no.
 For a causal system, changes in the output cannot precede changes in
the input. Thus, if x1 (n) = x2(n) for n  no, y1(n) must be equal to y2(n) for
n  n o.
 Causal systems are therefore referred to as non anticipatory. An LSI
system will be causal if and only if h(n) is equal to zero for n < 0.
Example 2.4 The system described by the equation y (n) = x(n) + x(n - 1) is
causal because the value of the output at any time n = no depends only on
the input x(n) at time no and at time no - 1.
The system described by y(n) = x(n) + x(n+ 1), on the other hand, is non
causal because the output at time n = no depends on the value of the input at
time no + 1.

4.2.6 Stability
 In many applications, it is important for a system to have a response, y(n),
that is bounded in amplitude whenever the input is bounded.
 A system with this property is said to be stable in the bounded input-
bounded output (BIBO) sense. Specifically.
2.2.7 lnvertibility
A system property that is important in applications such as channel
equalization and deconvolution is invertibility.
 A system is said to be invertible if the input to the system may be
uniquely determined from the output.
 In order for a system to be invertible, it is necessary for distinct
inputs to produce distinct outputs.
 In other words, given any two inputs x1(n) and x2(n) with x1(n) ≠ x2(n),
it must be true that y1(n) ≠ y2(n).

Example 2.6 The system defined by

y(n) = x(n)g(n)
is invertible if and only if g(n) ≠ 0 for all n. In particular, given y(n)
with g(n) nonzero for all n, x(n) may be recovered from y(n) as follows:
 2.2.8 Convolution
The relationship between the input to a linear shift-invariant system, x(n),
and the output, y(n), is given by the convolution sum

 Convolution is fundamental to the analysis and description of LSI systems.

Convolution Properties
Convolution is a linear operator and, therefore, has a number of important
properties including the commutative, associative, and distributive
properties.
Commutative Property
 The commutative property states that the order in which two sequences
are convolved is not important.
 Mathematically, the commutative property is

 This property states that a system with a unit sample response h(n)
and input x(n) behaves in exactly the same way as a system with unit
sample response x(n) and an input h(n).
Associative Property
 The convolution operator satisfies the associative property, which is

 The associative property states that if two systems with unit sample
responses h1(n) and h2(n) are connected in cascade as shown in Fig.
below (b).
 An equivalent system is one that has a unit sample response equal to
the convolution of h1 (n) and h2(n):
Fig. 2.5. The interpretation of convolution properties from a systems point of view.
Distributive Property
The distributive property of the convolution operator states that

 This property asserts that if two systems with unit sample responses
h1(n) and h2(n) are connected in parallel, as illustrated in Fig. 2.5(c).
 An equivalent system is one that has a unit sample response equal to
the sum of h1(n) and h2(n):
Example 2.6. Let us perform the convolution of the two signals

Example 2.7 Consider the convolution of the sequence

Examples