noc18-ee28 Lecture 50-Review of communication concepts-II

IIT Kanpur July 201
20 followers
Dec 3, 2023
[Music]
hello and welcome to nptel's MOOC on
fiber-optic communication systems and
techniques in this module we will
continue the review of some topics that
we started in the previous model so we
saw in the previous module that given X
I of T and X Q of T you can actually
construct the bandpass signal suppose X
I of T could be plus a rectangular
function of duration T this is our
notation for defining a rectangular
pulse of duration T which in the way
that we have defined will extend from
minus T by 2 to plus T by 2 okay and
then XY of T can also be minus a
rectangular pulse of the same duration
so if you want to send information in
the form of a digital communication
system when a 0 occurs you may choose to
transmit the baseband signal X I of T as
a rectangular function of T by T or the
rectangular pulse of duration T and if
you want to transmit a bit 1 in this
baseband you can transmit it with minus
a so the amplitude would be reverse so
minus a rectangular pulse of the same
duration would look like this okay so
this is alternatively or depending on
what is the sequence that you want to
transmit you can transmit these signals
in the baseband of course on a fiber
this is not going to go as a baseband
signal but it will actually go as a
passband signal or a bandpass signal so
which is X of T in our notation in the
previous module because X Q of T in this
case is actually equal to 0 X of T will
be either plus a rectangular pulse T by
D cosine 2 pi F naught T or it would be
minus a and the same thing out there
suppose I want to transmit a safe
which is say one zero one okay what I
have to do I have to first split up the
transmission into three slots each slot
is of a duration T seconds okay
depending on the rate at which we are
transmitting this T will be determined
so the rate at which you're transmitting
will be one by T so and so symbols per
second is what we would call this rate
as and what is the idea here so when you
want to transmit X of T with a bit 1 0
or 1 or 1 or 0 you have to choose
appropriately between a rectangular
pulse T by T times cosine 2 pi of 0 T or
minus a so in this case it is minus a so
what you actually have will be minus a
cos 2 pi F 0 T over the duration minus T
by 2 to plus T by 2 or essentially to
say that over one time duration you are
either transmitting minus a cos 2 pi of
0 T which is a time-limited function or
you're transmitting plus 8 cos 2 pi F 0
T over the same time limit that you are
looking at ok so if you were to look at
how this would be seen if you were to
plot that one so this is over time and
then this is minus T by 2 to plus T by 2
over which the rectangular pulse itself
is defined to be equal to amplitude 1
and here you now have a cosine wave also
being transmitted right so and since
this is a cosine wave obviously this
will start at max here at T equal to 0
and then go like that so this is your
cosine Omega 0 do you have to imagine
that there are a lot of periods out
there and then it will end up here okay
so this will be lot of those cycles up
there because f 0 is usually considered
to be very very large compared to 1 by T
here ok why 1 by T because that is a
bandwidth approximately that's the
bandwidth of the rectangular pulse one
half of the bandwidth of the rectangular
pulse so if you want to transmit in
sequence here what you have to do is to
essentially first draw the envelope here
and then start putting in either plus a
cosine 2 pi F naught T or minus a cosine
2 pi F naught T right so with this as my
center here
this is the cosine wave that I am
writing so that is how it would be so
you should also ensure that there are
integral number of cycles in this one
for various results this will be
integral number of cycles in one
direction but now when you actually have
a zero you now have to transmit - or
rather for one we should have
transmitted - a cosine T right so for
that one we should have actually
transmitted something like this so we
should have transmitted a - cosine wave
and then this is one cycle on this side
and then this is similarly the other
cycle on this side so I have just
reduced the value of f naught so that
you can clearly see the waveform but now
I have to transmit a zero and
transmitting a zero would mean I have to
now switch to plus a okay so at plus a
it would go like this and it would be in
this manner so one two and three this
should actually have been because I'm
starting with minus a this sequence here
should actually be like this so it
should have been this is your cycles or
if you want one more cycle this will be
the cycle okay whereas for the bit zero
since you're transmitting this one with
a plus a cosine T this would come here
so what is actually happening is there
is a sudden change in the phase when the
symbol makes a transition from one to
zero so perhaps it was not very clear
the way I have written but you know
hopefully when you look at and draw it
correctly or use MATLAB to get this one
correctly you will see that this is
alright and when you have a transition
again at one you will see that there is
a further 180 degree transition from
here to this one so this type of
transitions are the ones which actually
send you the information so if you were
to move your window appropriately and
sample the waveform in this case
sampling doesn't really help but the
idea here is essentially the same thing
you have to correlate this one with the
appropriate cosine waveform and then
extract this one but when you extract
them the samples you will essentially
obtain amplitudes which are like plus a
and then minus a in the absence of any
noise and by looking at the sine of your
output we can determine whether you
transmitted a 1 or a zero okay
incidentally this type of modulation is
called as binary phase-shift keying okay
this is in the class of those modulation
signals called as modulation formats
called as phase shift keying and this is
binary phase shift keying so we have
written X of T given that I had X I of
Terr did not start from X of T and then
come back to X I of T okay suppose you
take this as an exercise here suppose X
I of T can be plus or minus a okay and X
Q of T can also be plus or minus a over
the same time duration say minus T by 2
to plus T by 2 then what is the possible
values of this signal the carrier signal
or the bandpass signal X of T could be
okay and you can write down plot it and
then see for yourself that once have
specified the in-phase and quadrature
components it is rather easy for me to
obtain the bandpass signal so this will
bandpass signal X of T right so this is
an exercise for you you can take a look
at this
but don't be conditioned into thinking
that this is the only way we can
actually describe signals there is in
fact an important way of describing
signals which is called as the vector or
treating the signals as a vector okay
and then utilizing many of the
geometrical properties of a vector space
to understand more about the signals so
you have to understand something that is
very interesting out there so signals
are usually continuous quantities or may
be discrete quantities but if they are
continuous they would have an infinite
number of values over any finite
duration right because it's a cosine 2pi
l0t for example if you take a duration
of 10 seconds in that 10 seconds the
number of times is cosine 2 pi of 0 t or
the values that it would take but
actually be you know infinite the
numbers are so there are so many numbers
because it's changing continuously but
then when you think of a vector you
don't normally think of a vector having
this infinite number of values although
you can but this kind of a mapping or
understanding or equivalence between a
signal and a vector doesn't really seem
to be
that correct that's what you would think
at first right but that is not correct
there is in fact a one-to-one
correspondence between a signal and a
vector and whatever operations that you
can do to a vector you can actually
perform the same operations onto the
signal and inspect it a signal as a
vector okay the defining characteristic
of a vector is not its direction or
magnitude the defining characteristics
of the vector are slightly different and
we are going to talk about those things
so vectors are those objects which
usually follow certain rules for our
purpose if V is a vector and U is
another vector then there some V Plus u
should also be another vector W and they
should also fall in the same space as we
would call it it is possible to take a
vector multiplied by a scalar real or
complex and then get a new vector but
the most important thing about a vector
is its geometrical properties we define
the inner product of two vectors V and u
or the dot product of these two vectors
U and V as the length of U the length of
V okay and then the angle between the
two vectors here we have used the length
in a very specific manner up there so
this is a symbol called as norm okay and
in the norm essentially gives you the
length of that particular vector indeed
the norm is actually defined as the
inner product itself of the vector by
itself and the square root of that one
so the norm of V itself is defined by
first defining the inner product and
then talking about them or taking the
square root of the length and you can
show that when you are looking at vector
of the same direction I mean with the
inner product of a vector with itself
obviously that is going to give you the
length square of the vector correct this
is something that we already know from
our earlier vector space theory so the
norm is actually square root of the
length and it is usually taken to be a
positive quantity so these are all the
geometrical properties why because if
two vectors are orthogonal to each other
then the inner product of the two
vectors will be
equal to zero so the dot product being
equal to zero indicates that these are
vectors which are orthogonal or
perpendicular to each other so one of
them is V the other one is you okay then
there is another way of thinking about a
vector is to first think of a coordinate
system so in the three dimensional
Cartesian coordinate system we are all
familiar with it any vector here can be
described by giving its values or
components as we would call it VX X hat
plus V Y Y hat plus V Z Z hat
what are these values V X V Y and visit
we X is in fact given by the inner
product of the vector V with the unit
vector X and similarly dy and we said
okay and this vector X hat Y hat and z
hat are called as unit vectors that
completely specifies all the other
vectors in this three-dimensional space
so if you know this XY and Z any other
vector can be constructed as a linear
combination of these vectors XY and Z
and you can obtain the value of that
combination or the weight of that
combination V X V Y or we said by going
to this expression that is taking week
and then taking the inner product of
that V with respect to the appropriate
unit vector now in a in a similar manner
suppose I am given a signal F of T whose
energy is finite okay so this condition
I am imposing to avoid some mathematical
problems but there are various ways to
overcome this constraint let us not go
into too much of a detail there but if I
have F of T whose energy is finite
recall that the energy itself is given
by whatever the value of F of T that
exists over or the value of F of T
square or the magnitude of F of T square
over the time interval over which F of T
is defined in case it is not specified
we will let this be minus infinity to
plus infinity and we want this quantity
to be less than infinity so that we can
talk about energy of these signals okay
and I have used magnitude square to
denote the fact that F of T could be
complex for example a complex envelope
was actually a complex signal so if you
go back to the complex on
take the magnitude square of this
complex envelope and then integrate it
over whatever the range the complex
envelope is present or it is defined
then this will give you the energy of
the complex envelope and you know what
is X tilde of T that is actually X I of
T plus J X Q of T X I and X Q are reals
but if you take the magnitude square
you're going to get X I square plus X Q
square of course this is a function of
time here no doubt and then integrate
this one individually I mean integrate
this one over time to obtain the total
energy E X tilde and clearly you can see
that this energy will be split into the
in phase component energy plus the
energy in the quadrature component okay
so you have e X I plus e xq where X I
and X Q are the energies or e^x and DX u
are the energies of in-phase and
quadrature components so this is
something that is now intuitively
appealing to us because energy of a
complex signal has been defined what if
we define the inner product of two
complex signals which we will call this
as F of T and G of T in this manner
assume that both are energy signals and
both are complex so you have over the
integration range whatever that might be
if I define this as the inner product
will it satisfy all the properties of an
inner product yes because if you take
the inner product of F with itself what
are you going to get you are going to
get energy EF why do you get this one
because F conjugate times F will be
magnitude of F square DT integration
that is precisely what we have as EF and
EF being energy quantity will always be
greater than or equal to zero when will
it be equal to zero when F itself is
equal to zero right and to any signal F
of T you can add another signal G of T
and the sum will be denoted as the
point-wise addition of the two signals F
and G and you can also take a scalar
multiply that scalar to f of T
so it seems possible that those energy
class there is all signals whose energy
is finite actually define vector space
ok the dimension of the vector space is
something that we are going to talk
about it but the idea is that they can
be thought of as vectors ok and again in
the analogy of the perpendicular nature
of the vectors you can talk about it
right so if F of T if the integration of
these two waveforms turn out to be zero
then we will call f of T to be
orthogonal to G of T ok and given that
you have some M energy signals then it
is possible for you to find n energy
signals whose energies are actually
equal to 1 and n will be either less
than or equal to M depends on the
signals that have been given to you ok
suppose M signals are labeled F 1 of T F
2 of T and so on all the way up to F M
of T then it is possible for us to find
the signals Phi 1 of T Phi 2 of T all
the way up to Phi n of T the number n
will being smaller than M so you have 5
n of T Phi 2 of T and Phi n of T such
that any earth signal which where I can
be 1 2 all the way up to M can be
expressed as a linear combination of
this Phi J of T signals ok the meaning
of Phi G of T being orthogonal simply
means that integral of Phi R or you
knows now that we know the inner product
itself we don't need to go to the
integration thing so what it simply
means is that Phi I inner product with v
J will be equal to 0 unless i is equal
to j j ok so if i is not equal to j then
these two will be equal to 0 and of
course we know that these signals are
then called as orthogonal to each other
but because their energy is actually
equal to 1 they're also called as
orthonormal and because in terms of
these orthonormal functions you can
express any other signal f of T in the
same class of functions of finite energy
then these are called as orthonormal
basis functions ok and given
signals it is always possible for you to
find these end signals by utilizing many
procedures one of them is called as the
famous gram-schmidt orthogonalization
procedure so the grams per tonne instant
procedure will tell you this so as a
combination of F I of T or the
combination of Phi J of T with
appropriate weighting factors you can
express any signal F I of T of the
original M signal that you had now here
is an interesting thing so now look at
the way we write a vector V which is
written as V X X hat plus V Y Y hat plus
V Z deed had in three dimensions of
course otherwise you will have to write
this one in multiple dimension now it is
very clear that each of these Phi J's
are like the unit vectors each of these
CI J's are like the three components or
many components of the vector V along X
Y and set so clearly I can obtain C IJ
that is I can find out the component by
taking the inner product of the signal
fi of T with the signal Phi J of T ok if
they're all real then all the complex
conjugates will disappear so it your
life becomes much more simple ok so this
is something that we have I will give
this as an exercise to you given that
you have Phi 1 of T equals e to the
power J 2 pi F 1 T and then you have Phi
2 of T equals E power J 2 pi F 2 of T
defined over an interval of minus T by 2
to plus T by 2 that is all these signals
or all these two signals fiber of T and
Phi 2 of T are defined to be in this
manner over this duration can you show
that Phi 1 of T is orthogonal to Phi 2
of T that is their value or the inner
product will be equal to 0 unless F 2 is
equal to F 1 you can also further assume
F 1 to be an integer multiple of the
duration or inverse duration 1 by T and
F 2 is some M by T n is usually not
equal to M
okay so I believe this exercise this is
very very important exercise please do
workout this one so that you can better
appreciate the concepts of orthogonality
okay let me also give you two additional
functions one is square root of 2 by T
cosine 2 pi F naught T and the other is
minus square root of 2 by T sine 2 pi F
naught T what do you think about these
two functions this function is Phi 1 of
T and this function is Phi 2 of T again
I would like you to see whether these
two are orthogonal to each other
f 0 is the same for both so you would
actually expect that these two signals
to be orthogonal because they are
actually in quadrature and if you do
that you will be actually right okay and
you can use these 5 n of T and Phi 2 of
T to represent any modulated signal
because see any carrier signal X of T
would have been written as X I of T cos
2 pi F naught t minus X Q of T sine 2 pi
F naught T now in terms of you know
basic functions Phi 1 of T and Phi 2 of
T you can write this one as Phi 1 of T
and this one this minus sign including
as Phi 2 of T and in fact you can write
X of T as X I of T Phi 1 of T plus X Q
of T Phi 2 of T so in this case X Y of T
and X Q of T correspond to the time
varying coefficients that would get
multiplied to Phi 1 of T and Phi 2 of T
and of course the small difference
between the way we have written that
that when write the square root of 2 by
T factor is also present you can include
that square root of T by T factor either
in definition of 5 and Phi 2 or you can
push them in the definitions of X I and
X Q okay if you have done this exercise
ok after showing that these are
orthogonal and understood this one just
try your you know skills of writing a
signal as a vector provided X I of T for
this problem where we have given X I of
T as plus a
and xq of T we have given this as minus
a right so then you can find out what
would be your X of T in terms of I and Q
and then you can construct
two-dimensional space out there one
corresponds to the cosine or rather Phi
1 of t term and the other one would
correspond to Phi 2 of T okay you can
think of cos 2 pi F naught T as the
x-axis sine minus sine 2 pi of T as the
y-axis and then locate what would be the
corresponding X of T signal here and
whatever that you will locate will
actually correspond to what is called as
the symbol point and a collection of
such symbols would then be a
constellation okay and the spacing that
you have kept between these
constellations the way you have put them
up will all determine what kind of
modulation format you are actually going
to use and are we going to transmit okay
so in understanding the modulation
formats digital modulation formats it is
important for you to know that these
constellations are essentially
two-dimensional diagrams of the in-phase
and quadrature components which in this
case we have written as Phi 1 of T and
Phi 2 of T ok so this is the thing that
I wanted you to understand now we talked
a lot about signal but many similar
ideas also apply to noise meaning that
noise also can be expressed in the
canonical bandpass way that you have n I
of T cos 2 pi F 0 of t minus 10 Q of T
sine 2 pi F 0 T however noise is not
characterized by its Fourier transform
but it is characterized by its power
spectral density and you can actually
show that the power spectral density of
the in-phase and quadrature components
given that this power spectral density
of the noise itself is given by this
expression ok so that would be a sin of
F minus FC times SN of f plus FC okay
the exact roof or the better ideas about
this one you will not be you know
discussed here rather if you want to
know this particular thing you will have
to maybe this is not just one this is
about half so these factors of half one
they are all just scaling factors if you
you have to be consistent with that but
for you know understanding the ideas
these scaling factors are not really
important but this is exactly what you
are going to get for the interface and
quadrature components okay so that is
something that you should know and you
should also know that we are mostly
going to deal with noise which is
Gaussian in nature okay what it means is
that if this is a noise process when you
sample this noise at a time T came and
then sample the noise at another time T
I what you are going to generate will be
a noise random variables actually random
variables and these random variables are
assumed to be uncorrelated with each
other and the probability that or the
probability density function of this
random variable will actually be equal
to or given by e to the power minus NK
minus mu whole square divided by 2 Sigma
square 1 by Sigma root 2 pi where Sigma
or Sigma n let us say is the noise
variance as we would call it mu is the
average most of the times we are going
to deal with noise sources which are
actually having an average of 0 so in
that case mu will be equal to 0 and in
case the amplitude and what this
equation this is called as the Gaussian
probability density function what it
would tell you is that the amplitude NK
of the noise would follow a certain
equation and it would be most of the
times average it will be near average
but occasionally it will be deviating
from the average and again this is a
statistical description and I cannot you
know go to a lot of other interesting
things about noise here I will leave all
that thing to you know review for you
from the undergraduate textbook one of
the books that I recommend to know more
about these properties of signal as well
as noise as well as representation is
taken communication systems
third edition or second edition
okay so there's plentiful detail about
all these properties and please review
these properties because we are going to
use lot of these properties in
understanding the performance of optical
fiber communications in the next module
starting in the next module thank you
very much
[Music]
[Music]
