TYPE OF SIGNALS AND SYSTEMS

 Definitions of Signals
 A signal is a function of time representing a physical variable, e.g.
voltage, current, spring displacement, share market prices, number
of student asleep in the Lab, cash in the bank account.

 Typically we will use a mathematical function such as f(t), u(t) or y(t)

to describe a signal which is a continuous function of time.

 A continuous-time (or analog) signal exists at all instants of time.

The real word consists of continuous signals, and are usually
written as a function of t.

 A discrete-time signal exists only at discrete instants of time and is

usually derived from a continuous signal by the process of
sampling, e.g. measuring the temperature at 3 o'clock in the
afternoon.
 TYPE OF SIGNALS AND SYSTEMS

 A discrete-time signal is usually (not always) obtained from

sampling a continuous signal at a regular time period known as the
sampling period, which we will represent by the parameter T.

 Thus the discrete-time signal is equal in amplitude to the

continuous signal at the sampling instants.

f k   f kT
d c
(1)

 A discrete-time signal is essentially a sequence of numbers (0,1,2,

...) where each of those numbers represents the amplitude of a
continuous-time signal at a time equal to kT.

 Note that often we neglect to put in the sampling period and write
the signal simple as a function of the sample number k.
 TYPE OF SIGNALS AND SYSTEMS

 A continuous signal is continuous in amplitude. This is continuity in

t   f t   f t 
the traditional mathematical sense, i.e.
 
f (2)
i.e. the amplitude of the signal (function) is continuous.

 A discrete signal is discrete in amplitude, i.e. it only has discrete

levels
 A discrete signal is both discrete in amplitude and in time.
 A periodic signal repeats in time, i.e.

xt  P   xt  (3)

where P is known as the period.
 Note that in the case of discrete-time signal it is usual to describe
the period in terms of number of samples before the signal repeats.
 TYPE OF SIGNALS AND SYSTEMS

 A deterministic signal is known for all time and can be predicted

in advance exactly, e.g. a sinewave with known phase.

 A random signal cannot be predicted exactly, e.g. weather

temperatures.

 Random signals are usually dealt with by statistical rather than

analytical techniques.

 Noise is simply a signal we don't want.

 Sometime it is completely random, but on occasions it can be

someone else's signal, such as cross-talk on a telephone line. So,
since we don't want it we call it noise.
 TYPE OF SIGNALS AND SYSTEMS

 Example of converting between a continuous and discrete signal.

Consider the continuous-time signal (which also happens to be
continuous in amplitude) such as

x t   e
t
c (4)
The sampling interval T = 0.1 seconds. Start sampling at t = 0
  0
xd e  1.0000
0 

  1* T
xd e  0.9048
1 

   2* T
xd e  0.8187
2  (5)
   3* T
xd e  0.7408
3 

 
xd e e
k 
 k* T

0.1* k
 TYPE OF SIGNALS AND SYSTEMS

 We can also write


xd e
k  
 k* T

x k   e 
(6)
T
k

x k   0.9048 
k
d

Either method gives xd k   1.0000,0.9048,0.8187,0.7408, 

i.e. a sequence of number, known as the discrete-time signal.
 Note that the values of this sequence depend on the sampling
period.

 If the sampling interval T were doubled then x k   1.0000,0.8187, 

d

 Thus a single continuous function can yield an infinite number of

sampled sequences, depending on the value chosen for T.
 TYPE OF SIGNALS AND SYSTEMS
 Definition of System
 A system is mathematical model or abstraction of a physical that
relates inputs of the physical process to the outputs. Output and
inputs are defined in terms of signals.

 In general we will use u(t) to represent the input to a system and

y(t) to represent its output, although there will be exceptions to this.

 The system, which we will write S[], relates the input function u(t)
to the output function y(t)
y( t )  S [ u( t )] (7)
where S is some mathematical function.

 For example if S   
2
then y t   u t 
2

 In general systems can also have many inputs and outputs but we
will concentrate on the Single Input Single Output (SISO) system
 TYPE OF SIGNALS AND SYSTEMS
 Linear Systems
 Linear systems possess the property of Additivity or Superposition,
if
1
y u 
t   S 1 t  and 2 t   S 2 2
y u  (8)

then y t   y t   S u t   u 2
1 2 1 2
(9)

 Along with superposition of homogeneity. This implies that if

   
y t S u t ,
  
S u t  S u t  y t    
(10)

 Thus if we double the input signal, we double the output signal

level.

 Linear systems satisfy both additivity and homogeneity. If we

combine (9) and (10)we can test for both simultaneously,

 y t    y t   S  u t    u t 
1 2 1 2
(11)
 TYPE OF SIGNALS AND SYSTEMS

 For example : yt   ut  fails both homogeneity, since

2

ut   ut   yt 

2 2 2
(12)

and additivity since if

  
y1 t   u1 t
2
and y2 t   u 2t  2
(13)

then the condition required for additivity

u1t u2t  2
 u1t  u 2t   2u t u t   y t   y t 
2

2
1 2 1 2
(14)

does not hold.

 TYPE OF SIGNALS AND SYSTEMS

 TIME INVARIANT SYSTEMS

A time invariant system has parameters which are independent of
time.
Expressing this mathematically, if

y t   S u t  (15)
then
yt     S u t    (16)
for a time variant system.

Thus any system is involving switches or having a time varying

parameters is not time invariant. A simple example of a time varying
system is a system where the gain is equal to the time t.

As an example consider a system where

S u t   tu t  (17)
where the input is a pulse.
 TYPE OF SIGNALS AND SYSTEMS

The diagram below show how that delaying the pulse causes the output
to not only be delayed but also changed in shape and amplitude. As a
consequence the system is not time invariant.
u (t)
y(t)

S[]
u (t - )

We will nearly always consider time invariant systems. In reality nearly all
systems are time varying when one considers a sufficiently long period,
but can often be approximated over the period of interest by a time-
invariant system.

Note that when dealing with a discrete signal the sample number k is a
function of time since t  kT where T is the sampling period. Thus
the discrete system is also time varying.

yk   kuk  (18)

 TYPE OF SIGNALS AND SYSTEMS

 LINEAR TIME INVARIANT (LTI) SYSTEMS

 If a system is both linear and time invariant then we call it a linear

time invariant system. Most of the techniques for analysing systems
in this study rely on the system being (LTI) or a close
approximation.

 In practice analysis based on assuming a LTI is very powerful,

because
- it is good approximation to most real word systems
- it is easy to analyse and obtain results
- it is easy to predict its behaviour
 TYPE OF SIGNALS AND SYSTEMS

 CAUSAL SYSTEM
 No real word system can have an output which precedes the input,
e.g.
  
y t  x t  14 is a causal system.

yt   xt  14  is an anti-causal system

because the value of y(t) at time t = 0 depends on the value of x(t)

at time t = 14, i.e. at some time in the future.

 A bank account is an example of a causal system since before the

system gives you an output (interest) you have to insert an input
(i.e. deposit). Banks do not give interest on money that you have
yet to deposit and thus are firm believers in causality.
 TYPE OF SIGNALS AND SYSTEMS

 RESISTOR-CAPACITOR NETWORK
R
i(t)

+vo C e(t)

 i(t) is input signal, a current varying with time, e(t) is output, a

voltage varying with time. We changed notation of input and output
from u(t) and y(t) into i(t) and e(t) respectively.

 Both i(t) and e(t) are examples of signals and could be sinewaves,
square waves, exponential, etc. The system is the resistor and
capacitor and wires connecting them.
 TYPE OF SIGNALS AND SYSTEMS

 In reality, over the years the resistance and capacitance may

change slightly and strictly speaking the system is time varying.

 In practice, we are interested in a much shorter time span and

during that time the resistance and capacitance will not change
significantly.

 It is therefore a reasonable approximation to say the system is time

invariant. t
et   Rit    i t dt '  v0
1
C0
t
e1 t   Ri1 t   i1 t dt '  v0
1
C0
t
e2 t   Ri2 t   i2 t dt '  v0
1
C0 
 TYPE OF SIGNALS AND SYSTEMS

 Now input : it   i1 t   i2 t 

    
t
et   Ri1 t   i 2t  
1
i1 t '  i 2 t ' dt '  v0
C0
et   e1 t   e2 t   V0

 Thus strictly speaking the system is only linear if v0 is equal to

zero.
 SIGNAL POWER
 ENERGY and POWER in a SIGNAL
 In electrical systems the signal is usually a voltage or current. Most
other signals are converted to a voltage or current through a
transducer.

 The energy dissipated by a resistance R during a given time

interval between t1 and t 2 is,

V 2 t 
t2

E
t
R
dt
1
unit E is Joules, if V is in Volts and R is in Ohms.

 Some signals go on (practically) forever and so the integral in

above equation becomes infinite, as an example sin(t) is a function
of infinite energy.
 SIGNAL POWER
 In order to deal with signals such as this we use the concept of
power of a signal. When voltage V(t) is applied to a resistance its
power is given by
V 2 t 
t2
1
W
t1  t 2 t R dt
1

where W has been used to represent Watt and to avoid confusing

with using P which indicates the period of a signal.

 In signal processing theory it is common to use the term "Energy"

to refer to the quantity t2

E  x 2 t dt

t1
and "Power" to refer to the quantity
t2

x 2 t dt
1
W
t1  t 2 t 
1
 SIGNAL POWER
 If x(t) was a voltage we would need to divide by R to obtain the true
energy (or power), whilst if x(t) was a current we would need to
multiply by R.

 In practice, the R term is a constant can be safely left until the end,
whereupon if x(t) is actually a voltage or a current we can introduce
it to obtain the real energy or power.

 If x(t) is a share price, thus the above equations are common make
sense, with reckless disregard for wether we should multiply, divide
by or ignore the resistance.

 Thus if  1 

lim   x t dt   
2

  2  
the signal is said to be of finite power.
 SIGNAL POWER
 Power in a Sinwave
 2t 
y t   A sin  
 P 
where P is period of the sinwave.
The period can be taken from t1   P / 2 to t2  P / 2 or
t1  0 t2  P
to

 Usually we take the former  as it 2t 

often 2
makes the mathematics
P/2

easier. W
1
P / 2   P / 2   
A sin   dt
P / 2
  P 
  4t  
P/2
A2
W 1  cos
2P P / 2   P 
 dt

 4t 
P/2
2  P
W
A
t 
 4  P 
sin
 
2P
 P / 2
2
A
W
2P
 SIGNAL POWER
Note the following points :
- The power is not a function of the period
- The power is not a function of the phase
This is easy to show by redoing the calculation for
yt   A sin 2t / P   
- The power (in the signal theory sense) is the square of the
RMS value of a sinewave
 SIGNAL POWER
 The Decibel

 The ITU-T define the decibel in CCIR Rec 574 as follows :

 The bel (symbol B) expresses the ratio of two powers by the

decimal logarithmic of this ratio. This unit is not often used, having
been replaced by decibel (dB) which is one-tenth of a bel.

 The decibel may be used to express the ratio of two field

quantities, such as voltage, current, sound pressure, electric field,
charge velocity or density, the square of which linear system is
proportional to power.

 To obtain the same numerical value as a power ratio, the logarithm

of the field quantity ratio is multiplied by the factor 20, assuming
that the impedances are equal.
 SIGNAL POWER
 The relationship between a current or voltage ratio and that of the
corresponding power ratio is impedance dependent.

 Use of the decibel when the impedances are not equal is not
appropriate unless adequate information is given concerning the
impedances involved.

 For example, if P1 and P2 are two powers, the ratio expressed in

decibels is :
P 
10 log  1 
 P2 
 If P1 and P2 represent the powers dissipated by current I1 and I2 in
resistances R1 and R2. In this case, the logarithm of this ratio must
be
 P1   I12 R1   I1   R1 
10 log    10 log  2   20 log    10 log  
 P2   I 2 R2   I2   R2 
 SIGNAL POWER
 The important point to note is that the decibel expresses a ratio of
two powers. It is not a unit of power.

 When a value of power is quoted in logarithmic terms, the

reference power must also be quoted.

 For example, if the reference power is 1 miliwatt (mW), power

values are often quoted in dBm, where the suffix "m" means
referred to a power of 1 mW. Thus, "8 dBm" means the ratio of the
measured power to 1 mW is 8 dB.