Digital signal processing: Lecture 2
Linear and Time Invariant
Systems - I
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/1
Topics of last lecture
• Signals: Speech, audio, light, radio, TV, radar, etc.
• Signal processing: Enhancement, restoration,
reconstruction, synthesis, estimation, etc.
• Signal representation: Continues, discrete, and digital
• Fundamental signals: Unit sample, unit step,
exponential, sinusoid
• System representation: A functional module between
an input signal and an output signal
Produced by Qiangfu Zhao (sine1995), All rights reserved © DSP-Lec02/2
Topics of this lecture
• Linear and time- • 線形時不変システム
invariant system • インパルス応答
• Impulse response • 畳み込み和
• Convolution sum • システムの接続
• Connection methods • 因果的システム
• Causal system • 安定的システム
• Stable system
Chapter 5, start from page 66 of textbook
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/3
Linear system (線形システム）
The output y(n) is found
using the same linear
x1(n) y1(n) combination as that
S[ ] used for finding the
input x(n). This is called
superposition principle
x2(n) y2(n)
S[ ] (重ね合わせの原理)
x(n)=ax1(n)+bx2(n) y(n)=ay1(n)+by2(n)
S[ ]
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/4
Time-invariant system (時不変システム）
0 0
S[ ]
n0 n0
y ( n ) = S [ x ( n )]
y ( n − n0 ) = S [ x ( n − n0 )]
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/5
Example:
例題 Digital filter is a kind of system for signal processing.
Suppose that the LTI digital filter is defined by (a) or (b), try to
discuss its linearity (T=1). Taken from the reference book
（ただし、T=1とする）
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/6
Example: For the system given by (a) or (b), try to discuss if the
例題
system is time-invariant or not.
Taken from the reference book
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/7
Impulse response (インパルス応答）
1 n =0
δ ( n) = {
0 n≠0
This function is called the impulse signal, unit sample
signal, delta signal, and others (see pp. 13-14 in the text).
The output (response) of a system to the impulse signal
is called the impulse response （インパルス応答）.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/8
Response of an LTI system
(線形時不変システムの応答）
∞
x(n) = ∑ x(k )δ (n − k )
k = −∞
where
1 n =k
δ (n − k ) = {
0 n ≠k
suppose that the impulse response is h( n ). Since the
system is time - invariant,
S [δ ( n )] = h( n ) → S [δ ( n − k )] = h( n − k )
with these equations, we can find the output as follows.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/9
Convolution sum (畳み込み和）
linearity of the system
y ( n ) = S [ x ( n )]
∞ Time-invariance of the system
= S [ ∑ x ( k )δ ( n − k )]
k = −∞
∞
= ∑ x(k ) S[δ (n −k )]
k = −∞
∞
= ∑ x ( k )h ( n − k ) ∞
k = −∞
y (n) = ∑ x ( k )h ( n − k ) = x ( n ) ∗ h ( n )
k = −∞
x(n) y (n ) = x(n ) * h(n )
h (n )
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/10
Importance of impulse response
• The impulse response h(n) can determine the
characteristics of an LTI system completely.
• For any input x(n), the output y(n) is the convolution sum
of x(n) and h(n).
• Convolution sum is commutative (可換 ), that is
∞
y (n) = ∑ x ( k )h ( n − k ) = x ( n ) ∗ h ( n )
k = −∞
∞
= ∑ h(k ) x(n − k ) = h(n ) ∗ x(n )
k = −∞
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/11
Classification of systems
based on the impulse response
• Digital filters can be divided into two categories
based on the length (duration) of the impulse
response.
• If the impulse response is finite length, the
system is called an FIR (Finite Impulse
Response) system or FIR filter.
• If the length of the impulse response if infinite,
it is an IIR (Infinite Impulse Response) system,
or IIR filter.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/12
Example 5.2 pp. 70-71
• Suppose that the impulse response of a
digital filter is given below:
1 n≥0
h( n ) = α ⋅ u0 ( n )
n
u0 ( n ) =
0 otherwise
Represent the input-output relation using
convolution sum;
Find the response of the filter for the unit step
signal u0(n).
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/13
Example 5.2
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/14
Cascade connection of LTI systems
(線形時不変システムの縦続接続）
x(n) y (n)
h1 ( n) h2 ( n)
x(n) y (n)
h2 ( n) h1 ( n)
x(n) y (n)
h1 ( n) * h2 ( n)
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/15
Parallel connection of LTI systems
(線形時不変システムの並列接続）
h1 ( n)
x(n) y (n)
h2 ( n)
x(n) y (n)
h1 ( n) + h2 ( n)
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/16
Stable systems (安定的システム）
• A system is stable iff
∞
∑ | h(k ) |< ∞
k = −∞
• This is the so called bounded-input-bounded output
(BIBO) stability condition (有界入力有界出力）.
• The system can give meaningful output only if it is stable.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/17
Example 5.3 p. 74
• Try to discuss the stability of the system
given in Example 5.2.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/18
Causal systems (因果的システム）
• A causal system is a physically realizable (物理的に
実現可能な）system (not for off-line processing).
• An LTI system is causal if and only if
(iff: 必要十分条件）)
h( n) = 0 for n < 0
• A signal is called causal iff x(n)=0 for n<0.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/19
Causal systems
• A system is causal iff the current output y(n) is
calculated based only on already observed
data x(n), x(n-1), x(n-2), …
∞
y (n) = ∑ x ( k )h ( n − k )
k = −∞
∞ ∞
= ∑ x ( n − k )h ( k ) = ∑ x ( n − k )h ( k )
k = −∞ k =0
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/20
Homework-1
• Suppose that the impulse response of a
digital filter is given by
h( n ) = a u0 ( n ) n
• Discuss the stability and causality of the
filter.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/21
Homework-2
• Suppose that the impulse response of a
digital filter is given by
nα
h( n ) = e u0 ( n )
• find the output y(n) of the filter using
convolution sum when the input x(n) is
defined by x(0)=1, x(1)=1, x(2)=1, x(3)=1,
and x(n)=0, for n>3.
Produced by Qiangfu Zhao (since 1995), All rights reserved © DSP-Lec02/22
Quiz and self-evaluation
T1
• For the system given in 1.2
1
homework-2, tell if it is 0.8
causal or not. T6 0.6 T2
0.4
0.2
• For the same system,
give the condition for the T5 T3
system to be stable.
T4
Name: Student ID: .
