Description & Analysis of
Systems
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Introduction
A system can be defined as anything that responds when
stimulated or excited.
A system can be an electrical system, a mechanical system,
a biological system,, an economic system, etc.
There exist artificial systems (i.e., designed by engineers)
and natural systems.
Some can be easily analyzed using mathematics, others are
extremely difficult to analyze thoroughly.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
In this chapter, we
1.
introduce nomenclature which describes important
system characteristics
2.
develop techniques to classify systems according to their
characteristics
3.
develop methods of finding responses to arbitrary inputs
of linear time-invariant systems.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Block Diagram and System Terminology
A simple system of one input and one output, represented as
a block diagram, is displayed below
An example of a system would be a boat steered by a rudder
Outputs: boat
heading and
speed.
Inputs: propeller
thrust, rudder
position and water
current.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Another example of a system is an automobile suspension
Inputs: surface topology
of the road.
F06/F07/F08/F09/F10/S13/S14/S15
Outputs: position of
the chassis relative
to the road.
ELEG315: Description & Analysis of Systems
The process of describing and analyzing a system without
building it is called modeling.
Modeling is important in designing large and expensive systems.
Systems can be classified in different ways:
open-loop and closed-loop systems.
continuous-time and discrete-time systems.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Systems Characteristics
Consider a very simple CT system: an RC circuit.
Assume the system has no stored energy before t = 0.
The differential equation that describes the circuit (refer to
Circuits I & II) is
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
It is obtained by using fundamental characteristics of a resistor
and capacitor
Mathematical tools allow us to solve the D.E. and obtain the
solution vout(t) for an arbitrary input vin(t).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Now, consider the following DT system
Mathematically, the output is related to the input by the
difference equation
Once again, mathematical tools allow us to easily find the
solution y[n] for an arbitrary input signal x[n].
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
Homogeneity:Inahomogeneoussystem,multiplyingtheexcitation
byany(complex)constantmultipliestheresponsebythesame
constant,i.e.,
or
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
10
Additivity(Superposition):Ifanarbitraryinputofasystemx1 producesa
responsey1 ,anarbitraryinputx2 producesaresponsey2,andthe
inputx1+x2 alwaysproducestheresponsey1 +y2 thesystemissaid
tobeadditive.
or
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
11
Linearity:Asystemislinear ifitishomogeneousand additive.,i.e.,
wherey1 istheoutputduetox1 and y2 istheoutputduetox2.
Example: Let a system be described by y(t) = x(2t). Determine
whether it is a linear system or not.
Solution: We see that an arbitrary input x1(t) produces x1(2t) = y1(t)
and x1(t) (same as x1(t) but with different amplitude) produces x1(2t) = y1(t).
Similarly, an input x2(t) produces x2(2t) = y2(t) and x2(t) produces
x2(2t) = y2(t).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
12
Now, if the input is
the output must be x(2t)
which is exactly,
Therefore, we conclude that the system is a linear system.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
13
Time Invariance: Ifanarbitraryexcitationofasystemx(t) (x[n])
causesaresponsey(t) (y[n]) andanexcitationofthesystem
x(t-t0) (x[n-n0]) causesaresponsey(t-t0) (y[n-n0]) foran
arbitrary t0 (n0), then,thesystemissaidtobetimeinvariant.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
14
An example of a system which is not time-invariant is y[n] = x[2n].
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
15
Causality:Anysystemforwhichtheresponseoccursonlyduringor
afterthetimeinwhichtheexcitationisappliediscalledacausal
system.
Note: Allphysicalsystemsarecausal becausetheyareunableto
anticipateanexcitationthatwillbeappliedlater.
Asystemiscausaliftheoutputdependsonlyonpresentandpast
[Link]
Asystemisnoncausaliftheoutputdependsonfuture valuesofthe
[Link]
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
16
Memory:Ifanysystemsresponseatanarbitrarytimet0 depends
onlyontheinputatthattime,andnotonthevalueoftheinputor
outputatanyothertime,thesystemhasnomemory.
An example of a system with memory
An example of a memoryless system
Stability:Asystemissaidtobeboundedinputboundedoutput
(BIBO)stableifandonlyifeveryboundedinput(i.e., |x(t)|<
forallt or |x[n]|< foralln) resultsinboundedoutput.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
17
An example of a BIBO stable system is
An example of a BIBO unstable system is
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
18
LTI Systems: The Convolution Sum
One way to determine a system response is to solve the
differential /difference equation that describes it.
An alternative way, which uses the linearity & time-invariance
properties of the system, uses the convolution integral or sum.
First, let us note that the response of a linear system to an input
is given by
where y1(t) is the response to x1(t), and y2(t) to x2(t).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
19
More generally, if the input is
the output will be
Now, recall the sifting property of the function
This expression means that any signal x(t) can be expressed
as a continuum of weighted impulses (i.e., a combination of impulses).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
20
Consequently, the output of a linear system is a linear
combination of the responses of the system to shifted impulses.
That is,
where h(t,) denotes the response of the linear system to the
shifted impulse (t-).
If, in addition, the linear system is time-invariant h(t,) does
not depend on but on t- (i.e., origin is not important but the difference is)
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
21
Therefore, we end up with
The above integral is called the convolution integral. It relates the
input & output of a system by means of its impulse response h(t).
Usually, the convolution is represented by
The convolution is commutative
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
22
or, equivalently, (refer to Appendix D, change of variable)
It is also
associative (change of variable & swap of integrals)
and distributive (separation of integrals)
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
23
For a DT system, the procedure is similar and the output is
given by the convolution sum
The commutative, associative & distributive properties also
hold in this case. That is,
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
24
In conclusion, for any LTI system the output is found as
In particular, if x(t) = (t) (or x[n] = [n]), then
Note that, in this case, the output is just the
impulse response of the system. This is because
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
25
Example: Evaluate the response of a DT LTI system whose
impulse response, h[n], and input, x[n], are displayed below.
Solution: The convolution could be evaluated analytically or
graphically , as shown next.
The system output is given by
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
26
For each value of n, we need to multiply h[n-m] & x[m], and
then, sum over all values of m.
For n < -1 or n > 2 , the functions h[n-m] & x[m] do not overlap
their product is 0 the result of the sum would be 0.
Thus, for n < -1 or n > 2, we have y[n] = 0.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
27
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
28
Example: Plot the response of a CT LTI system whose
impulse response and input are given by
x1(t) = A1rect[(t-2.5)/5]
x2(t) = -A1rect[(t-2.5)/5].
Solution: The output y(t) is the convolution of x1(t) & x2(t):
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
29
First, we need to plot x1() and x2(-)
Now, for an arbitrary t, we evaluate the product x1()x2(t-)
and integrate over .
We repeat this step for all possible values of t, as shown next
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
30
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
31
LTI Systems Interconnections
Parallel Connection: Consider 2 LTI systems with impulse
responses h1(t) & h2(t) connected in parallel
The output y(t) can be written as
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
32
Since x() is a common input, we can write
where h(t) = h1(t)+h2(t) is the equivalent system representing
the parallel connection of the 2 systems.
Note: The above result is in fact the distributive property of the
convolution operation
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
33
An identical result hold for discrete-time systems.
Cascade Connection: Consider 2 LTI systems with impulse
responses h1(t) & h2(t) connected in cascade
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
34
The output y(t) can be expressed in terms of z(t) as
Since z() is the output of the 1st system, it is expressed in terms
of x() as
Substituting the last expression in the previous one yields
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
35
Now, we use = - and swap the order of integration
The inner integral is just the convolution of h1(t) with h2(t)
evaluated at t-. That is, if we define h(t) = h1(t)* h2(t), then
Consequently,
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
36
Notes: 1. The order of the systems in a cascade is not important
(commutative property of the convolution). That is,
2. A similar result holds for discrete-time systems in cascade
Note: The above demonstration is in fact the associative property
of the convolution.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
37
Relationship Between LTI System
Properties & Its Impulse Response
Memoeyless LTI Systems: The output of a DT LTI system is
(1)
Expanding the above sum yields
For the system to be memoryless, y[n] must depend only on
x[n] h[n] must verify h[n-k] = 0 for k 0, i.e.,
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
38
Similarly, a CT LTI system is memoryless if
Note:AllmemorylessLTIsystemssimplyperformscalarmultiplicationon
theinput.
Causal LTI Systems: The output of a causal LTI system depends
only on past or present values of the input.
From Eq. (1), we see that a LTI system is causal if its impulse
response verifies
Similarly, for a CT LTI system
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
39
BIBO Stable LTI Systems: Formally, a bounded input verifies
and a bounded output verifies
To relate the stability to the impulse response, we write
Using
the previous expression becomes
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
40
Since the input is bounded
then,
The output is also bounded (BIBO stable), if
Similarly, we can show that for a BIBO CT LTI system, its
impulse response is absolutely integrable
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
41
Step Response
Here, we examine a system response to a unit step (or unit
sequence for a DT system).
The response of a CT system to a unit step
Thus, the response to a unit step is the integral of the response
to a unit impulse (recall that the unit step is the integral of the unit impulse).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
42
In general, the response of a CT system to the integral of any
excitation is the integral of the response to that excitation.
The converse holds true also. If we differentiate the excitation,
we also differentiate the response.
Graphically, the situation can be represented as follows.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
43
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
44
For a DT system, the step response is given by
On the other hand, a unit sequence can be expressed in terms of
unit impulses as
where we used q = n-k as a change of variable.
Therefore, the top equation can be re-written as
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
45
Observe that when m < 0 the term [q] in the inner sum is
always=0 (because [q] only at q=0 which is outside the limits).
In this case, the inner sum is always zero.
Also, observe that for m 0, the inner sum reduces to only one
point (which corresponds to [q]=[0] =1). All the other terms
are 0.
Consequently, the double sum can be written as
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
46
Once again, using a change of variable q = n-m, we obtain
which the equivalent operation of an integral for DT functions.
Therefore, the response at any discrete-time n of an LTI system
excited by a unit sequence is the accumulation of the impulse
response (up to that time).
This relationship holds for any excitation. That is, if the input is
changed to its accumulation, the response also changes to its
accumulation (similarly, for the backward difference, which is the
inverse operation of the accumulation).
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
47
Response To a Sinusoid
We note that a response of a LTI system to a sinusoid is the
same sinusoid multiplied by a complex constant. To show this,
let the input be a complex sinusoid est, the output y(t) is given
by
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
48
Block Diagram Representation of Differential
& Difference Equations
Consider a system characterized by the differential equation
This expression can be re-written as
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
49
The system can thus be represented by
y(t)
y(t)
or
y(t)
y(t)
The bottom block
diagram is the preferred
option because it does
not include differentiators,
which are prone to highfrequency noise
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
50
For a DT realizable system, we have only past values. These
can be obtained using a simple delay in the system.
For example, the difference equation
can be re-written as
and, consequently, may be implemented using
y[n-1]
y[n-2]
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
51
Summary
1.
2.
3.
4.
5.
6.
Asystemwhichisbothhomogeneousandadditiveislinear.
AsystemwhichisbothlinearandtimeinvariantiscalledLTI
system.
AsystemissaidtobeBIBOstableifboundedexcitations
alwaysproduceboundedresponses.
Allrealphysicalsystemsarecausal.
TheresponseofaLTIsystemexcitedbyasinusoidisaalsoa
sinusoidwiththesamefrequency(generally,withdifferent
amplitudeandinitialphase).
Systemsofdifferentphysicalstructuremaybedescribedby
differentialordifferenceequationsofthesamemathematical
form.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
52
Summary
7.
AnLTIsystemresponsetoanarbitraryexcitationorinputcanbe
foundbyconvolvingtheinputwiththesystemsimpulse
response.
8. TheimpulseresponseofacascadeconnectionofLTIsystemsis
theconvolutionoftheindividualimpulseresponses.
9. TheimpulseresponseofaparallelconnectionofLTIsystemsis
thesumoftheindividualimpulseresponses.
10. AnLTIsystemisBIBOstableifitsimpulseresponseisabsolutely
summableorabsolutelyintegrable.
11. LTIsystemscanberepresentedusingblockdiagram,whichare
usefulforanalysis.
F06/F07/F08/F09/F10/S13/S14/S15
ELEG315: Description & Analysis of Systems
53