EEE304

Week 1: Review of Signals and Systems

Fundamental Concepts

H
X Y
EEE304
Lecture 1.1a: Review of basic signals
and their properties: Steps and Impulses
Unit Step
1 if t  0
• Unit step: u (t )  
0 otherwise
• The unit step serves as a set indicator, i.e., whether an argument belongs to a set or not.
It is useful in writing compact expressions for “rule-based” functions
 sin t if t  0
• E.g.., x(t )    sin t u (t )  cos(2t ) (1  u (t ))
cos 2t otherwise

• A similar definition in Discrete-time signals (sequences)

1 if n  0
• Unit step: u ( n)  
0 otherwise
• Observation: In CT, u(t ), 1  u(t ) are almost the same (they differ at single
point). But in DT, u(n), 1  u(n) are substantially different.
Unit Impulse
du (t )
• Unit delta, or (Dirac) impulse:  (t ) 
dt
• The unit impulse can be viewed as the limit of the derivative of a continuous
approximation of the unit step; one possible choice is un (t )  max[0, min[1, nt  1/ 2]]

du1 (t )
u1 (t ) 1 (t ) 
dt
…

…
du5 (t )
u5 (t )  5 (t ) 
dt
…

…
dfn
u(t )  lim n un (t )  (t )  lim n  n (t )
 Some common applications of the impulse

• Key properties of the impulse (sampling) x( ) ( )  x(0) ( ),   ( )d  1


Product with a continuous function : cos( ) ( )  cos(0) ( )   ( )

  
Product and integratio n :  cos( ) (  1)d   cos(1) (  1)d  cos(1)   (  1)d  cos(1)
  

u (0  )  u (0  ) 1
Product with a discontinu ous function : u (t ) (t )   (t )   (t ), ( is even)
2 2
  
Product and integratio n :  u ( ) (  1)d   u (1) (  1)d  1   (  1)d  1
  
   
u (0 )  u (0 ) 1
Product and integratio n around a discontinu ity :  u (  1) (  1)d    (  1)d 
  2 2
 
 1
Integration with the impulse on the boundary :  u ( ) ( )d   u (0 ) ( )d 
0 0 2
Some common applications of the step
• Using steps to alter the interval of integration
 
Lower limit of integratio n and forward steps :  x( )u (  1)d   x( )d
 1
 1
Upper limit of integratio n and backward steps :  x( )u (1   )d   x( )d
 
 min(1,t )
Upper limit of integratio n and backward steps :  x( )u (t   )u (1   )d   x( )d
 

t 
Upper and lower limit of integratio n and steps :  x( )u (t   )u (  1)d   x( )d  u(t - 1)
 1 
 Signal parametrization
• The response of an LTI system to an impulse characterizes its response to
arbitrary signals. At the heart of this result is the so-called “signal
parametrization” in terms of shifted impulses
  

 x( ) (t   )d   x(t ) (t   )d  x(t )   (t   )d  x(t ) 1  x(t )

  

Shifted Approximate
impulses ramp
(with offset) d0*0+d1*1+
d0, d1, d2, d2*2+d3*3+
… d4*4+…
Discrete-time development
1 if n  0
• Unit delta, or (Kronecker) impulse:  (n)  u (n)  u (n  1)  
0 otherwise
(an ordinary sequence)
• Signal parametrization in terms of shifted impulses

 x(k ) (n  k )   x(n) (n  k )  x(n) (n  k )  x(n) 1  x(n)

k k k

• These CT/DT identities lead to the convolution integral and convolution

sum, describing the output of LTI systems in terms of the input and a
characteristic of the system, the “impulse response.”
EEE304
Lecture 1.1b: Review of basic systems
and their properties: Impulse response
Impulses and LTI systems (CT)
• For the parametrization of the output of an LTI system in terms of shifted
impulse responses (convolution integral) we operate with the system on x:
    
H [ x(t )]  H   x( ) (t   )d    H [ x( ) (t   )]d   x( )H [ (t   )]d   x( )h(t   )d  y(t )
     

1
H ( s) 
s 1

Adding the signals to produce a “pulse” and its response

1
H ( s) 
s 1
Computation details (CT)
It is a good practice to derive here the formulae of the various responses
mentioned in the last example:
 t
(t  ) (t  ) ( t )  t
H [u (t )]   u ( )e u (t   )d   e d u (t )  e e u (t  0)  (1  et )u (t )
0
 0

t  (t 1)
H [ pulse(t )]  H [u(t )  u(t  1)]  (1  e )u(t )  (1  e )u(t  1)

Now, adding the five one-second pulse responses to obtain the five-second
pulse response is a tedious but straightforward exercise (the
intermediate terms cancel out).
Impulses and LTI systems (DT)
• The discrete-time analog also starts with the input parametrization:
x(n)   x(k ) (n  k )
k

Then, operating with the system on both expressions of the input signal
 
H [ x(n)]  H  x(k ) (n  k )   x(k )H [ (n  k )]   x(k )h(n  k )  y(n)
k  k k

• The last expression is the parametrization of the output of linear systems

in terms of shifted impulse responses (convolution sum)
Computation Details (DT)
• Performing the computations for the analogous pulse example in DT:
 ( n1) ( n1)
 n 1   1  
H [u (n)]   u (k )nk u (n  k )   nk u (n  0)  n 1
u (n  0)  u ( n)
k  k 0 1  1 

• Here, we recall the well-known computation for geometric series:

n
Sn           ...  
k 0 1 2 n

k 0
n
Sn            ...  
k 1 2 3 n 1

k 0

(1   ) S n  0  n1
n 1
1 
Sn 
(1   )
 Computational Details (DT)
• Thus,
( n1)
1  1  ( n)
 ( n)
H [ pulse(n)]  H [u (n)]  H [u (n  1)]  u ( n)  u (n  1) 
1
 ( n)  u(n)  u(n  1) 
1  1  1  1 
(n)
(n)

1
 ( n)  u(n)  u(n  1)  1
 ( n)  (  1)u(n)   (n)   nu(n)
1  1  1  1 

• This is simply the impulse response since u(n) – u(n-1) = (n)

 
H [ pulse2 (n)]  H [u(n)]  H [u(n  2)]  ...  nu(n)  n1u(n  1)   (n)  n  n1 u(n  1)

z
H ( z) 
z  0.5
MATLAB sample code for the examples
t=[-10:.01:10]; % define a time vector
n=1; % pick n for the step approximation
u0=max(0,min(1,n*t+1/2)); % define shifted steps
u1=max(0,min(1,n*(t-1)+1/2));
u2=max(0,min(1,n*(t-2)+1/2));
u3=max(0,min(1,n*(t-3)+1/2));
u4=max(0,min(1,n*(t-4)+1/2));
d1=([diff(u1) 0]*100); % compute derivatives (approx)
d0=([diff(u0) 0]*100);
d2=([diff(u2) 0]*100);
d3=([diff(u3) 0]*100);
d4=([diff(u4) 0]*100);
plot(t,[d0 ;d1-2 ;d2-4 ;d3-6; d4-8]); pause % plot results

H=tf(1,[1 1]) % define a system by its transfer function

y0=lsim(H,d0,t); % compute the system response to input d0
y1=lsim(H,d1,t);
y2=lsim(H,d2,t);
y3=lsim(H,d3,t);
y4=lsim(H,d4,t);
plot(t,[y0 y1-2 y2-4 y3-6 y4-8]); pause % plot results
MATLAB sample code for the examples
N=[-10:10]; % define a time vector
n=10; % pick n for the step approximation
u0=max(0,min(1,n*N+1)); % define shifted steps
u1=max(0,min(1,n*(N-1)+1));
u2=max(0,min(1,n*(N-2)+1));
u3=max(0,min(1,n*(N-3)+1));
u4=max(0,min(1,n*(N-4)+1));
d1=([diff(u1) 0]); % compute differences(approx)
d0=([diff(u0) 0]);
d2=([diff(u2) 0]);
d3=([diff(u3) 0]);
d4=([diff(u4) 0]);
plot(N,[d0 ;d1-2 ;d2-4 ;d3-6; d4-8],'o'); pause % plot results

H=tf([1 0],[1 -.5],1) % define a system by its transfer function

y0=lsim(H,d0,N); % compute the system response to input d0
y1=lsim(H,d1,N);
y2=lsim(H,d2,N);
y3=lsim(H,d3,N);
y4=lsim(H,d4,N);
plot(N,[y0 y1-2 y2-4 y3-6 y4-8],'o'); pause % plot results
EEE304
Lecture 1.2: Review of basic signals and
their properties: Exponentials
Exponentials
 t  jt  t jt t
• CT Exponential: e  e
st
e e  e (cost  j sin t )

• When  = 0, the exponential is periodic with period 2/

• When  = 0, the exponential has magnitude 1:
jt
e  cos t  sin t  1
2 2

n jn
• DT Exponential: e sTn
z  e
n
  (cosn  j sin n)
n

2
• When  = 1, the exponential is periodic with period N if and only if 
N
• When  = 1, the exponential has magnitude 1.
• Note:  in rad/sec,   T in rad/sec  sec/sample  rad/sample
Exponentials and LTI systems (CT)
• Parametrization of the output of LTI systems in terms of exponentials
(transfer function)
   
H [ x(t )  e st ]   x( )h(t   )d   x(t   )h( )d   e s (t  ) h( )d  e st  es h( )d  H ( s)e st  y (t )
   

st st
e 1 H ( s )e e( 0.1 j )t
e ( 0.1 j )t H ( s) 
s 1 (0.1  j )  1

Here, the Laplace transform of the impulse response (h(t)), also known as the
transfer function, arises naturally as the coefficient of the output exponential.

I.e., in terms of exponentials, LTI systems are described by multipliers, while in

terms of shifted impulses they are described by convolution integrals
Frequency-domain response of LTI systems
The corresponding parametrization of the output of LTI systems in the
frequency domain (complex exponentials) uses the inverse Laplace
transform
c  j
1
x(t )   
st Bromwich integral,contour in ROC
X ( s ) e ds
2j c j
 1 c  j  c  j
H x(t )  H 
1
 X ( s)e ds    X ( s )H [e ]ds 
st st

 2j c j  2j c j
c  j c  j
1 1
    ds  Y ( s )  H ( s ) X ( s )
st st
X ( s ) H ( s )e ds Y ( s )e
2j c j 2j c j

Note: H[x(t)] is bad notation. Writing it, we simply mean that this is the response of the system H with input x, a time
function. The response, or output, is itself a time function. A more precise and unambiguous notation would be
y = H[x] and its value at time t would be y(t) = H[x](t).
Discrete-time LTI systems in the
frequency domain
• Parametrization of the output of linear systems in terms of exponentials
(transfer function)
H [ x ( n)  z ]   x ( k ) h( n  k )   x ( n  k ) h( k )   z
n n k
h( k )  z n
z k
h( k )  z H ( z )  y ( n)
n

k k k k

Similarly, the output in the discrete-time frequency domain is

Y ( z)  H ( z) X ( z)
Again, the Z -transform of the impulse response (h(n)), also known as the transfer
function, arises naturally as the coefficient of the output exponential.

• Thus, in terms of exponentials, LTI systems are described by multipliers,

both in CT and DT. When the exponentials are sinusoids, the description is
related to the Fourier “frequency response” of the system.
Exponentials and LTI systems
• A more familiar look:
• In continuous-time, the cosine is expressed in terms of complex
exponentials and the response to a cosine can be found as a simple
linear combination of responses to exponentials
H [ x(t )  cost ]  H e jt e jt
[ 2 ]
1
2
 jt
H ( j )e  H ( j )e  jt

 ... | H ( j ) | cost  H ( j )

• Here, the derivation makes use of properties of the transfer function

of real systems to express the output as a cosine of different
amplitude and phase.
• Analogous statements can be made for discrete-time systems
H [ x(n)  cosn]  H [ 2 ]
e jn e jn 1
2
  
H (e jn )e jn  H (e jn )e jn  ... | H (e jn ) | cos n  H (e jn ) 
MATLAB sample code for the examples
X=exp(-0.1*t).*exp(j*t); % define a complex exponential
plot3(t,imag(X),real(X)); pause % plot it in 3D (Re, Im, time)
s=-0.1+j; % define the complex frequency
Hs=1/(s+1); % evaluate the transfer function
plot3(t,imag(Hs*X),real(Hs*X),'r');pause % plot the 3D output
EEE304
Lecture 1.3: Review of Basic Systems
and their properties
Systems: Basic definitions
• Systems are maps from a space of functions to another space of functions

H
X Y

• In continuous time, typical examples of the spaces X, Y are the spaces of

piece-wise continuous functions, energy functions, bounded functions. In
discrete time, typical spaces are bounded sequences, etc.
• We write H : X  Y , y  H [ x], y(t )  H [ x](t ), y(n)  H [ x](n)
• For LTI, we also have y(t )  (h * x)(t ), Y (s)  H (s) X (s), Y ( z)  H ( z) X ( z),
System properties

• We say that a system is

• Linear, if H [ax1  bx2 ]  aH [ x1]  bH [ x2 ], x1, x2 , a, b

• Time-Invariant, if H T  TH ,  , where T : x(t )  T [ x](t )  x(t   ) is the

“shift operator” or “time-delay system”
• Memoryless, if the value of y (t ) depends only on the value of x(t )

• Causal, if the value of y (t ) depends only on the values of x( ),   t

• Stable, if bounded inputs produce bounded outputs.

x : { B  0 :| x(t ) | B, t}  y  H [ x] : { C  0 :| y(t ) | C, t}
System properties

• Analyzing system properties from the definition:

• Linearity: H [ x1  x2 ]  H [ x1 ]  H [ x2 ]; H [ax]  aH [ x]

• Time-Invariance: H [ x] T H [ x]
H T
T [ x] H T [ x]
=?

T H
• Stability: Bounding of system output with a sequence of inequalities,
e.g.,
   
| y (t ) |  x( )h(t   )d   | x( ) || h(t   ) | d   Bx | h(t   ) | d  Bx  | h(t   ) | d  Bx (H )
   
System properties: Examples I
• Some examples of systems and the properties they satisfy are:

• y(t )   x( )h(t   )d Linear, Time-Invariant (LTI)


• y(t )   x( )u (t   )d LTI, Causal, Memory, Unstable



• y(t )   x( )u (t   )d L, TV, Not-Causal, Memory, Unstable


t 
• y(t )   x( )e u (t   )d LTI, Not-Causal (Anti-Causal), Memory, Stable

System properties: Examples II
• Some examples of systems and the properties they satisfy are:
• y(t )  dx(t ) / dt LTI, Causal, Memory, Unstable

• y(t )  x(at ), a  1 LTV, Not-Causal, Memory, Stable

• y(t )  sin(5t ) x(t ) LTV, Causal, Memoryless, Stable

• y(t )  x(t  t0 ), t0  0 LTI, Causal, Memory, Stable

• y(t )  x(t  t0 ), t0  0 LTI, Not-Causal (Anti-Causal), Memory, Stable

System properties: Examples III
• More examples of systems and the properties they satisfy are:
• y(t )  sin(5x(t )) Non-Linear, TI, Causal, Memoryless, Stable

• y(t )  sin(5x(t  1)) Nonlinear, TI, Not-Causal, Memory, Stable

• y(t )  sin(5t  1) x(t ) LTV, Causal, Memoryless, Stable

• y(t )  et1x(t ) LTV, Causal, Memoryless, Unstable

• y(t )  e x (t 1) NL, TI, Causal, Memory, Stable

EEE304
Lecture 1.3 Addendum: Derivations of
the system properties
System properties: Examples I
• Some examples of systems and the properties they satisfy are:

• y(t )   x( )h(t   )d Linear, Time-Invariant (LTI)


• y(t )   x( )u (t   )d LTI, Causal, Memory, Unstable



• y(t )   x( )u (t   )d L, TV, Not-Causal, Memory, Unstable


t 
• y(t )   x( )e u (t   )d LTI, Not-Causal (Anti-Causal), Memory, Stable


H [ x] T H [ x]
H T
T [ x] H T [ x]
T H
System properties: Examples I
• Some examples of systems and the properties they satisfy are:

• y(t )   x( )h(t   )d Linear, Time-Invariant (LTI)


• y(t )   x( )u (t   )d LTI, Causal, Memory, Unstable



• y(t )   x( )u (t   )d L, TV, Not-Causal, Memory, Unstable


t 
• y(t )   x( )e u (t   )d LTI, Not-Causal (Anti-Causal), Memory, Stable

System properties: Examples I
• Some examples of systems and the properties they satisfy are:

• y(t )   x( )h(t   )d Linear, Time-Invariant (LTI)


• y(t )   x( )u (t   )d LTI, Causal, Memory, Unstable



• y(t )   x( )u (t   )d L, TV, Not-Causal, Memory, Unstable


t 
• y(t )   x( )e u (t   )d LTI, Not-Causal (Anti-Causal), Memory, Stable

System properties: Examples II
• Some examples of systems and the properties they satisfy are:
• y(t )  dx(t ) / dt LTI, Causal, Memory, Unstable

• y(t )  x(at ), a  1 LTV, Not-Causal, Memory, Stable

• y(t )  sin(5t ) x(t ) LTV, Causal, Memoryless, Stable

• y(t )  x(t  t0 ), t0  0 LTI, Causal, Memory, Stable

• y(t )  x(t  t0 ), t0  0 LTI, Not-Causal (Anti-Causal), Memory, Stable

System properties: Examples III
• More examples of systems and the properties they satisfy are:
• y(t )  sin(5x(t )) Non-Linear, TI, Causal, Memoryless, Stable

• y(t )  sin(5x(t  1)) Nonlinear, TI, Not-Causal, Memory, Stable

• y(t )  sin(5t  1) x(t ) LTV, Causal, Memoryless, Stable

• y(t )  et1x(t ) LTV, Causal, Memoryless, Unstable

• y(t )  e x (t 1) NL, TI, Causal, Memory, Stable

EEE304
Lecture 1.4: Properties of LTI systems in
the Time-Domain
 y  H [ x]  h * x


Properties of LTI systems (CT) y (t )   h(t   ) x( )d ; Y ( s)  H ( s) X ( s)


y (n)   h(n  k ) x(k ); Y ( z )  H ( z ) X ( z )
k

• Here, we focus on LTI systems. Using the general parametrization in terms

of their impulse response, we can obtain equivalent conditions for the
remaining properties. For the continuous-time (CT) case,

y  H [ x]  h * x  y (t )   h(t   ) x( )d


• The above LTI system is:

• Causal, if and only if h(t )  0, for t  0 . Alternatively, h(t )  h(t )u(t ).

• Memoryless, if and only if h(t )  k (t ) , for some constant k. Here,  (t ) is the

Dirac delta.

• Stable, if and only if  | h(t ) | d   . In this case, we say that h is “absolutely

integrable.
 y  H [ x]  h * x


Properties of LTI systems (DT) y (t )   h(t   ) x( )d ; Y ( s)  H ( s) X ( s)


y (n)   h(n  k ) x(k ); Y ( z )  H ( z ) X ( z )
k

• Analogous statements are valid for discrete-time (DT) systems


 
y  H [ x]  h * x  y (n)   h(n  k ) x(k )  alt.  h(n  k ) x(k ) 
k   k 
• The above LTI system is:
• Causal, if and only if h(n)  0, for n  0. Alternatively, h(n)  h(n)u(n)

• Memoryless, if and only if h(n)  k (n), for some constant k. Here,  (n) is the
Kronecker delta.

• Stable, if and only if | h(n) |  . In this case, we say that h is “absolutely

n
summable.”
Examples of LTI system properties I
• Determine the properties of an LTI system from its impulse response:
2t
• h(t )  (e )u(t ) Memory, Causal, Stable
2t
• h(t )  (e )u(t ) Memory, Anti-Causal, Unstable
• h(t )  (e2t )u(t ) Memory, Anti-Causal, Stable
• h(t )  (e2t )u(t ) Memory, Causal, Unstable
• h(t )  (17) (t ) Memoryless, Causal, Stable
Examples of LTI system properties II
• Determine the properties of an LTI system from its impulse response:
• h(n)  (3) u(n)
n
Memory, Causal, Unstable
• h(n)  0.3 u(n)
n
Memory, Anti-Causal, Unstable
• h(n)  (0.3)n u(n) Memory, Anti-Causal, Unstable
• h(n)  (0.3)n u(n) Memory, Causal, Stable
• h(n)  (17) (n) Memoryless, Causal, Stable
EEE304
Lecture 1.5a: Properties of CT LTI
systems in the Frequency Domain: ROC
System properties from the transfer function:
Region of Convergence
• The development of equivalent conditions for the properties of LTI
systems from its transfer function (TF) hinges on the concept of the region
of convergence (ROC) of the corresponding transform (L,Z)
• The ROC of the transform is the region where the corresponding integral
or summation converge.

 st
X ( s)   e x(t )dt X ( z )   z  k x(k )
 k

By convergence we mean the formal existence of the limits

N N
X ( s)  lim N ,M  e
 st
x(t )dt X ( z )  lim N ,M  z k
x(k )
M k  M
Region of Convergence of transforms
• The analysis of convergence also makes use of the concept of left- and
right- sided functions and sequences:
• x(t) is right sided if x(t) = x(t)u(t), e.g., a right sided CT signal is

• x(k) is left sided if x(n) = x(n)u(-n), e.g., a left-sided DT signal is

etc.

• Fact: Any signal can be written as a sum of a right-sided and a left-sided one.
x(t )  x (t )  x (t ) x(n)  x (n)  x (n)

The analysis of the ROC of the transforms uses this decomposition

Region of Convergence of transforms (CT)
• It is now straightforward to see that if the Laplace transform of a right-
sided function converges for s  s0 , then it converges for all s with Re s  Re s0
Furthermore, it also converges for all s with Re s  Re s0
• For left-sided signals, the Laplace transform converges for s with Re s  Re s0
Properties of ROC of transforms (CT)
• It follows that right-sided signals have ROC that extends to  and left-
sided signals have ROC that extends to   .
• Since the ROC cannot (by definition) contain singularities, it is limited by
the singularities (poles) of the transform function.
• Combining the left- and right-sided parts of the function, the ROC of the
Laplace transform must have the form of vertical strips, with singularities
on its boundary.
• The singularities to the left of the ROC correspond to right-sided time
functions, and singularities to the right of the ROC correspond to left-sided
time functions.
 Examples of ROC of transforms (CT)
( s  1)(s  2)
• As an example, consider the Laplace transform X ( s) 
( s  3)(s  1)(s  2)
• We sketch its pole-zero plot (poles = ‘x’, zeros = ‘o’)

The possible ROC’s are highlighted with 4 3 2 1

different colors. For example, ROC 1
corresponds to a right-sided function, ROC 4 to
a left-sided, etc. x x o o x
For the ROC 2, say, the pole at 3 should be
inverted as left-sided and the poles at -1, -2 as
right-sided. Thus, writing the PFE of X(s), we
have
( s  1)(s  2) 0.1 2.4  1.5 2t t
X ( s)      x(t )  0.1e u (t )  2.4e u (t )  1.5e u (t )
3t
( s  3)(s  1)(s  2) ( s  3) ( s  2) ( s  1)
 Examples of ROC of transforms (CT)
( s  1)(s  2)
• As an example, consider the Laplace transform X ( s) 
( s  3)(s  1)(s  2)
• We sketch its pole-zero plot (poles = ‘x’, zeros = ‘o’)

4 3 2 1
Another example:
x x o o x
For the ROC 1, all poles should be inverted as
right-sided. Thus, writing the PFE of X(s), we
have (same PFE):

( s  1)(s  2) 0.1 2.4  1.5  2t t

X ( s)      x(t )  0.1e u (t )  2.4e u (t )  1.5e u (t )
3t
( s  3)(s  1)(s  2) ( s  3) ( s  2) ( s  1)
EEE304
Lecture 1.5b: Properties of DT LTI
systems in the Frequency Domain: ROC
Region of Convergence of transforms (DT)
• Analogous statements hold in DT and the Z-transform
X ( z )   z x(k )
k

• Here, the quantity that determines the convergence of the Z transform is

the magnitude of the discrete complex frequency z.
• Right-sided sequences have Z-transforms that converge for | z || z0 | . Their
ROC extends to infinity.
• Left-sided sequences have Z-transforms that converge for | z || z0 | . Their
ROC extends to zero.
Region of Convergence of transforms (DT)
• For general sequences, the ROC of the Z transform is a so-called annular
region, that is limited by the singularities of the function.
• Poles smaller than the inner radius correspond to right-sided sequences,
• Poles larger than the outer radius correspond to left-sided sequences.
• Notice the difference in DT between right- and left-sided sequences:

1
Z {z /( z  a) RS }  a nu (n)

1
But, Z {z /( z  a) LS }  a nu (n  1)
 Examples of ROC of transforms (DT)
( z  1)( z  2)
• As an example, consider the Z transform X ( z) 
( z  3)( z  1)( z  2)
• We sketch its pole-zero plot (poles = ‘x’, zeros = ‘o’)

The possible ROC’s are indicated by the numbers. For

example, ROC 1 corresponds to a right-sided function, ROC 4
to a left-sided, etc. 3 2 1

For the ROC 2, say, the pole at 3 should be inverted as left- 4

x x o o x
sided and the poles at -1, -2 as right-sided. Thus, writing the
PFE of X(z), we have
( z  1)( z  2) 0.1 2.4  1.5
X ( z)    
( z  3)( z  1)( z  2) ( z  3) ( z  2) ( z  1)
n 1 n 1 n 1
 x(n)  0.1(3) u (n)  2.4(2) u (n  1)  1.5(1) u (n  1)
1
Note : Z {z /( z  a) LS }  a u (n  1)
n
 Examples of ROC of transforms (DT)
( z  1)( z  2)
• As an example, consider the Z transform X ( z) 
( z  3)( z  1)( z  2)
• We sketch its pole-zero plot (poles = ‘x’, zeros = ‘o’)

As another example

For the ROC 1, all the poles should be inverted as right-sided. 3 2 1

Thus, writing the same PFE of X(z), we have
4
x x o o x
( z  1)( z  2) 0.1 2.4  1.5
X ( z)    
( z  3)( z  1)( z  2) ( z  3) ( z  2) ( z  1)
 x(n)  0.1(3) n1u (n)  2.4(2)n1u (n  1)  1.5(1) n1u (n  1)
EEE304
Lecture 1.6: Properties of LTI systems in
the Frequency Domain: Examples
 y  H [ x]  h * x


Properties of LTI systems (CT) y (t )   h(t   ) x( )d ; Y ( s)  H ( s) X ( s)


y (n)   h(n  k ) x(k ); Y ( z )  H ( z ) X ( z )
k

• In our framework, LTI systems are associated with an impulse response or

a transfer function:

y  H [ x]  h * x  y (t )   h(t   ) x( )d , Y ( s)  H ( s) X ( s); ROCH ( s )  {s} M


• Consequently, we can ascertain all the system properties from its Transfer
Function (TF), as well as, its Impulse Response (IR)
• Some common problem statements:
o Given an IR, determine the ROC of the corresponding TF
o Given a TF and ROC, determine the corresponding IR
o Given a TF and ROC determine the system properties (without explicitly finding the IR)
o Given a TF determine the ROC that corresponds to certain system properties.
 y  H [ x]  h * x


Properties of LTI systems (CT) y (t )   h(t   ) x( )d ; Y ( s)  H ( s) X ( s)


y (n)   h(n  k ) x(k ); Y ( z )  H ( z ) X ( z )
k

• Thus, the system


y  H [ x]  h * x  y (t )   h(t   ) x( )d , Y ( s)  H ( s) X ( s); ROCH ( s )  {s} M


with transfer function H(s) is:

• Causal, iff* its ROC extends to   . Anti-Causal, iff* its ROC extends to   .

• Memoryless, if and only if H (s)  k, for some constant k.

• Stable, if and only if its ROC contains the j axis. In this case, poles with
negative real parts correspond to right-sided functions and poles with positive
real parts correspond to left-sided functions.
*: iff for rational transfer functions, otherwise  /   in ROC only implies right/left-sided
impulse response.
Examples of LTI system properties
(Frequency Domain, CT)
Determine the transfer function ROC for an LTI system to be stable/causal:
1
• H (s) 
( s  1)(s  2) Causal: ROC= {s : 1  Re s} Stable: ROC= {s : 1  Re s}

1
• H (s) 
( s  1)(s  2) Causal: ROC= {s : 2  Re s} Stable: ROC= {s : Re s  1}

1
• H (s) 
( s )(s  2) Causal: ROC= {s : 2  Re s} Stable: ROC= 

1
• H (s) 
( s  1)(s  2) Causal: ROC= {s : 2  Re s} Stable: ROC= {s : 1  Re s  2}
 y  H [ x]  h * x


Properties of LTI systems (DT) y (t )   h(t   ) x( )d ; Y ( s)  H ( s) X ( s)


y (n)   h(n  k ) x(k ); Y ( z )  H ( z ) X ( z )
k

• Analogous statements are valid for discrete-time (DT) systems


y  H [ x ]  h * x  y ( n)   h(n  k ) x(k ), Y ( z )  H ( z ) X ( z ), ROCH ( z )  | z |  M
k 

• The above LTI system with transfer function H(z) is:

• Causal, iff its ROC extends to . Anti-Causal, iff its ROC extends to 0.

• Memoryless, if and only if H ( z)  k , for some constant k.

• Stable, if and only if its ROC contains the unit circle. In this case, poles with
modulus (magnitude) less than one correspond to right-sided functions and
poles with modulus greater than one correspond to left-sided functions.
Examples of LTI system properties
(Frequency Domain, DT)
Determine the transfer function ROC for an LTI system to be stable/causal:
1
• H ( z) 
( z  0.1)( z  0.2)
Causal: ROC= {z : 0.2 | z |} Stable: ROC= {z : 0.2 | z |}

1
• H ( z) 
( z  1.1)( z  2)
Causal: ROC= {z : 2 | z |} Stable: ROC= {z :| z | 1.1}

1
• H ( z) 
( z  1)( z  2)
Causal: ROC= {z : 2 | z |} Stable: ROC= 

1
• H ( z) 
( z  0.1)( z  2)
Causal: ROC= {z : 2 | z |} Stable: ROC= {z : 0.1 | z | 2}
Notes on system properties and
response computations
• Fourier transform = Laplace evaluated at s = jw, if ROC includes the jw-
axis. E.g., impulse responses of stable systems, energy signals
• Power signals have Fourier transforms containing impulses and their
Laplace transforms are different (at s = jw).
• Increasing exponentials or polynomials do not have a Fourier transform
• Analysis of transient responses is easier with Laplace
• After PFE, poles should be inverted according to their location relative to the
ROC

• Periodic signals, if necessary, should be handled after projection to right sided

and left sided components.
EEE304
Lecture 1.7a: Examples of system
response computations
Notes on system properties and
response computations
• Analysis of periodic responses is easier with Fourier
• FT of periodic signals is composed of impulses (through Fourier Series expansion). Then,
system output is also impulses weighted by the T.F. (sampling property of impulses)

• Analysis of transient signals, would involve the convolution of the function

transforms with, e.g., steps, and as such, it can be easier with Laplace

• Similarly in Discrete-time with Z- and DTFT

• Stability: ROC (of IR) must include the jw axis (CT) or the unit circle (DT). For causal
systems (ROC includes +inf) this implies that all the poles must be in the left half-
plane (CT) or inside the unit circle (DT).
Computation of system responses
Computation of system responses
Computation of system responses
Computation of system responses
EEE304
Lecture 1.7b: Examples of system
response computations
Computation of system responses
Computation of system responses
Computation of system responses
Computation of system responses
MATLAB code for the examples
display('Problem 1')
e=exp(1);
H=1/e*tf(1,[1 1]);
H.iodelay=1
HH=ss(H)-ss(tf(1,[1 2])); % convert to state space for MATLAB to handle delays
[m,p]=bode(HH,1)
x=1/e*(e^-j)/(j+1)-1/(j+2) % alternative computation
abs(x)
angle(x)*180/pi

display('Problem 2')
num=[1 0],den=conv([1 0],conv([1 -1],[1 -2]))
[r,p,k]=residue(num,den)

display('Problem 3')
num=[1 0],den=conv([1 -1],conv([1 -0.1],[1 -0.2])) % factor out one z in num
[r,p,k]=residue(num,den)

display('Problem 4')
H=tf([1 0],[1 -.2],1)*tf(1,[1 -.1],1)
[m,p]=bode(H,2*pi/12)
z=exp(j*2*pi/12) % alternative computation
x=z/(z-.1)/(z-.2)
angle(x)
abs(x)