Chapter 2: DT Signals and Systems

DT Signals
Some DT signals we encounter frequently are the unit sample,
unit step, ramp, and exponential functions.

1. Unit sample
The unit sample signal is defined as
1 for n = 0
(
δ(n) =
0 otherwise
This is also referred to as the unit impulse signal.

2. Unit step
The unit step signal is defined as
(
1 for n ≥ 0
u(n) =
0 otherwise

1
DT Signals (ctd)

The unit sample and the unit step signals are related as:
n
X ∞
X
u(n) = δ(k) = δ(n − m)
k=−∞ m=0

δ(n)
[ ] = u(n)
[ ] − u(n
[ − 1)
]

3. Unit ramp
The unit ramp signal is defined as
(
n for n ≥ 0
ur (n) =
0 otherwise

4. Exponential
The exponential signal is defined as
x(n) = an for all n
If a is real, then x(n) is a real signal. If a is complex, then a can
be expressed as
a = re jθ
and
x(n) = rn(cos θn + j sin θn)
2
Classification of DT Signals

1.Energy and Power Signals

The Energy E of a DT signal is defined as
∞
X
E= |x(n)|2
n=−∞
which can be finite or infinite.
If E is finite, the signal is called an Energy signal.
The power of a signal is defined as
N
1 X
P = lim |x(n)|2
N→∞ 2N + 1
n=−N
If E is finite, then P will be zero.
If E is infinite, then P can be finite or infinite.

2. Periodic and aperiodic signals

A signal is said to be periodic if and only if
x(n) = x(n + N) for all n
The lowest value of N for which the above holds is the
fundamental period. Periodic signals are power signals.

3. Symmetricity
A signal is said to be symmetric or even-symmetric if
x(−n) = x(n)
and antisymmetric or odd-symmetric if
x(−n) = −x(n)
3
Simple manipulations

1. Time shift

2. Time reversal and shift

4
Classification of DT systems

A DT system takes DT signals as inputs and produces DT sig-

nals as outputs. In this course we will consider only single-input
single-output systems.

We use the following notations to indicate the relationship be-

tween x(n), y(n) and the system.
y(n) = T [x(n)]
T
x(n) −→ y(n)
where T denotes the transformation performed by the system.

1. Linearity
For any linear DT system, if
an input x1(n) produces an output y1(n) and
an input x2(n) produces an output y2(n),
then an input αx1(n) + βx2(n) should produce an output
αy1(n) + βy2(n).
T
αx1(n) + βx2(n) −→ αy1(n) + βy2(n)
or

T [αx1(n) + βx2(n)] = αT [x1(n)] + βT [x2(n)]

5
Classification of DT systems(ctd)

2. Time-invariance
If an input x1(n) produces an output y1(n) , then the same input
applied at a later time, should produce the original output de-
layed by the same amount of time.
i.e. The input x(n − n0) should produce the output y(n − n0)
T
x(n − n0) −→ y(n − n0)

3. Static vs. Dynamic systems (Memorylessness)

For a static (memoryless) system, the output at any given time
depends only on the current value of the input, not on previous
or future values of inputs. Therefore, at an instance n = n0, the
output y(n0) depends only on x(n0).
For dynamic systems, the output at any given instance can de-
pend on current and past values of the input. An example of a
dynamic system is
∞
X
y(n) = x(n − k)
k=0

6
Classification of DT systems (ctd)

4. Causality
For a causal system, the output at any given time does not de-
pend on the future values of the input. It depends only on the
current value and the past values of the input.
If we consider time instance n = n0, the output y(n0) depends
only on x(n) where n ≤ n0. All physical systems are causal.

5. Stability
Every Bounded Input to a stable system must result in a Bounded
Output. This is called BIBO stability.
If |x(n)| ≤ B x < ∞ for all n,
then |y(n)| ≤ By < ∞ for all n.

7
DT System models

In this course, we will only deal with linear and time-invariant

(LTI) systems.

For CT systems, we have seen that the system model is a dif-

ferential equation. For LTI DT systems, the system model is a
difference equation of the form:

aN y(n − N) + aN−1y(n − N − 1) + . . . + a0y(n) =

b M x(n − M) + b M−1 x(n − M − 1) + . . . + b0 x(n)

N
X M
X
ak y(n − k) = b j x(n − j)
k=0 j=0
This type of difference equations are called constant-coefficient
difference equations because the coefficients ak and b j are all
constants.

Impulse response
The Impulse Response of a system is its response to a unit
impulse at the input. It is normally denoted by h(n)

Step response
The response of a system to a unit step input is called the Step
Response of the system. It is normally denoted by a(n)

8
Response to arbitrary inputs

Any input x(n) can be represented as a weighted sum of

impulse functions
+∞
X
x(n) = x(k) δ(n − k)
k=−∞
For example, sequence x(n) = [2, 0, 5, 3, 7] can be expressed as

x(n) = 2δ(n) + 5δ(n − 2) + 3δ(n − 3) + 7δ(n − 4)

If the system is linear and time-invariant, its response to x(n)
can be obtained as the combination of its responses to the
weighted sum of impulses. Therefore
 +∞ 
 X 
y(n) = T [x(n)] = T   x(k)δ(n − k)
k=−∞
+∞
X
= x(k)T [δ(n − k)]
k=−∞
+∞
X
= x(k) h(n − k)
k=−∞
= x(n) ∗ h(n)
This is called the convolution sum and is expressed using the
notation x(n) ∗ h(n).
The convolution operation is commutative. i.e.

x(n) ∗ h(n) = h(n) ∗ x(n)

+∞
X +∞
X
x(k) h(n − k) = h(k) x(n − k)
k=−∞ k=−∞

9
Evaluation of the Convolution Sum

To obtain the value of y(n) at n = n0, it is necessary to determine

the sum +∞
X
y(n0) = x(k) h(n0 − k)
k=−∞
It should be noted that, in the above expression k is the index
of summation and the time variable for x(k) and h(n − k).
The evaluation of y(n0) requires 4 intermediate steps.

1. Folding or Time Reversal

h(k) is time-reversed to obtain h(−k)
2. Shifting
The signal h(−k) is shifted to the right to obtain h(n0 − k)
(right shift if n0 is positive, and left shift otherwise).
3. Multiplication
Multiply x(k) and h(n0 − k) to obtain the sequence
vn0 (k) = x(k) h(n0 − k).
4. Summation
Sum the sequence vn0 (k) w.r.t. k to determine
+∞
X
y(n0) = vn0 (k)
k=−∞

These four steps must be repeated for all values of n to deter-

mine y(n). If the durations of the signals x(n) and h(n) are N1
and N2 respectively, the duration of y(n) would be N1 + N2 − 1
and the above steps must be repeated that many times.

10
Graphical Evaluation of the Convolution Sum

Consider h(n) = [1, 2, 1, −1] and x(n) = [1, 2, 3, 1]

↑ ↑

We first time reverse h(k) to obtain h(−k).

We can compute y(0) as
∞
X ∞
X
y(0) = x(k) h(−k) = v0(k) = 4
k=−∞ k=−∞

Similarly, we can compute y(1) as

∞
X ∞
X
y(1) = x(k) h(1 − k) = v1(k) = 8
k=−∞ k=−∞

To determine y(−1), we compute

∞
X ∞
X
y(−1) = x(k) h(−1 − k) = v−1(k) = 1
k=−∞ k=−∞

The value of y(n) for other n can be computed in a similar

manner.

11
12
Analytical Evaluation of the Convolution Sum

Consider h(n) = u(n) − u(n − N) and x(n) = anu(n)

Case 1: n < 0
There is no overlap between x(k) and h(n − k) (figure 2.10(a))
and the product x(k)h(n − k) is zero for all k.
Therefore, y(n) = 0 for n < 0.

Case 2: 0 ≤ n ≤ N − 1
As seen from figure 2.10(b), x(k) and h(n − k) overlap from k = 0
n+1
to k = n. Therefore y(n) = nk=0 ak = 1−a
P
1−a

Case 3: n ≥ N
The overlap between x(k) and h(n − k) is now from
k = n − N + 1 to k = n (figure 2.10(c)). Therefore
n
X
y(n) = ak
k=n−N+1
an−N+1 − an+1
=
1−a
1 − a N!
= an−N+1
1−a

The sequence y(n) is shown in figure 2.10(d)

13
 h[n – k]
x [k]

n 0 k
n – (N – 1) (a)

0 n k
n – (N – 1)
(b)

0 n k
n – (N – 1)
(c)

y [n]

0 k
N–1
(d)

Figure 2.10 Sequence involved in computing a discrete convolution. (a)–(c) The

sequences x [k ] and h [n − k ] as a function of k for different values of n . (Only
nonzero samples are shown.) (d) Corresponding output sequence as a function
of n .
From Discrete-Time Signal Processing, 2e by Oppenheim, Schafer, and Buck ©1999-2000 Prentice Hall, Inc.
Properties of the Convolution

1. Identity
x(n) ∗ δ(n) = x(n)
2. Shifting
x(n) ∗ δ(n − k) = x(n − k)
3. Commutative law

x(n) ∗ h(n) = h(n) ∗ x(n)

4. Associative law

[x(n) ∗ h1(n)] ∗ h2(n) = x(n) ∗ [h1(n) ∗ h2(n)]

5. Distributive law

x(n) ∗ [h1(n) + h2(n)] = x(n) ∗ h1(n) + x(n) ∗ h2(n)

Causal LTI systems

For a causal LTI system, the impulse response h(n) would be a
causal sequence. (i.e. h(n) = 0 for n < 0.)
The convolution sum in this case reduces to
+∞
X n
X
y(n) = h(k) x(n − k) = x(k) h(n − k)
k=0 k=−∞
If the sequence x(n) is also causal, (i.e. x(n) = 0 for n < 0), then
the convolution sum further reduces to
n
X n
X
y(n) = h(k) x(n − k) = x(k) h(n − k)
k=0 k=0

15
DT systems described by difference equations

A system whose output y(n) depends on past outputs is called

a recursive system. The parameter N is called the order of the
system. For a non-recursive system, ak = 0 for k > 0.

For recursive systems, we can express y(n) as

N
X M
X
y(n) = − ak y(n − k) + b j x(n − j)
k=1 j=0
where the parameters ak and b j have been normalized w.r.t a0.

The output y(n) of a LTI system can be considered to consist of

two components, namely the zero-input response and the zero-
state response.

Zero-Input Response (ZIR)

This is the response of the system due to energy stored within
the system (initial conditions of the system). It is also called
the natural response or the free response. The ZIR does not
depend on the input signal x(n).

Zero-State Response (ZSR)

This is the response of the system due to the input signal x(n)
only. It is also called the forced response. The ZSR does not
depend on the initial conditions of the system.

For LTI systems, the total response is equal to the sum of the
zero-input and zero-state responses.
y(n) = yzi(n) + yzs(n)
16
Solution of linear, constant-coefficient difference equations

Solving a linear, constant-coefficient difference equation can be

done in two ways. The direct method is solving the equation
in the time domain. The indirect method is solving it using the
z-transform.
The direct method involves finding the solution as the sum of
two parts, namely the homogeneous or complementary solu-
tion and the particular solution:

y(n) = yh(n) + y p(n)

Homogeneous solution
This is the solution to the homogeneous difference equation
N
X
ak y(n − k) = 0
k=0
To find the homogeneous solution, we assume that yh(n) is ex-
ponential. i.e. yh(n) = λn. Substitution in the above equation
yields
N
X
ak λn−k = 0
k=0
This equation is called the characteristic polynomial. In general
it has N roots, which may be real or complex. For simplicity, we
will assume that all the roots are distinct. Thus, yh(n) can be
expressed as
yh(n) = c1λn1 + c2λn2 + . . . + cN λnN
The coefficients c1, c2, . . . cN are determined using the initial con-
ditions specified for the system.
17
Example

Consider the first-order system described by the difference equa-

tion

y(n) + a1y(n − 1) = x(n)

and the initial condition y(−1) = A. The homogeneous difference
equation for the system is

y(n) + a1y(n − 1) = 0
and the characteristic polynomial is

λn + a1λn−1 = 0
λn−1(λ + a1) = 0
λ = −a1
Therefore, the homogenous solution is

yh(n) = c1(−a1)n
Using the initial conditions, we find c1 to be

c1 = yh(0) = −a1y(−1) = −a1 A

This gives us the zero-input response of the system

yzi(n) = A(−a1)n+1 for n ≥ 0

18
Multiple Roots for the Characteristic Polynomial

The form of the homogeneous solution

yh(n) = c1λn1 + c2λn2 + . . . + cN λnN
is valid only when all the roots of the characteristic polynomial
are distinct. If there are mutiple roots, this needs to be modified.

If there is a root λ1 of multiplicity m, then the homogeneous

solution becomes:
yh(n) =c1λn1 + c2nλn1 + c3n2λn1 + . . . + cmnm−1λn1
+ cm−1λnm+1 + . . . + cN λnN

Example
Consider the third order system described by
y(n) − 5y(n − 1) + 8y(n − 2) − 4y(n − 3) = x(n) + 2x(n − 1)
The characteristic polynomial for the system is

λn − 5λn−1 + 8λn−2 − 4λn−3 = 0

λn−3(λ3 − 5λ2 + 8λ − 4) = 0
The roots of the polynomial are λ = 2, 2, 1.
The root λ = 2 is a multiple root. Hence, the homogeneous
solution is
yh(n) = c1(2)n + c2n(2)n + c3(1)n

19
Particular solution

The particular solution y p(n) should satisfy the system differ-

ence equation for the particular input x(n).
To find the particular solution, we assume for y p(n) a form that
depends on the input signal x(n).

x(n) y p(n)
A (a constant) K
AM n K Mn
An M K0n M + K1n M−1 + . . . + K M
Ann M An(K0n M + K1n M−1 + . . . + K M )
A cos(ω0n) K1 cos(ω0n) + K2 sin(ω0n)
A sin(ω0n) K1 cos(ω0n) + K2 sin(ω0n)

Example:
Consider the difference equation y(n) + a1y(n − 1) = x(n) and a
unit step signal as the input x(n) = u(n).
We assume that the particular solution is of the form
y p(n) = Ku(n). By substituting this we obtain
Ku(n) + a1 Ku(n − 1) = u(n)
To determine the constant K , we must substitute a value for n
so that none of the terms in the equation vanish.
In this case, for any n ≥ 0,
K + a1 K = 1
1
K=
1 + a1
Therefore
1
y p(n) = u(n)
1 + a1
20
Total solution

For a LTI system, the total solution is the sum of the homoge-
neous solution and the particular solution.

y(n) = yh(n) + y p(n)

Here, y(n) contains the constant parameters c1, c2, . . . cN
contained in the homogeneous solution. The values of these
constants must be determined to satisfy the initial conditions.

Example
Consider the total solution to the difference equation
y(n) + a1y(n − 1) = x(n) for a unit step input x(n) = u(n)
and the initial condition y(−1) = A.
From previous work, we obtain
1
y(n) = c1(−a1)n + u(n) for n ≥ 0
1 + a1
By substituting n = 0, we obtain
1
y(0) = −a1 A + 1 = c1 +
1 + a1
a1
c1 = −a1 A +
1 + a1
The total solution is
1 − (−a )n+1
1
y(n) = A(−a1)n+1 + for n ≥ 0
1 + a1

21
Summary

We now summarize the procedures for obtaining the zero-state

response, zero-input reponse, and the total response of a sys-
tem, when the system difference equation is given.

1. The homogeneous solution yh(n) is the solution of the

homogeneous difference equation. It has undetermined
constants c1, c2, . . . cN .
2. The zero-input response of the system can be found by
determining the values of the constants c1, c2, . . . cN of the
homogeneous solution using the initial conditions of the
system.
3. The particular solution y p(n) is found by assuming a suit-
able form for y p(n) depending on the input x(n). The partic-
ular solution does not have any undetermined constants.
4. The sum of the homogeneous solution and the particular
solution gives the total response of the system.
y(n) = yh(n) + y p(n)
5. The zero-state response of the system can be found by
determining the values of c1, c2, . . . cN in yh(n) + y p(n)
assuming an initially relaxed system. i.e.
y(−1) = y(−2) = . . . = y(−N + 1) = 0
6. The total response of the system can be determined as the
sum of the zero-input and zero-state responses.
7. The total response can also be found by determining the
values of c1, c2, . . . cN in yh(n) + y p(n), using the actual initial
conditions of the system, y(−1) = A, y(−2) = B, etc.

22
Impulse Response

The Impulse response of a system is defined as its zero-state

response to a unit impulse input.

When the input signal x(n) is the unit impulse δ(n), the particular
solution of the difference equation is zero, as x(n) = 0 for n > 0.

Therefore, the impulse response consists only of the homoge-

neous solution, with the values of the constants c1, c2, . . . cN de-
termined to satisfy the initial conditions dictated by the impluse
input, for an initially relaxed system.

Example
Consider the system
y(n) − 3y(n − 1) − 4y(n − 2) = x(n) + 2x(n − 1)
The homogeneous soultion for this system is
yh(n) = c1(−1)n + c2(4)n for n ≥ 0
When x(n) = δ(n), for an initially relaxed system
y(0) = x(0) = 1
y(1) = 3y(0) + 2x(0) = 5
By substituting n = 0 and n = 1 in yh(n) we obtain
y(0) = c1 + c2
y(1) = −c1 + 4c2
This gives the values c1 = − 15 and c2 = 6 . The impulse reponse
5
of the system is " #
1 6
h(n) = − (−1)n + (4)n u(n)
5 5
23