noc18-ee28 Lecture 49-Review of communication concepts-I

IIT Kanpur July 201
20 followers
1 Video View·Dec 3, 2023
Hello and welcome to NPTEL MOOC on Fiber - Optic Communication Systems and Techniques . In
the previous module, we discussed a generic optical fiber communication system, point
to point optical communication system and then we looked at the modulator structures
ah which can be used to modulate any of the carrier characteristics such as amplitude,
phase and frequency. Although as I have told you in this course,
we will be looking at only those modulations which will change the amplitude and free ah,
amplitude and phase ah in what is called as quadrature amplitude modulation . Ah that
is the modulation technique that we are going to discuss in detail because that is the prevalent
ah technique which is ah used today in all almost all optical fiber communication systems
. Now, if the change of the carrier characteristic
is continuous or is it it is analogous to whatever the change that the information is
undergoing where the information is usually a continuous time signal , then what we have
is the analogue communication system and you can actually perform analogue communication
system or you can implement analogue communication system by biasing the modulator, the intensity
modulator at what is called as the quadrature point that is at point v phi by 2 if you were
to bias and then apply a small signal which is continuous time , then your output intensity
will be kind of faithful reproduction of the message signal ok. Of course, when the message
signal is 0 , the output will be non zero, it will be having some amount of power coming
out because you have bio sated the quadrature point of the modulator and then when any when
the input signal changes, then the output power will also change.
So, that is the analogue modulation and as I have told you analogue modulation is something
that is not quite preferred in today's optical communication systems. In fact, in many of
communication systems today , communication is kind in communication is mostly digital
communication system . In digital communication system, the message signal is not continuous
rather it would be represented by binary numbers that is a sequence of 1's and 0's . And you
can either code you know 1, 0 with a waveform or 1 with a waveform or you can take the groups
of this 0's and 1's and give an appropriate waveform or generate an appropriate waveform
for it . And at the output, you just have to sample the received signal ok. Of course,
when I say you just have to sample, there are lot of operations that need to be taken
place ah in the receiver which we will some of those operations, we are going to talk
about it but the net result of that receiver is that you are going to sample the signal
and by looking at the sample you should be able to guess which of the group of bits that
were transmitted or in case of a single bit being transmitted in terms of a a waveform
mapping, then you are essentially looking at what was actually being transmitted whether
it was a single bit or a group of bits depending on the sample value that you have received
ok. Because information and the reception is not
exactly continuous, it is in the form of bits which are kind of discontinuous or discrete,
this is called as digital modulation and digital modulation is what we are going to study mainly
; however, even though we are using digital modulation, the signal that comes out of the
optical transmitter will still be a waveform. Of course, it will be waveform in the form
of electric field which will then be coupled into the optical fiber after shaping a little
bit of that signal that comes out of the transmitter and maybe combining it with other channels
that are present in the same system , but in the fiber as soon as this ah signal is
coupled, the signal is actually a continuous time signal.
So, ah all though we do talk about discrete levels and other things, the actual waveform
that is being propagated from point a to point b is continuous. But luckily, we do not have
to track over all the changes that the pulse is undergoing as it propagates, the pulse
being representing some amount of mean representing the digital information , you do not have
to keep track of every aspect of the pulse because at the end ah at the I mean at the
receiver, you are going to simply sample the signal and use the sample .
So, which means that if the pulse changes it is amplitude slightly or you know gets
distorted a little bit, it is alright because we are not interested in those small details
, but as long as these distortions and changes are not too drastic to change the pulse shape
itself, then you are almost ok and this is the reason why digital communication offers
a higher immunity compared to analogue communications , ok. We will talked a little bit about analogue
communications , we talked a little bit of digital communications, we need some mathematical
tools to you know discuss further and this module is actually concerned with review of
some basic concepts in both deterministic and random signals that you would have seen
, but perhaps kind of not being familiar right away .
Ah this is a good refresher, but because of the nature of this course being you know not
the focus on these concepts , the review will be very fast that is ah the phase of the review
will be fast , but these are standard topics in undergraduate communication system. So,
you can actually look at any standard textbook and then you know brush up your knowledge
of these concept. So, with that ah caution in I mean caution that I have given you, let
us look at some concepts that are absolutely critical ah which will allow us to represent
to talk about the signals and then to study what happens as the signals pass through different
elements, different elements of the optical communication system. What happens when noise
is added, what exactly is noise, all those things we are going to review by first stating
that signals actually come in two varieties; a signal is essentially a function of time.
We have used these signals a lot in optical communication systems as well, we had for
an optical fiber, we had a certain pulse that was transmitted and at the output you had
a pulse and a shape of these pulses were different because of the dispersion in the fiber.
But never the less, we did represent the electric field at the input of the fiber as a way you
know , as a time domain function and this time domain function presumably carry some
information therefore, this can be considered as a signal. In this case, this is an optical
signal or an electric field and at the output again, you had at the output of the fiber,
you had ah for z equal to L, you had another time domain function which was you know E
of L comma t . So, these are all signals which hopefully convey some amount of information
to you. If we consider a simple signal such as x of t, this signal x of t could be of
two types broadly ok; one is the deterministic signal and the other one is the random signal.
The deterministic signal is one whose properties are reasonably easy to predict or rather.
In fact, a proper deterministic signal is one whose properties can be specified to any
accuracy as you want and if these deterministic signals passed through, deterministic systems
that is transformations that would take x of t as input and transform it into some output
which we will denote as y of t . So, if this rule of transformation or mapping
is actually kind of fixed, then you actually have a deterministic system and once you have
been given a deterministic signal x of t and the deterministic system transfer function
or the impulse response of the system, then it is easy to find out what will happen to
y of t . Now, it would have been very nice for our communication system to have been
composed only of deterministic signals or and deterministic systems . Unfortunately,
that is not the case ok because you may actually have, so for arguments sake, I am going to
assume that the fiber is deterministic. So, if I have sent in a pulse, I know exactly
what will happen to the pulse at the output, but this is the pulse that is I am not going
to measure this pulse right here right, what actually finding at the output of the fiber
is the electric field and to convert electric field into something that can be measured
in the form of a current, you will have to use of photodiode , right.
So, once you use the photodiode and there is some resistance R L which is the load resistor
let us say you are monitoring the output across this resistor R L . So, this optical pulse
which is sent into the photodiode would then generate a photo current and this photo current
would multiplied through the load resistor R L to generate the photo voltage to so to
speak . But unfortunately, a real register this is your R L which is an ideal resistor
, but a real resistor is actually a physical register it is made out of physical elements
and this elements of course are atoms coming from whatever the materials that are used
to make up a resistor and these physical resistors ah because they are atoms and other things,
they actually have some kinetic energy when the temperature t is greater than 0 and you
know that one of the properties of this ah you know atoms is and the electrons that are
there is that if they have been given some amount of energy, that they would like to
jump from the valence band and going to the conduction band ok. I am just assuming the
these are crystals . Therefore I am talking about band structure
here, but it is not really necessary that way. So, the idea here is that if whatever
material that you take , if it is ah temperature is greater than 0, then there is an associated
thermal energy which will then be imparted to the electrons, which will then allow this
electrons to move. So, it will move collide with the atoms and then come back to rest
and because there is some energy again available it will move. So, there are this random motions
that keep happening in the resistor giving rise to what is called as a thermal noise
right and this noise is an example of a signal that is random in nature which means that
if I were to measure, if I were to measure this thermal noise across a resistor with
an ideal voltmeter or whatever the meter that I am using right and if I have no other uncertainties
in this measurement scheme; that is I have fix the measurement correctly, I have the
voltage voltmeter also working correctly, the value that the voltmeter will actually
show is not predictable that is if I look at the value I know what is the value, but
before hand I cannot say with 100 percent certainty that this will be the amplitude
of the voltage that you are going to measure and I cannot do this even if by chance I am
actually correct once, it does not matter much because if I were to repeat this experiment
a million and billion times, then my chances of predicting or being right will actually
drastically go down ok. Because I cannot predict the amplitude but
I can predict that the amplitude would more or less be this particular value on average
that is I look at results of million of experiments and then I say on an average the amplitude
should be 3.5 micro volts and the variance that is the deviation from this average be
bounded to about plus or minus 1 microvolt. I mean I can say these statements and when
I am making these statements, I have a picture in my mind in which these type of experiments
are being repeated a million million million million times indicating that I am not able
to ah you know ah give the results of one single experiment but on an average, I can
predict the properties of this ah noise processes as well so.
So, the difference between deterministic and random signals is this whenever I perform
an experiment with a deterministic signal passing through a deterministic system, I
can always predict the output no matter how many times I repeat that experiment. On the
other hand, if I take noise which is an example of a random signal and then I pass it through
a deterministic system, let us say the system is known very well, but because the input
itself is unpredictable the output is also not predictable and therefore, the output
is also random. So, that is not precise, but that is as close as we will get to define
what is the random signal and what is the deterministic signal ok in our modulators
in the communication system. You can represent the output of a laser ok
as the electric field E of t ; for now let us ignore the ah polarization aspect of it
and this output can be given by ah say square root of P s e to the power j omega 0 t ok
and where omega 0 is of course the carrier and I am using what is called as the complex
notation . So, the signal that comes out will have to be a real signal. So, let us say this
signal is about route P s cosine omega 0 t as it would be I cannot obtain E to the power
j omega 0 t as the output . But I do not want to keep writing cosine and sine, so therefore,
this is a short hand notation that we use to represent the output that would come out.
So, this is in fact, related to what is called as a Phasor and in terms of the phasor, you
can also write down you mean you can write down whatever the signals that are coming
out of the transmitter and further as they propagate through the various elements, you
can look at what is happening to these phasors or to these complex signals and analyze the
system ok . So, let us just before going back to the complex domain, so let us actually
look at this square root P s cos omega 0 t now this is my carrier correct. So, this is
the carrier that I have and here the two parameters that you can see are this power P s I have
use square root because I am writing the amplitude out there and then I have omega naught which
is the carrier frequency. If these two are well defined and they actually have a unique
value , then the carrier will be a deterministic signal unfortunately this is an idealized
expression because I have to account for the fact that this omega 0 itself can be a function
of time called as the drift of the central frequency in the lasers and moreover the spectrum
will not be exactly a impulse like spectrum centered at omega 0. In fact, they will be
a small amount of curve around this frequency omega 0 which will then cause a certain noise
to appear and this noise appears . Noise remember is just another a function but that function
is non deterministic or random in nature. So, that noise would appear here as what is
called as the phase noise of the laser ok. So, in the complex representation that would
be E to the power j omega 0 t plus theta n of t .
So, you can in fact, right down that the overall phase psi of T of a non ideal laser. So, this
was a non ideal laser or non ideal oscillator will be ah the linear phase that comes because
of the centre frequency plus theta n of t which is the phase noise plus delta omega
of T times T . Of course, I could have combined omega 0 and ah t and delta omega of t into
t , but that anyway is ok alright we can put them separately out .
So, this of course, is the deterministic linear phase that you are going to get if the laser
were to be ideal, then the other two terms would not exist . If you were to plot this
psi of t as a function of time t , it would be a straight line passing through the origin
and the slope of this would have been equal to omega naught but because of the phase noise,
this total phase will not be exactly same it will be having some kind of ah change or
you know different or you know slope and other things . And if delta omega of t , E which
is ah which would then referred to the ah which is which usually referred to the frequency
drift what the the effect of this frequency drift is to actually change the slope right.
So, you may actually have this kind of a behavior overtime the slope may actually change. So,
maybe this is also another behavior and as we can see, I cannot write down a single graph
and then say this is how the total phase would change because all these terms theta n of
t and delta omega of t times t are random signals ok . So, that is what we actually
have anyway . So, as this laser comes out, you are going to modulate the laser correct.
So, you are going to modulate the laser with whatever the modulation signal that will be
there. So, let us say the modulation signal is sum m of t. It does not mean that m of
t is a continuous time signal only, it could be a discrete time or digital signal as well,
but that is we have just at this point left it as it is.
So, what will be the output of this modulator? Well, depends on what kind of modulation we
are performing right. So, ah, but if you were to consider the simple a tumble sideband suppressed
carrier type of modulation . So, what you have to do is to imagine that this m of t
is going to multiply this ah carrier which I am going to write this as square root of
P s E to the power j omega 0 t and for now , I am going to imagine that this theta n
of t and delta omega t of e are equal to 0 ok because I want to first look at what happens
to the deterministic case and then come back to the random case.
So, here is what I am doing. So, I have this m of t and then when it you know when it gets
multiplied to this ah carrier , you get square root P s E power j omega 0 t m of t ok. This
is a signal that would come out . Now, in terms of frequency domain what exactly
has taken place. So, this is your multiplier, this is your carrier . Now square root P s
E power j omega 0 t is you know, we will have a spectrum which is center at omega naught
and it spectrum will be a sharp P cut omega naught called as the impulse function here.
It will have an amplitude of square root P s, but you do not need to worry about the
amplitude at this part . I mean at this point. So, when this spectrum
goes through and then you have an m of t signal multiplying with respect to I mean multiplying
this input signal or the input spectrum here, we need to specify slightly more about this
m of t. Let us assume that this m of t spectrum would look like this ah maybe this is deterministic
or maybe it is not deterministic, but you can assume that this is the spectrum that
would be coming out of this m of t. For now , we will just take it to be deterministic
m of t ok. So, I have this m of t spectrum. Let us assume that the spectrum is centered
at 0 frequency ok and instead of working with omega I will switch and work with the frequency
f. So, similarly the carrier will not be omega
0 , but it could actually be f naught ok . And then, the maximum ah m of f , so, this is
m of f. I am only showing you the absolute spectrum the phase spectrum I am not showing
at there, but it will be there ok. So, it will have both magnitude as well as phase
spectra. So, we will assume that the spectrum is non zero only over this band which is say
from minus w to plus w and because the significant spectrum is concentrated around 0 frequency,
this is called as Base band signal ok. That is , this is the spectrum is that of a baseband
signal . So, m of t is a baseband signal and then it has a bandwidth of w ok. So, it has
a bandwidth of W because the maximum frequency is W Hertz .
So, it has a bandwidth of W Hertz and why did I write W and naught the total width to
W because in base bands signal, the bandwidth is counted only from positive frequency part
ok. So, it will be 0 to W. So, this is bandwidth of W Hertz ok and this operation of multiplication
by the square root what it would do is to you know take this was baseband signal and
then , you know shift the centre of that one shift the center instead of at 0 it would
have shifted the center to f 0 . So, what you know how is an impulse f naught, but of
course, you do not actually get to have the impulse the way I have written it .
So, so, this is just used to place hold where is f naught is this is on the f axis of course,
and around this will be your baseband spectrum ok. So, around this is your base band spectrum
of whatever the band width that you now have the highest frequency here will be f 0 plus
W; the lowest frequency here will b f 0 minus w I am of course, that f 0 is much larger
than W ok and this signal which I have shown has only positive frequency components and
therefore, of course, is complex is not really going to be there in your actual signals .
Because in actual or systems , in actual systems you will have not E power j omega 0 t, but
you will have cos omega 0 t or cos 2 phi f naught t that would be multiplying , so therefore,
the spectrum of the carrier would be having two impulses. The amplitude of course will
reduce by half ok. This is all coming from simple Fourier transform theory that you must
be familiar with and if you look at the carrier the carrier has two impulse one at g naught
and the other at minus f naught and when you multiply that one with m of t , you are going
to shift the spectrum m of f into two portions ; one to the right that is centered at f naught
and the other one centered at minus f naught. So, again you will have your baseband signal
both will actually have the same aptitude , this will have the same amplitude and this
frequency will now be minus f naught plus W and this one will be minus f naught minus
W . So, this signal which we have written let us call this signal as x of t and the
spectrum of this one as x of f and if you look at the significant spectral content of
x of f that does not occur at 0 Hertz rather it occurs at f 0 which is greater than 0 Hertz
and therefore this is called as pass band signal , sometimes also called as band pass
signal. Usually, this band pass signal or pass band
signals are not represented in this ah manner because you do not want to keep carrying around
everywhere e to the power j 2 phi of 0 t . So, rather you want to be able to talk about what
is called as the canonical representation of the band pass signal. So, canonical representation
means that any band pass signal whose carrier frequency f 0 is greater than 0 and usually
much much greater than 0 can actually be written as x of t as the real part of x ah tilde of
t E to the power j 2 phi f 0 t ok where this x tilde of t is called as the complex low
pass equivalent signal ok. The complex low pass equivalent signal and this is the complex
low pass equivalent of the band pass signal. So, in terms of pictures , this was a band
pass signal that I had and if you were to obtain the equivalent low pass signal of this
one which in this case turns out to be the m of t signal, but in general you do not know
if you do not know m of t , that is fine you can start with the band pass signal and then
go back to the base band. So, when you do that band pass to base band transformation
or transition , then you essentially generate this complex low pass equivalent signal . And
if you take that complex low pass equivalent signal multiplied by e power j to phi f 0
t and then take the real part of it , that will give you the real band pass signal ok.
So, that will give you the real pass signal and this complex low pass equivalent is sometimes
simply called as the base band equivalent signal or the complex envelop. Because this
is already complex , you can write this as x i of t plus j x q of t where you identify
this i and q as the in phase and quadrature components . So, in phase component and quadrature
components are again time varying functions and the information is usually present in
these two arms; utilizing these two you can construct first the complex envelope and from
the complex envelop , you can construct the band pass signal ok.
So, the by substituting for x the complex envelop x of t can be explicitly written down
in terms of x i ok. So, this is x i of t cosine 2 phi f 0 t minus x q of t sin 2 phi f 0 t
ok. So, this cosine term that is that ah E signal which is or the function which is multiplying
the cosine term is called as the in phase term and the one that is multiplied to the
sine is called as the quadrature term. You can recover x i and x q independently by multiplying
on one branch of x of t by cosine 2 phi f 0 t and on the other branch by multiplying
cosine with sin 2 phi f 0 t . So, what would that would do it would actually
correlate these two and ah f 0 t and then it would actually extract out x i , so the
output would be proportional not exactly equal, but it is amplitude will be slightly different
, but it is possible for you to actually extract these signals as you know you can verify from
your communication system takes place ok . So, this is what the ah equation wise was,
but what actually happened out here was you started off with some base sorry pass band
signal or the band pass signal which had ah centers at f 0 and then, you can multiply
this one this entire spectrum . If this is x of f , you can multiply the spectrum by
a unit step function ok , a unit step function would be exactly one for f greater than 0
and it would be 0 for f less than 0. What it will do is it will eliminate this set of
frequencies right and the result of applying this or multiplying these two will be a single
spectrum which is centered at f naught you can call this fellow as x plus of f . And
next what you do is to actually shift this x plus of f on to the left by the factor f
0. So, if you take this x plus of f plus f 0 . What you would have done is to actually
push this spectrum which is originally centered at f naught onto center at 0 and of course,
this x plus of f plus f naught would actually be in time domain look like x plus of t , E
to the power minus the j 2 phi f 0 t and once you have done, this one you have obtained
the complex envelop x tilde of f. So, therefore, in terms of time domain this
would be the complex envelop x tilde of t . So, x tilde of t is x plus of t E power
minus j 2 f 0 t and exactly what is x plus of t ; x plus of t is the inverse Fourier
transform of x of f multiplied by u of f ok. I will leave this as an exercise for you to
figure this out. So, we will rap this module now and consider several other concepts in
the next module. Thank you very much .
