Sie sind auf Seite 1von 27

Chapter One

Discrete-Time Signals and Systems


Lecture #3

Rediet Million

AAiT, School Of Electrical and Computer Engineering


rediet.million@aait.edu.et

March, 2018

(Rediet Million) DSP-Lecture #3 March, 2018 1 / 27


1.2.Discrete-Time Signals and Systems

1.2.2. Discrete-Time Systems


A discrete-time system is defined mathematically as a transformation
(processor or algorithm) that transforms an input sequence x(n) into
an output sequence y (n) according to some input/output rule.

y (n) = T {x(n)}

Discrete-time system examples:


I An ideal delay: y (n) = x(n − nd ) , nd is the delay of the system.
I Compressor (down-sampler): y (n) = x(Mn)
n
I A rate expander(up-sampler): y (n) = x( M )
I Modulator:y (n) = x(n)cos(nωo )

(Rediet Million) DSP-Lecture #3 March, 2018 2 / 27


Discrete-Time Systems...

I An accumulator system is defied as


n−1
P
y [n] = x[k] + x(n) i.e y (n) = y (n − 1) + x(n)
k=−∞

I A moving average system is defied by


M2
1 P
y [n] = M1 +M 2 +1
x(n − k)
k=M1
- This system computes the nth sample of the output sequence as the
1
average of M1 +M 2 +1
samples of the input sequence around the nth sample.

(Rediet Million) DSP-Lecture #3 March, 2018 3 / 27


Discrete-Time Systems...

I Moving average smoothes out rapid variations e.g

using 5 point moving average


4
x(n)= s(n)signal +d(n)noise y [n] = 15
P
x(n − k)
k=0

(Rediet Million) DSP-Lecture #3 March, 2018 4 / 27


Discrete-Time Systems...
Classification of DT Systems

1.Linear system
A system is linear if it satisfies superposition principle.
I Additive property:

T {x1 (n) + x2 (n)} = T {x1 (n)} + T {x2 (n)} = y1 (n) + y2 (n)

I Homogeneity or scaling property:


T {ax(n)} = aT {x(n)} = ay (n) where a is arbitrary constant.
I These two properties are combined into the principle of
superposition:

T {ax1 (n) + bx2 (n)} = aT {x1 (n)} + bT {x2 (n)} = ay1 (n) + by2 (n)

(Rediet Million) DSP-Lecture #3 March, 2018 5 / 27


Discrete-Time Systems...
Classification of DT Systems...

Example-1
- Accumulator: Example-2
n
- Offset accumulator:
P
y [n] = x[k] n
k=−∞
P
y [n] = C + x[k]
x(n) = a · x1 (n) + b · x2 (n) k=−∞
Pn n
y [n] = [a · x1 (k) + b · x2 (k)]
P
y [n] = C + [a · x1 (k) + b · x2 (k)]
k=−∞ k=−∞
n
P n
P n n
= [a · x1 (k)] + [b · x2 (k)]
P P
= C +a· [x1 (k)]+b· [x2 (k)]
k=−∞ k=−∞ k=−∞ k=−∞
n n
=a·
P
[x1 (k)] + b ·
P
[x2 (k)] 6= a · y1 (n) + b · y2 (n)
k=−∞ k=−∞ I Thus,the system is non-linear
= a · y1 (n) + b · y2 (n) unless C=0.
I Thus,the system is linear.

(Rediet Million) DSP-Lecture #3 March, 2018 6 / 27


Discrete-Time Systems...
Classification of DT Systems...

2.Time(shift)-invariant system
Time shift or delay of the input sequence causes a corresponding shift
in the output sequence.
I if x(n) −→ y (n) implies that
x(n − no ) −→ y (n − no )
Example :
Take the system y (n) = n · x(n)
If x(n) = x1 (n − no ) then
y (n) = n · x1 (n − no )
6=
y (n − no ) = (n − no ) · x1 (n − no )
⇒ Hence, the system is not shift invariant system.

(Rediet Million) DSP-Lecture #3 March, 2018 7 / 27


Discrete-time systems...
Classification of DT Systems...

3.Causal system
A system is causal if the output value at n = no depends only on the
input sequence values for n ≤ no .
In other words, the output of the system at any time depends only on
I present and past values of the input x(n)
i.e[x(n), x(n − 1), x(n − 2), ..x(n − no )]
I but does not depends on future inputs
i.e[x(n + 1), x(n + 2), ..x(n + no )] → non-causal system.
Example :
-Forward difference system,y (n) = x(n − 1) − x(n), is causal system.
-Backward difference system,y (n) = x(n + 1) − x(n), is non-causal
system.

(Rediet Million) DSP-Lecture #3 March, 2018 8 / 27


Discrete-Time Systems...
Classification of DT Systems...

4.Stable system
A system is said to be stable in the Bounded Input-Bounded Output
sense if, for any input that is bounded,|x(n)| ≤ A < ∞ , the output
will be bounded,|y (n)| ≤ B < ∞.
If for some bounded input x(n) the output is unbounded (infinite)
,then the system is classified as unbounded.
Pn
Example : Consider the accumulator system y [n] = x[k]
k=−∞
When x(n) = u(n) ,which is clearly bounded by A = 1. For this
input,the output of the
( accumulator is
Pn (n + 1) , n ≥ 0
y [n] = u[k] =
k=−∞ 0 ,n < 0
There is no finite choice for B such that ,(n − 1) ≤ B < ∞ for all n.
⇒ Thus,the system is unstable.
(Rediet Million) DSP-Lecture #3 March, 2018 9 / 27
Discrete-Time Systems| Classification of DT Systems

(#2) Class exercises & Assignment


1) For each of the following systems,determine whether the system is
linear system and shift invariant system .
a.y (n) = x(n)sin( 2πn π
7 + 6)
b.y (n) = log [x(n)]
c.y (n) = 2x(n) + 3
d.y (n) = [x(n)]2
e.y (n) = e x(n)
2) For each of the following systems,determine whether the system is
stable and causal system .
a.y (n) = [x(n + 1)]2
e x(n)
b.y (n) = x(n−1)
c.y (n) = cos[x(n + 2)]

(Rediet Million) DSP-Lecture #3 March, 2018 10 / 27


1.3.LTI System and Discrete-Time Fourier Transform

1.3.1 Linear Time-Invariant (LTI) Systems


This is a particularly important class of systems that combines the
properties of linearity and time-invariance.
Important due to convenient representations and significant
applications.
A linear time-invariant system is completely characterized by its
impulse response h(n).
I Given a system :

If x(n) = δ(n) then y (n) ≡ h(n) → impulse response

(Rediet Million) DSP-Lecture #3 March, 2018 11 / 27


Linear Time-Invariant (LTI) Systems...

I For example take a simple system:

I Using signal decomposition we know that we can express any sequence


by unit sample (impulse) δ(n) as :

P
x(n) = x(k)δ(n − k)
k=−∞

(Rediet Million) DSP-Lecture #3 March, 2018 12 / 27


Linear Time-Invariant (LTI) Systems...

I When we apply the above sequence x(n) as an input to LTI system


- Due to linearity :

X
y (n) = T [x(n)] = T [ x(k)δ(n − k)]
k=−∞

X
y (n) = T [x(n)] = x(k)T [δ(n − k)]
k=−∞

- Due to Time(shift) invariance :


T [δ(n)] = h(n) ⇒ T [δ(n − k)] = h(n − k)
-Therefore,

P
y (n) = T [x(n)] = x(k)h(n − k) ≡ x(n) ∗ h(n)
k=−∞

P
= h(k)x(n − k).... show the proof !
k=−∞
This expression is called convolution sum.
(Rediet Million) DSP-Lecture #3 March, 2018 13 / 27
Linear Time-Invariant (LTI) Systems...
Steps involved to perform the convolution sum


X
y (n) = x(n) ∗ h(n) = x(k)h(n − k)
k=−∞

1 Consider x(k) and h(k) as a function of k.


2 Obtain the sequence h(n − k)
I Reflect h(k) about the origin to get h(−k).
I Shift the reflected sequence by n(to the right if n > 0 and to the
left if n < 0).
3 Multiply the two sequence x(k) & h(n − k) and sum the product for
all values of K.

(Rediet Million) DSP-Lecture #3 March, 2018 14 / 27


Linear Time-Invariant (LTI) Systems...
Examples

1. Analytically convolution
( of two signals:
n
a ,n ≥ 0
x(n) = an u(n) = and h(n) = u(n)
0 ,n < 0
Using direct evaluation of the convolution sum we find
∞ ∞
ak u(k)u(n − k)
P P
y (n) = x(n) ∗ h(n) = x(k)h(n − k) =
k=−∞ k=−∞
From the definition of unit step u(k) = 0 for k < 0 and u(n − k) = 0 for
k > n.Thus,
n 1 − an+1 1 − an+1
ak =
P
y (n) = =( )u(n)
k=0 1−a 1−a
Some useful formulas:
∞ 1 N−1 1 − aN
an =
P P n
for |a| < 1 a =
n=0 1−a n=0 1−a
N−1 1 ∞ a
nan =
P P
n = N(N − 1) for |a| < 1
n=0 2 n=0 (1 − a)2
(Rediet Million) DSP-Lecture #3 March, 2018 15 / 27
Linear Time-Invariant (LTI) Systems...
Examples

2.Graphical convolution of two signals:

(Rediet Million) DSP-Lecture #3 March, 2018 16 / 27


Linear Time-Invariant (LTI) Systems...
Examples

(Rediet Million) DSP-Lecture #3 March, 2018 17 / 27


Linear Time-Invariant (LTI) Systems...
Properties of LTI-systems

LTI systems are completely characterized by the impulse response


function,h(n). Therefore, we can determine causality and stability
from h(n):
I Causality: h(n) = 0 for n < 0.

P
I Stability: |h(n)| < ∞.
n=−∞
For the convolution operator,y (n) = x(n) ∗ h(n),there are several
properties of interest :
I Commutative: x(n) ∗ h(n) = h(n) ∗ x(n)
I Associative: [x(n) ∗ h1 (n)] ∗ h2 (n) = x(n) ∗ [h1 (n) ∗ h2 (n)]
I Distributive: x(n) ∗ [h1 (n) + h2 (n)] = x(n) ∗ h1 (n) + x(n) ∗ h2 (n)
Discrete-time convolution will be discussed detail in chapter three.

(Rediet Million) DSP-Lecture #3 March, 2018 18 / 27


Linear Time-Invariant (LTI) Systems...

(#3) Class exercises & Assignment


For each of the following pairs of sequences,x(n), represents the input to
an LTI system with impulse response h(n).Determine each output y (n)
using discrete time convolution.
a.
a.

(
αn , 0 ≤ n ≤ 10
( b. h(n) =
1 , 0 ≤ n ≤ (N − 1) 0 , otherwise
b. h(n) =
0 , otherwise (
1 ,0 ≤ n ≤ 5
x(n) = an u(n) x(n) =
0 , otherwise
(Rediet Million) DSP-Lecture #3 March, 2018 19 / 27
1.3.2 Linear Constant Coefficient Difference Equation

From a computational point of view sometimes convolution sum


representation is not very efficient.
For example,take a system that has a unit sample response
h(n) = αn u(n):
I Using convolution sum we can express y (n) as :

αn x(n − k)
P
y (n) =
k=0
I But we can express the output more efficiently in terms of past
values of the output in addition to the current and past values of the
input .
y (n) = αy (n − 1) + x(n)
- This equation is a special case of what is known as a linear constant
coefficient difference equation (LCCDE).

(Rediet Million) DSP-Lecture #3 March, 2018 20 / 27


Linear Constant Coefficient Difference Equation...

The general form of LCCDE is :


M
X N
X
y (n) = b(k)x(n − k) − a(k)y (n − k)
k=0 k=1

- Where the coefficients a(k) and b(k) are constants that define the
system.
- If a(k) 6= 0 → recursive difference equation.
- If a(k) = 0 → non-recursive difference equation.
I Before solving LCCDE,it is necessary to specify a set of an initial
conditions.
For LTI system that is described by a difference equation, the unit
sample response,h(n), is found by solving the difference equation for
x(n) = δ(n) assuming initial condition is zero.

(Rediet Million) DSP-Lecture #3 March, 2018 21 / 27


Linear Constant Coefficient Difference Equation...

For a non-recursive system ,a(k) = 0, the difference equation becomes


M
X N
X
y (n) = b(k)x(n − k) ⇒ h(n) = b(k)δ(n − k)
k=0 k=0

I Thus, h(n) is finite in length and the system is referred to as a


finite impulse response(FIR) system.
IHowever, if a(k) 6= 0, the unit sample response h(n) is infinite in
length and the system is referred to as an infinite impulse response
(IIR)system.
There are several difference methods that one may use to solve
LCCDEs for a general input x(n).
IHomogeneous and particular solution approach.
I Z-transform approach→ will discussed in chapter 4.

(Rediet Million) DSP-Lecture #3 March, 2018 22 / 27


Linear Constant Coefficient Difference Equation...
Homogeneous and Particular Solutions

Given an LCCDE, the general solution is a sum of two parts:


y (n) = yh (n) + yp (n)
- yh (n) is known as the homogeneous solution and yp (n) is the
particular solution.
Homogeneous solutions (zero input response):
I The homogeneous solution is the response of the the system to the
initial conditions, assuming that the input x(n) = 0.
I The solution is found by solving the difference equation as :
N
X
a(k)y (n − k) = 0
k=0
when a0 = 1
N
X
y (n) + a(k)y (n − k) = 0
k=1
(Rediet Million) DSP-Lecture #3 March, 2018 23 / 27
Linear Constant Coefficient Difference Equation...
Homogeneous and Particular Solutions...

I The general method of solution to the homogeneous equation is to


assume a homogeneous solution, y (n), as

yh (n) = λn

I If the assumed solution is a solution, then it has to satisfy the equation:


N
X
ak λn−k = 0
k=0

assuming a0 = 1
λn + a1 λn−1 + a2 λn−2 + .....aN λn−N = 0
λn−N [λN + a1 λN−1 + a2 λN−2 + .....aN ] = 0
I The polynomial in braces is called the characteristics polynomial.
Because it is of degree N, it may have N roots (distinct or repeated),
which may be either real or complex.
(Rediet Million) DSP-Lecture #3 March, 2018 24 / 27
Linear Constant Coefficient Difference Equation...
Homogeneous and Particular Solutions...

I Distinct roots: The form of the most general solution will be a linear
combination of those of each characteristics equations as

yh (n) = c1 λn1 + c2 λn2 + c3 λn3 + ...cN λnN


- ci is determined from the initial conditions.

I Repeated roots: If some of the roots of the characteristics equations


are repeated then the form of the solution will be modified as

yh (n) = c11 λn1 + c12 nλn1 + c13 n2 λn1 + ... + c1m nm−1 λn1 + c2 λn2 + ...cN λnN

(Rediet Million) DSP-Lecture #3 March, 2018 25 / 27


Linear Constant Coefficient Difference Equation...
Homogeneous and Particular Solutions

Particular solutions (zero state response):


I It is the solution response to the applied input signal.
I To determine the particular solution yp (n) is to assume that the
response will have the same form as the input itself within a multiplier
constant.

(Rediet Million) DSP-Lecture #3 March, 2018 26 / 27


Linear Constant Coefficient Difference Equation...

(#4 ) Class exercises & Assignment


1) Determine the total response,y (n),of the following causal systems
when the forcing function is x(n). Assume initial conditions are zero.
5 1
a.y (n) = y (n − 1) − y (n − 2) + x(n) , x(n) = 2n u(n)
6 6
b.y (n) − 3y (n − 1) − 4y (n − 2) = x(n) + 2x(n − 1) , x(n) = δ(n)

2) Determine the impulse response and the unit step response of the
system described by the following difference equations for n ≥ 0.

a.y (n) = 0.7y (n − 1) − 0.1y (n − 2) + 2x(n) − x(n − 2)

b.y (n) = 0.6y (n − 1) − 0.08y (n − 2) + x(n)

(Rediet Million) DSP-Lecture #3 March, 2018 27 / 27

Das könnte Ihnen auch gefallen