CHAPTER TWO

DISCRETE-TIME SIGNALS AND SYSTEMS

This chapter presents the basic definitions, establish notation, develop and review the basic
concepts associated with discrete-time signals and systems. The major emphasis in this chapter is
the characterization of discrete time systems in general and the class of linear time invariant
(LTI) systems in particular.

A number of important time-domain properties of LTI systems are defined and developed, and
an important formula, called the convolution formula is derived which allows us to determine the
output of an LTI system to any given arbitrary input signal. In addition to the convolution
formula, difference equations are introduced as an alternative method for describing the input-
output relationship of an LTI system. Finally, we review Fourier transform concept.

2.1 Discrete time signals

Discrete-time signals are data sequences. They are represented mathematically as sequences of
numbers. A sequence of numbers x, in which the nth number in the sequence is denoted x[n], is
formally written as
x= { x [n] } , −∞ <n< ∞ (2.1)
where n is an integer. In a practical setting, such sequences can often arise from periodic
sampling of an analog (or continuous-time) signal x a (t). Although sequences do not always arise
from sampling analog waveforms, it is convenient to refer to x [n] as the “nth sample” of the
sequence.
There are several representations of discrete-time signals (i.e., sequences). One method is the
graphical representation as shown in Figure 2.1. Although the abscissa is drawn as a continuous
line, it is important to recognize that x [ n ] is defined only for integer values of n. It is not correct
to think of x [n] as being zero when n is not an integer; x [n] is simply undefined for non-integer
values of n .

Figure 2.1 Graphical representation of a discrete time signal.

By Niguse B. Page 1
Besides the graphical representation of a discrete-time signal or sequence as illustrated in Fig.
2.1, there are some alternative representations that are often more convenient to use. These are:
1. Functional representation, such as

{
1 , for n=1 ,3
x ( n )= 4 , for n=2
0 , elsewhere

The time origin for a sequence x [ n ], which is zero for n< 0, is understood to be the first
(leftmost) point in the sequence.

2.2.1 Some elementary discrete-time signals

There are some sequences that we shall encounter frequently. They are described here.
1. Unit sample sequence
Unit sample sequence defined in figure 2.2 (a) is the sequence defined as

δ [n ]= {10,,n=0
n ≠0
(2.3)

The unit sample sequence plays the same role for discrete-time signals and systems that the
unit impulse function (Dirac delta function) does for continuous-time signals and systems.
For convenience, we often refer to the unit sample sequence as a discrete-time impulse or
simply as an impulse. It is important to note that a discrete-time impulse does not suffer from
the mathematic complications as the continuous-time impulse; its definition in Eq. (2.3) is
simple and precise.

By Niguse B. Page 2
 Figure 2.2 some basic sequences.
One of the important aspects of the impulse sequence is that an arbitrary sequence can be
represented as a sum of scaled, delayed impulses. For example, the sequence p[n] in Figure
2.3 below can be expressed as

Figure 2.3 Example of a sequence to be represented as a sum of scaled, delayed impulses

By Niguse B. Page 3
2. The unit step sequence
The unit step sequence is one that has amplitude of zero for negative indices and amplitude
of one for non-negative indices. The unit step sequence is denoted as u[n] and
mathematically written as

u [ n ]= {10 ,,forfor nn<0≥ 0 (2.4)

The unit step is related to the unit impulse by

n
u [ n ]= ∑ δ [k ] (2.5)
k=−∞

that is, the value of the unit step sequence at (time) index n is equal to the accumulated sum
of the value at index n and all previous values of the impulse sequence. An alternative
representation of the unit step in terms of the impulse is obtained by interpreting the unit step
in terms of a sum of delayed impulses, as in this case, the nonzero values are all unity, so

As yet another alternative, the impulse sequence can be expressed as the first backward
difference of the unit step sequence, i.e.,
δ [ n ] =u [ n ] −u[n−1] (2.6)

3. Sinusoidal Sequences
A sinusoidal sequence has the form

By Niguse B. Page 4
 This function can also be decomposed into its in-phase x i [ n ] and quadrature x q [ n ]
components.

4. Exponential Sequence
Exponential sequences are another important class of basic signals. The general form of an
exponential sequence is
x [ n ] =A α
n
(2.7)
If A and α are real numbers, then the sequence is real. If 0 < α < 1 and A is positive, then the
sequence values are positive and decrease with increasing n.

2.2 Discrete-Time Systems

In many applications of digital signal processing we wish to design a device or an algorithm that
performs some prescribed operation on a discrete-time signal. Such a device or algorithm is
called a discrete-time system. More specifically, a discrete-time system is a device or algorithm
that operates on a discrete-time signal, called input or excitation, according to some well-defined
rule, to produce another discrete-time signal called the output or response of the system.

A discrete-time system is defined mathematically as a transformation or operator that maps an

input sequence with values x [n] into an output sequence with values y [n]. This can be denoted
as

y [ n ] =T { x [n] } or x [ n ] T y [n] (2.8)

→

and is indicated pictorially in Figure 2.3. Equation (2.8) represents a rule or formula for
computing the output sequence values from the input sequence values. It should be emphasized
that the value of the output sequence at each value of the index n may depend on input samples
x [n] for all values of n, i.e., y at time n can depend on all or part of the entire sequence x .

Figure 2.3 Representation of a discrete-time system.

Some examples of discrete time systems

By Niguse B. Page 5
a) Moving (Running) Average system

The general moving-average system is defined by the equation

This system computes the nth sample of the output sequence as the average of (M 1 + M 2+1)
samples of the input sequence around the nth sample.
b) The ideal delay system is defined by the equation
y [ n ] =x [ n−n d ] ,−∞ < n<∞

where n d is a fixed positive integer representing the delay of the system. In other words, the
ideal delay system shifts the input sequence to the right by n d samples to form the output. If,
in the above equation n d is a fixed negative integer, then the system would shift the input to
the left by |n d| samples, corresponding to a time advance.
c) Identity system
y [ n ] =x [n]
d) Compressor systems is defined by the input-output relation by
y [ n ] =x [ Mn]

2.4 Classification of Discrete-Time Systems

Classes of systems are defined by placing constraints on the properties of the transformation T {. }.
Doing so often leads to very general mathematical representations, as we will see.

2.4.1 Memoryless systems

A system is referred to as memoryless if the output y [n] at every value of n depends only on the
input x [n] at the same value ofn .
Examples 2.4.1
a) y [ n ] =( x [n] )2 c) y [ n ] =x [ n−n d ]
b) y [ n ] =ax [n] d) y [ n ] =x [ 2 n ]

By Niguse B. Page 6
(a) and (b) are both static or memoryless systems. On the other hand the ideal delay system is not
memoryless unless, n d=0 ; in particular, that system is referred to as having “memory” whether
n d is positive (a time delay) or negative (a time advance).

2.4.2 Linearity
The general class of systems can also be subdivided into linear and nonlinear systems. A linear
system is one that satisfies the superposition principle. Simply stated, the principle of
superposition requires that the response of the system to a weighted sum of input signals be equal
to the corresponding weighted sum of the responses of the system to each of the individual input
signals.
For a certain system, let the outputs corresponding to inputs x 1 [n] and x 2 [n] are y 1 [n] and y 2 [n]
respectively. Then the relaxed system is linear if and only if

wherea is an arbitrary constant. The first property is the additivity property, and the second the
homogeneity or scaling property. These two properties together comprise the principle of
superposition, stated as

for arbitrary constants a and b .

This equation can be generalized to the superposition of many inputs. Specifically, if

where y k [n] is the system response to the input x k [n].

Example 2.4.2
Determine if the systems described by the following input-output equations are linear or
nonlinear.

By Niguse B. Page 7
 n
c) y [ n ] =x [n2 ]
a) y [n]= ∑ x [k ]
k=−∞
d) y [ n ] =x 2 [n]
b) y [ n ] =nx [n] e) y [ n ] = Ax [ n ] +B
The system of (a) is called the accumulator system, since the output at time n is the accumulation
or sum of the present and all previous input samples.
We begin by defining two arbitrary inputs x 1 [n] and x 2 [n] and their corresponding outputs

When the input is x 3 [ n ] =a x 1 [ n ] +b x 2 [n], the superposition principle requires the output
y 3 [ n ] =a y 1 [ n ] + b y 2 [n] for all possible choices of a and b. we can show this by starting from the
equations.

Thus, the accumulator system satisfies the superposition principle for all inputs and is therefore
linear. Thus it’s a linear system. Do others as an exercise on your own.

2.4.3 Time-Invariant System

A time-invariant (often referred to equivalently as a shift-invariant) system is a system for which
time shift of the input sequence causes a corresponding shift in the output sequence. Thus, for
time invariant systems its input-output characteristics do not change with time.
To determine if any given system is time invariant, first we excite the system with an arbitrary
input sequence x [n], which produces an output denoted as y [n] and delay the output by the
amount n 0 to produce y [ n−n0 ].
Next we delay the input sequence by same amount n 0 and recomputed the output as

By Niguse B. Page 8
 y [ n , n0 ]=T { x 1 [ n ] }=T { x [n−n0 ] }
Now if the outputs
y [ n , n0 ]= y [ n−n0 ] , for all possible values of n 0, the system is time invariant. On the other hand,

if the output y [ n , n0 ] ≠ y [ n−n0 ] , even for single value ofn 0, the system is time variant.

Examples 2.4.3
Determine if the systems defined by the following input-output relations are time invariant or
not.
i) y [ n ] =x [ n ] −x [n−1]
ii) y [ n ] =nx [n]
iii) y [ n ] =x [ n ] cos ω0 n
iv) y [ n ] =x [ Mn ] for all questions−∞< n<∞ .
The system defined in (i) is time invariant and the remaining systems are not.

2.4.4 Causality
A system is said to be causal if for every choice of n 0, the output sequence value at the index
n=n0 depends only on the input sequence values for n ≤ n0. That means the output of the system
at any time n depends only on the past and present inputs, but does not depend on future inputs.

Examples 2.4.4
Determine if the systems described by the following input-output equations are causal or non-
causal.
a) y [ n ] =x [ n ] −x [n−1] d) y [ n ] =ax [n]
b) y [ n ] =x [ n ] +3 x [n+ 4] e) y [ n ] =x [2n ]
n
f) y [ n ] =x [n2 ]
c) y [ n ] = ∑ x [k ]
k=−∞ g) y [ n ] =x [−n]
Solution: The systems described in parts (a), (c), and (d) are clearly causal since the output
depends only on the present and past inputs. On the other hand, the systems in parts (b), (e), and
(f) are clearly non-causal. Since the output depends on the future values of the input

2.4.5 Stability

By Niguse B. Page 9
A number of somewhat different definitions are commonly used for stability of a system.
Throughout this text, we specifically use bounded-input bounded-output stability. A system is
stable in the bounded-input, bounded-output (BIBO) sense if and only if every bounded input
sequence produces a bounded output sequence.
The input x [n] is bounded if there exists a fixed positive finite value Bx such that
¿ x [n]∨≤ Bx< ∞ , for all n .
Stability requires that, for every bounded input, there exists a fixed positive finite value B y such
that
¿ y [n]∨≤ By <∞ , for all n .
It is important to emphasize that the properties we have defined in this section are properties of
systems, not of the inputs to a system. That is, we may be able to find inputs for which the
properties hold, but the existence of the property for some inputs does not mean that the system
has the property. For the system to have the property, it must hold for all inputs.

2.5 Analysis of LTI systems

Most of practical systems are LTI or can be approximated by LTI system. Due to this reason we
consider only the analysis of LIT systems. There are two basic methods for analyzing the
behavior or response of a system to a given input signal.
The first method for analyzing the behavior of a linear system to a given input signal is first to
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.
Decomposing input signal into a weighted sum of impulse sequences proves to be
mathematically convenient and completely general. Then, using 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.
Let us describe this method mathematically: Let h [n , k ] be the response of the system to the
input δ [n−k ], an impulse occurring at n=k . That is,
h [ n , k ] =T { δ [n−k ] }
In equation above we note that n is time index and k is a parameter showing the location of the
input impulse.

By Niguse B. Page 10
We know that any arbitrary signals can be represented in terms of weighted sum of delayed
impulses and written as
∞
x [ n ] = ∑ x [ k ] δ [n−k ]
k=−∞

Then the response of the system to x [n] is the corresponding sum of weighted outputs, that is

{∑ }
∞
y [ n ] =T { x [n] }=T x [ k ] δ [n−k ]
k=−∞

Using superposition principle of linearity property, we can write as

∞
y [ n]= ∑ x [ k ] T { δ [n−k ] }
k=−∞

∞
y [ n]= ∑ x [ k ] h[n , k ] (2.14)
k=−∞

According to equation (2.14), the system response to any input can be expressed in terms of the
responses of the system to the sequencesδ [n−k ]. If only linearity is imposed, then h [n , k ] will
depend on both n and k , in which case the computational usefulness is somewhat limited. We
obtain a more useful result if we impose the additional constraint of time-invariance.

The property of time invariance implies that if h [n ] is the response to δ [n], then the response to
δ [n−k ] is h [n−k ] .With this additional constraint, equation (2.14) becomes

∞
y [ n]= ∑ x [ k ] h[n−k ], for all n (2. 15)
k=−∞

Now we observe that an LTI system is completely characterized by its impulse response h [n ],
since for the given sequences x [n] and h [n ] for all n , it is possible to compute each sample of the
output sequence y [n]. Equation (2.15) is referred to as the convolution sum, and represent by the
shorthand operator notation as
y [ n ] =x [n]∗h[n]. (2.16)
The derivation of Eq. (2.15) suggests the interpretation that the input sample at n=k , represented
as x [k ]δ [n−k ], is transformed by the system into an output sequence x [ k ] h [ n−k ] ,for−∞ <n< ∞,
and that, for each k , these sequences are superimposed (summed) to form the overall output
sequence.

By Niguse B. Page 11
However, a slightly different way of looking at Eq. (2.15) leads to a particularly useful
computational interpretation. When viewed as a formula for computing a single value of the
output sequence, Eq. (2.15) dictates that y [n] (i.e., the nth value of the output) is obtained by
multiplying the input sequence (expressed as a function of k ) by the sequence whose values are
h [n−k ] , −∞ < k <∞ for any fixed value of n, and then summing all the values of the products
x [k ]h[n−k ], with k a counting index in the summation process.

The process of computing the convolution between x [ n ] ∧h[n] for a single output y [ n ] , at n=n0
involves the following four steps.
1) Folding: fold h [k ] about k =0 to obtain h [−k ].
2) Shifting: shift h [−k ] by n 0 to the right (left) if n 0 is positive (negative), to obtain h [n 0−k ] .
Note thath [ n0 −k ] =h[−(k−n 0)].
3) Multiplication: multiply x [k ] by h [n 0−k ] to obtain the product sequence,
v n [ k ]=x [ k ] h[n0−k ].
0

4) Summation: sum all the values of the product sequence v n [ k ] to obtain the value of the
0

output y [ n0 ] at time n=n0 .

In general the preceding procedure is repeated for each value of n , thus generating the complete
output sequence y [n], for −∞ <n< ∞.

Example-1: Let us consider the discrete time system with impulse response
h [ n ] =0.8 δ [ n ] + 0.4 δ [ n−1 ] +0.2 δ [ n−2 ] +0.1 δ [ n−3 ]
what would be the system output y [ n ] if the input is
x [ n ] =δ [ n ] +0.5 δ [ n−1 ] +2 δ [ n−2 ]
The result is
y [ n ] ={ 0.8 , 0.8 , 2 ,1 , 0.45 , 0.2 }

Example-2: Consider a system with impulse response

By Niguse B. Page 12
To find the output, it is helpful to sketch the sequences x [k ] and h [n−k ] as functions of k for
different representative values of n . Figure 2.4(a) shows the sequences x [k ] and h [n−k ] , plotted
for n a negative integer. Clearly, all negative values of n give a similar picture; i.e., the nonzero
portions of the sequences x [k ] and h [n−k ] do not overlap, so
y [ n ] =0 , for n< 0.
Figure 2.4(b) illustrates the two sequences when n ≥ 0 and n−N +1≤ 0 . These two conditions can
be combined into the single condition0 ≤ n ≤ N −1.
n n
y [ n ] =∑ x [ k ] h [ n−k ]=∑ a k , for 0 ≤ n≤ N−1
k=0 k=0

Preceding equation shows that y [n] is the sum of n+1 terms of a geometric series in which the
ratio of terms is a . This sum can be expressed in closed form using the general formula

Applying this formula to the equation above, we obtain

Finally, Figure 2.4(c) shows the two sequences when 0< n−N +1 or N−1< n. As before,

but now the lower limit on the sum is n−N +1, as seen in Figure 2.1(c). Thus,

Using the geometric summation rule, the above equation can be written as

By Niguse B. Page 13
The overall output is

By Niguse B. Page 14
Figure 2.4 Sequence involved in computing a discrete convolution (a)–(c). Corresponding output
sequence as a function of n , as in (d).

2.6 Properties of LTI systems

Since all LTI systems are described by the convolution sum of Eq. (2.15), the properties of this
class of systems are defined by the properties of discrete-time convolution. Some general
properties of the class of LTI systems can be found by considering properties of the convolution
operation.

By Niguse B. Page 15
Commutative Law: the convolution operation is commutative.
x [ n ]∗h [ n ] =h [n]∗x [n]
This can be shown by applying a substitution of variables to the summation index in Eq. (2.15).
Specifically, with m=n−k ,

Hence, the system output is the same if the roles of the input and impulse responses are reversed.
Accordingly, an LTI system with input x [n] and impulse response h [n ] will have the same
output as an LTI system with input h [n ] and impulse response x [n].

Distributive Law
The convolution operation also distributes over addition; i.e.,
x [ n ]∗( h 1 [ n ] +h2 [ n ] )=x [ n ]∗h 1 [ n ] + x [n]∗h2 ¿
This follows in a straightforward way from Eq. (2.15) and is a direct result of the linearity and
commutativity of convolution. The above equation is represented pictorially in Figure 2.5, where
Figure (a) represents the right-hand side of equation and Figure (b) the left-hand side.

Associative Law
The convolution operation also satisfies the associative property, i.e.,
y [n]=(x [n]∗h1 [n])∗h2 [n]=x [n]∗(h1 [n]∗h2 [n]).
Also since the convolution operation is commutative, the preceding equation is equivalent to
y [n]=x [n]∗(h2 [n]∗h1 [n])=(x [n]∗h2 [n])∗h 1 [n ].
These equivalences are represented pictorially in Figures 2.6. Also, equations above clearly
imply that if two LTI systems with impulse responses h1 [n] and h2 [n] are cascaded in either
order, the equivalent overall impulse response h [n ] is
h [n ]=h1 [n]∗h2 [n]=h2 [n]∗h1 [n]
In a parallel combination, the systems have the same input, and their outputs are summed to
produce an overall output. It follows from the distributive property of convolution that the
connection of two LTI systems in parallel is equivalent to a single system whose impulse
response is the sum of the individual impulse responses; i.e.,
h [n ]=h1 [n]+ h2 [n].

By Niguse B. Page 16
 Figure 2.5 (a) Parallel combination of LTI systems. (b) An equivalent system.

Figure 2.6 (a) Cascade combination of two LTI systems. (b) Equivalentcascade. (c) Single
equivalent system.

Causal LTI systems

In section 2.4 we have defined the condition of causality. In this case of an LTI system, causality
can be translated to a condition on the impulse response. To determine this relationship, let us
consider a linear time-invariant system having an output at time n=n0 given by the convolution
formula
∞
y [ n0 ] = ∑ h [ k ] x [n0−k ]
k=−∞

Suppose that we subdivide the sum into two sets of terms, one set involving present and past
values of the input (i.e.. x [ n ] for n ≤ n0) and one set involving future values of the input (i.e..
x [ n ] for n ≥ n0). Thus we obtain
∞ −1
y [ n0 ]=∑ h[k ] x [ n0−k ] + ∑ h[k ] x [ n 0−k ]
k=0 k=−∞

¿ ( h [ 0 ] x [ n 0 ]+ h [ 1 ] x [ n 0−1 ] +h [ 2 ] x [ n0−2 ]+ … )

+ ( h [−1 ] x [ n 0+ 1 ]+ h [−2 ] x [ n 0+ 2 ]+ … )
We observe that the terms in first sum involve the present and past values of the input signal. On
the other hand, the terms in the second sum involve the input signal components of the future
values. Now, if the input at time n=n0 is to depend only on the present and past inputs, then,
clearly, the impulse response of the system must satisfy the condition

By Niguse B. Page 17
 h [ n ] =0 ,n< 0
Since h [n ] is the response of the LTI system to unit impulse applied at n=0 , it follows that
h [ n ] =0 for n< 0 is both a necessary and a sufficient condition for causality.
Hence an LTI system is causal if and only if its impulse response is zero for negative values of n.
It is sometimes convenient to call a sequence that is zero for n< 0 , a causal sequence, and one
that is nonzero for n< 0∧n> 0 , a non-causal sequence
Example-1
Determine the unit step response of the linear time-invariant system with impulse response
h [ n ] =a u [ n ]|a|<1
n

Stability of LTI systems

As indicated previously, stability is an important property that must be considered in any
practical implementation of a system. We defined an arbitrary relaxed system as BIBO stable if
and only if its output sequence y [n] is bounded for every bounded input x [n].
If x [n] is bounded, there exists a constant Bx such that
|x [n]|≤ B x <∞
Similarly, if the output is bounded, there exists a constant B y such that
| y [n]|≤ B y <∞ for all n
Now, given such bounded input sequence x [n] to an LTI systems, let us investigate the
implications of the definition of stability on the characteristics of the system. If we take the
absolute value of both sides of the convolution sum equation, we obtain

|∑ |
∞
| y [n]|= h [ k ] x [n−k ]
k=−∞

Now, the absolute value of the sum of terms is always less than or equal to the sum of the
absolute values of the terms. Hence
∞
| y [n]|≤ ∑ |h [k ]||x [n−k ]|
k=−∞

If the input is bounded, there exists a finite number Bx such that |x [n]|≤ B x , by substituting this
upper bound for x [n] in the equation above, we obtain
∞
| y [n]|≤ M x ∑ |h[k ]|
k=−∞

By Niguse B. Page 18
From this expression we observe that the output is bounded if the impulse response of the
satisfies the condition
∞
B h= ∑ |h [k ]|< ∞
k=−∞

That is, a LTI system is stable if its impulse response is absolutely summable. This condition is
not only suffient but necessary to ensure the stability of the system.
To show that it is also a necessary condition, we must show that if Bh=∞, then a bounded input
can be found that will cause an unbounded output. Such an input is the sequence with values

where h∗[n] is the complex conjugate of h [n ]. The sequence x [n] is clearly bounded by unity.
However, the value of the output at n=0 is
|h[k ]|2
∞
y [ 0]= ∑ =Bh
k=−∞ |h [k ]|

Therefore, if Bh=∞, it is possible for a bounded input sequence to produce an unbounded output
sequence.
Examples- Determine the following systems defined in prevoius examples are stable or not.
a) Ideal delay system
h [ n ] =δ [ n−nd ] , n d a positive fixed integer
b) Moving average system
M2
1
h [ n ]= ∑ δ[n−k ]
M 1+ M 2 +1 k=−M 1

c) Accumulator system.

d) Forward difference
h [ n ] =δ [ n+1 ] −δ [n].
e) Backward difference
h [ n ] =δ [ n ] −δ [n−1].

By Niguse B. Page 19
Given the impulse responses of these basic systems, we can test the stability of each one by
computing the sum
∞
B h= ∑ |h [n ]|
n=−∞

For the ideal delay, moving-average, forward difference, and backward difference examples, it is
clear that Bh <∞ , since the impulse response has only a finite number of nonzero samples. In
general, a system with a finite-duration impulse response (henceforth referred to as an FIR
system) will always be stable, as long as each of the impulse response values is finite in
magnitude. The accumulator, however, is unstable because
∞
Bh=∑ u [n]=∞
n=0

The impulse response of the accumulator has infinite duration. This is an example of the class of
systems referred to as infinite-duration impulse response (IIR) systems.

Exercises
Determine the range of values of the parameters a and b for which the linear time-invariant
system with impulse response
a) h [ n ] =an u [n]

{
n
a ,n≥0
b) h [ n ] =
bn , n<0
is stable.

2.7 LINEAR CONSTANT-COEFFICIENT DIFFERENCE EQUATIONS

In section 2.5 we treated linear time-invariant systems and characterized them interms of their
impulse responses. In this subsection we focus our attention on a family of linear time-invariant
systems described by an input-output relation called a difference equation with constant
coefficients.
An important class of LTI systems consists of those systems for which the input x [n] and the
output y [n] satisfy an N th-order linear constant-coefficient difference equation of the form
N M

∑ ak y [n−k ]= ∑ bm x [n−m]. (2.25)

k=0 m =0

Consider difference equation representations for some of the LTI systems that we have defined.

By Niguse B. Page 20
Example- Difference equation representation of the accumulator
The accumulator system is defined by
n
y [ n]= ∑ x [k ]
k=−∞

To show that the input and output satisfy a difference equation of the form of
Eq. (2.25), we rewrite the accumulator system equation as
n−1
y [ n]= ∑ x [k ]+ x [n]
k=−∞
n−1
y [ n−1 ]= ∑ x [k ]
k=−∞
Substituting this equation in the above relation
y [ n ] = y [ n−1 ] + x [n]
and equivalently,
y [ n ] − y [ n−1 ]=x [n]
Thus, in addition to satisfying the defining relationship of accumulator
relation, the input and output of an accumulator satisfy a linear constant-
coefficient difference equation of the form Eq. (2.25), with
N=1 , a 0=1 , a1=−1 , M =0 ,∧b0=1.
The difference equation in the form of accumulator in the form of difference
equation suggests a simple implementation of the accumulator system.
According to this equation, for each value of n , we add the current input
value x [n] to the previously accumulated sum y [n−1]. This interpretation of
the accumulator is represented in block diagram form in Figure below.
Equation above and the block diagram in Figure are referred to as a
recursive representation of the system, since each value is computed using
previously computed values. This general notion will be explored in more
detail later in this section.

By Niguse B. Page 21
 Block diagram of a recursive difference equation representing an
accumulator.

Given a linear constant coefficient difference equation as the input-output

relationship describing a LTI system, our objective in this section is to
determine an explicit expression for the output y [ n ] . Basically the goal is to
determine the output y [ n ] , for n ≥ 0, the system given a specific input x [ n ] , n ≥ 0,
and a set of initial conditions. The direct solution method assumes that the
total solution is the sum of two parts:
y [ n ]= yh [ n ]+ y p [n ]
(2.26)
where y h [ n ] is any solution to Eq. (2.25) with x [n]=0 , i.e., a solution to the
equation
N

∑ ak y h [n−k ]=0 .
k=0

(2.27)
Equation (2.28) is called the homogeneous difference equation and y h [n] the
homogeneous solution. The sequence y h [n] is in fact a member of a family of
solutions of the form
N
y h [ n ] = ∑ A m z nm ,
m=1

(2.28)
where the coefficients Am can be chosen to satisfy a set of auxiliary
conditions on y [n]. Substituting Eq. (2.28) into Eq. (2.27) shows that the
complex numbers z m must be roots of the polynomial
N
A ( z )=∑ ak z−k .
k=0

(2.39)
i.e., A ( z m )=0 , for m=1 , 2 , … , N . Equation (2.29) assumes that all N roots of the
polynomial in Eq. (2.30) are distinct. The form of terms associated with

By Niguse B. Page 22
multiple roots is slightly different, but there are always N undetermined
coefficients.
Alternatively, if the auxiliary conditions are a set of auxiliary values of y [n],
the other values of y[n] can be generated by rewriting Eq. (2.25) as a
recurrence formula, i.e., in the form
N M
ak b
y [ n ] =∑ y [n−k ]+ ∑ k x [n−k ] (2.30)
k=1 a0 k=0 a0

If the input x [n] for all n, together with a set of auxiliary values, say,
y [−1], y [−2],. . . , y [−N ], is specified, then y [0] can be determined from Eq.
(2.30).With y [0], y [−1], .. . , y [−N + 1] now available, y [1] can then be calculated,
and so on. When this procedure is used, y[n] is said to be computed
recursively; i.e., the output computation involves not only the input
sequence, but also previous values of the output sequence.
To generate values of y [n] for n<−N (again assuming that the values
y [−1], y [−2],. . . , y [−N ] are given as auxiliary conditions), we can rearrange Eq.
(2.25) in the form
N −1 M
ak b
y [ n−N ]=− ∑ y [ n−k ] + ∑ k x [n−k ]
k=0 aN k=0 a N

from which y [−N−1], y [−N−2],. . . can be computed recursively in the

backward direction.
Our principal interest in this text is in systems that are linear and time
invariant, in which case the auxiliary conditions must be consistent with
these additional requirements.
If a system is characterized by a linear constant-coefficient difference
equation and is further specified to be linear, time invariant, and causal, then
the solution is unique. In this case, the auxiliary conditions are often stated
as initial-rest conditions. In other words, the auxiliary information is that if
the input x[n] is zero for n less than some time n 0, then the output y[n] is
constrained to be zero for n less than n 0. This then provides sufficient initial
conditions to obtain y[n] for n ≥ n 0 recursively using Eq. (2.31).

By Niguse B. Page 23
To summarize, for a system for which the input and output satisfy a linear
constant coefficient difference equation:
 The output for a given input is not uniquely specified. Auxiliary
information or conditions are required.
 If the auxiliary information is in the form of N sequential values of the
output, later values can be obtained by rearranging the difference
equation as a recursive relation running forward in n, and prior values
can be obtained by rearranging the difference equation as a recursive
relation running backward in n.
 Linearity, time invariance, and causality of the system will depend on
the auxiliary conditions. If an additional condition is that the system is
initially at rest, then the system will be linear, time invariant, and causal.
Example-1: Determine the homogeneous solution of a system described by
the first order difference equation:
y [ n ] =a 1 y [ n−1 ]=x [ n ]

By Niguse B. Page 24