All About Modulation
All About Modulation
All About Modulation
Figure 3 - Binary FSK Carrier
In PSK, we change the phase of the sinusoidal carrier to indicate information.
Phase in this context is the starting angle at which the sinusoid starts. To transmit
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
3
0, we shift the phase of the sinusoid by 180. Phase shift represents the change in
the state of the information in this case.
sin(2 ) 1
( )
sin(2 ) 0
f t for bit
PSK t
f t for bit
!
=
"
+
#
Figure 4 - Binary PSK Carrier (Note the 180 phase shifts at bit edges)
ASK techniques are most susceptible to the effects of non-linear devices which
compress and distort signal amplitude. To avoid such distortion, the system must
be operated in the linear range, away from the point of maximum power where
most of the non-linear behavior occurs. Despite this problem in high frequency
carrier systems, Amplitude Shift Keying is often used in wire-based radio
signaling both with or without a carrier.
ASK is also combined with PSK to create hybrid systems such as Quadrature
Amplitude Modulation (QAM) where both the amplitude and the phase are
changed at the same time.
What is digital, what is analog?
There are three parts to a communications system.
1. The information, also called the baseband
2. The medium
3. The carrier
Information can be defined in two forms, digital or analog. Analog signal is
considered continuous. Its signal amplitude can take on any number of values
between the signal maximum and minimum. Voice is analog and can take any
number of volume levels between its dynamic-range which is the range of
volumes your vocal cords can produce. Digital devices convert analog voice to a
digital signal by process of sampling and quantization. The analog signal is first
sampled and then quantized in levels and then each level is converted to a binary
number. For example, we may quantize your voice in 16 levels. Each of these
levels can be represented by four bits.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
4
Perhaps you remember when your telephone system went to the tone dialing. It
went from being a pure analog system to a digital system based on sampling and
quantization. Other examples of analog information are music and voice
transmitted via FM and AM radio transmissions. Nearly everything else
nowadays is digital.
The medium is thing the signal travels through. It can be air, space or wires of all
sorts. Each of these mediums offers its own unique set of advantages and
distortions that determine what is used as a carrier. A short wire in a chip for
example may not need a carrier at all. A signal through space such as for satellite
transmission may need a very high frequency carrier that can overcome space loss
and other losses.
If medium is the road taken, then carrier is the truck that carries the information
hence we call it Carrier. Depending on the medium, it can be light as in optical
communications or a microwave signal as for mobile communications. An
electromagnetic carrier can be of any frequency depending on the medium and the
communication needs. Most mediums dictate what type of carrier (its frequency,
amplitude) can propagate through it and the type of distortions it will suffer while
traveling through it.
Anything that is wireless is analog always. Wired signals can be digital or
analog. Communications inside a computer are examples of pure digital
communications, digital data over digital medium. LAN communications are
digital data over analog medium. The radios are examples of analog data over
analog medium.
In general when we talk about a digital system, we are usually talking about
digital information over an analog medium. However, there are exceptions. Pulse
Coded modulation (PCM) is a form of modulation where there is no carrier, so
that makes it a pure digital system.
The Shift Keying the second two terms in the name of these modulations imply
that they are digital modulations, i.e. the information is digital.
Modulation is the process by which we map information on to a carrier. How
many ways can you fit your bags in car? Can you repack your stuff to make it fit
better? These are the conceptual questions we tackle when we talk about
modulation.
Signal Spaces and basis functions
Study of signal spaces provides us with a geometric method of conceptualizing
the modulation process. In a physical space when we describe a vector by its
coordinates (x, y); the vector is being described by a linear combination of two
functions (1, 0) and (0, 1). Any vector can be written as a linear combination of
these two functions which are called basis functions and are orthogonal to each
other.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
5
Another example of such a family of functions is the unit width pulses separated
in time shown below. Each of these is independent of others and clearly we can
use these functions to create any random data sequence consisting of square
pulses. Each one of these single pulses is a basis function. However, this is not a
very efficient set of basis functions as it takes a large number of these functions to
create a random signal.
0
(t)
1
(t)
2
(t)
3
(t)
Figure 5 - Ortho-normal basis set
Ideally we want as few basis functions as possible which when combined can
create a large number of independent signals, both digital and analog. In addition,
basis functions should
Have unit energy, such as the (1, 0) and the (0, 1) vectors and the above
unit pulses.
They should be orthogonal to every other function in the set, represented
mathematically by
i j
t t
i j
i j
+
$
=
=
!
"
#
( ) ( )
1
0
Another important example is the pair of sine and cosine functions of unit
amplitude. This special basis set is used as carriers in all real communications
systems.
1
(t)
1
-1
0
1 2
2
(t)
1
-1
0
1 2
t t
Figure 6 - Sine and cosine, two orthogonal functions are the basis set for all
modern communications
The concept of I and Q Channels
We can write real signals as a vector sum of two signals in quadrature called I and
Q. You can think of I and Q as the x, y axis projection of a signal.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
6
[ ] S I Q =
Below you see two views of a signal space. One shows a signal in I-Q space and
the other in its polar form.
2
(t)
1
(t)
s
11
s
12
2
(t)
1
(t)
s
11
, s
21
40.4
(a) I and Q projections (b) Polar form
Figure 7 - Signal vector plotted on signal space
In Figure 7(a) the x and y-axis are I and Q projections of the signal. Quantity s
11
is
I channel projection and s
12
is the Q projection of the signal. Figure 7 (b) shows
the same signal in polar form, with its length equal to its amplitude and the angle
is equal to its phase.
The coefficients s
11
represent the amplitude of I signal and s
12
the amplitude of the
Q signal. These amplitudes when plotted on the x and y axis respectively, give the
signal vector. The angle the signal vector makes with the x-axis is the phase of
this signal.
2 2
Magnitude of signal S I Q = +
1
Phase of the signal tan
I
Q
=
Constellation diagrams
In a spectrum analyzer, we obtain the constellation diagram by sampling both I
and Q channels at the same instant and then plotting I value against Q value on a
x-y diagram. The time axis can be imagined as coming out of the analyzer so new
points lie on top of the old ones at the transmitter. But when the constellation
diagram is created for a signal after it has gone through the medium, it shows
some offset from the transmitted points because of noise and other effects.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
7
The tip of the signal vector is called the constellation point. The length of this
vector or the distance from the origin is the energy of that signal.
[ ]
E S s s
s i i i i
= = +
2
1
2
2
2
From above equation we can compute the energy of each signal shown below by
adding the squared I and Q coefficients. The points that lie further from the origin
have higher energy than those that lies close in. In a hybrid modulation with the
following constellation pattern, the outlying signal sig2, has higher energy (i.e.
higher amplitude, higher E
s
) than inner signal, sig1. Another way to think is to
realize that we are really plotting the amplitude. So a longer vector has a larger
amplitude and hence higher energy.
1
(t)
2
(t)
sig1 sig2
. .
Figure 8 - The energy of a signal is proportional to its length
The distance between two signals (distance between the tips of the vectors) can be
written as the dot product between the two signals.
E s s S S
s s
1 2
1 2 1 2
= = cos
The distance between the tips is the difference in the energy of two signals. The
angle between the two vectors is the correlation of the two signals. So if two
signals are 90 to each other than, then cos is 0 and the two signals are un-
correlated or are said to be orthogonal. In this example, the angle is not 90, so a
correlation exists between these two signals and this makes it harder to resolve the
differences between the signals when received.
All carrier signals other than BPSK and QPSK have some level of correlation
between their signals. This is the reason why BPSK and QPSK form the limit of
the lowest bit error rate possible.
Symbols, bits and bauds
A symbol is quite apart from a bit in concept although both can be represented by
sinusoidal or wave functions. Where bit is the unit of information, the symbol is a
unit of transmission energy. It is the representation of the bit that the medium
transmits to convey the information. Imagine bits as widgets, and symbols as
boxes in which the widgets travel on a truck. We can have one widget per box or
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
8
we can have more. Packing of widgets (bits) per box (symbols) is what
modulation is all about.
In communications, the analog signal shape, by pre-agreed convention, stands for
a certain number of bits and is called a symbol.
Each of these is one bit
of information, 0's and
1's.
This is a carrier symbol
that represents one
information bit.
Time
Carrier amplitude Bit baseband amplitude
Process of modulation
converts bits to symbols.
Digital world of bits and
information
Analog world of signal
symbols
Figure 9 Digital information travels on analog carrier
A symbol is just a symbol. It can stand for any number of bits, not just one bit.
The bits that it stands for are not being transmitted, what is transmitted is the
symbol or actually the little signal packet shown above. The frequency of this
packet is usually quite high. The 1 Hz signal shown above is just an abstraction.
A baud is same as the symbol rate of a communication system.
PSK modulations
BPSK
Lets imagine a ship lost at sea with no communication system. It sees an
airplanes flying overhead and wants to communicate its plight to the airplane
while it is overhead. The captain marks two spots on each side of the mast as
shown below.
Now he holds a bright light and runs back and forth between the marked spots to
signal a message. Spot to the right means a 1 and spot to the left means a 0. We
assume that all airplanes seeing this know that what each light stands.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
9
Figure 10 Two signaling spots, a simple modulation system
This is a one dimensional signal, because the captain uses only one dimension
(running from left to right) to indicate a symbol change.
The shining of the light is a symbol. Simplest thing is to have the symbol stand
for just one bit and we call this BPSK modulation. We utilize just one sinusoid as
the basis function. We vary the phase of this signal to transmit information which
is identical in concept to the example of shining the light from the deck. Each
symbol is signaled by a change in phase or position of the light as in this example.
In BPSK we define two little packets of the cosine wave, one with zero phase and
second one with a 180 degree different phase.
Table 1 lists the two symbols and the signals used to represent them. (The carrier
signal shown is for f = 1.) The I and Q amplitudes are the x and y projections
computed as follows.
( exp ) cos( )
( exp ) sin( )
amplitude
amplitude
I symbol ression x phase
Q symbol ression x phase
=
=
From this we get, I = 1 for the first symbol and -1 for the second symbol. Q
amplitude is zero for both symbols because sin of both 0 and 180 is zero.
Symbol Bit Expression Carrier Signal I Q
S1 0
2
cos( )
s
E
t
T
1
0
S2 1
2
cos( )
s
E
t
T
+
-1
0
Table 1 Mapping rules for BPSK which uses two signals, both a variation
of a cosine signal
Look at the carrier signals in the above table. This is what is transmitted in
response to the bits not the bits themselves.
What are those funny coefficients in front of the expressions above?
Recall from Tutorial 1 that energy of a signal is equal to
2
2
s
A T
E =
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
10
So instead of writing a amplitude term to make the expression general, we write
the equation in terms of energy, where A = 2E T
s
/ . When referring to carrier
signals, we typically talk in terms of signal or bit energy, so it makes sense to
write the equations in terms of energy, which is what this scaling factor is. Now
we can scale the carrier signal for the power with which it is transmitted. Later
when we talk about bit energy, and E
b
/N
0
this will become clear.
Creating a BPSK carrier
How would we send a bit sequence 0111 0101 0010 1011 using BPSK signaling
technique? To transmit this sequence, we need 16 symbols since each BPSK
symbol stands for one bit. These are
s1 s2 s2 s2 s1 s2 s1 s2 s1 s1 s2 s1 s2 s1 s2 s2
Now string together the appropriate symbol signal packets from Table 1 in the
right order. Figure below is the modulated carrier that would be transmitted for
this sequence if we use the mapping in Table I.
2
2
carbpsk t ( )
s t ( )
16 0 t
0 2 4 6 8 10 12 14 16
2
0
2
Figure 11 - A BPSK signal for bit sequence 0111 0101 0010 1011
If you could catch the modulated carrier and look at it on a network analyzer, you
would see the above. However, the above picture is at a carrier frequency of 1 Hz,
which is not realistic. In real systems, the carrier frequency is very high and we
would see a signal that covers a lot of cycles between each transition.
What is a transition? A transition is the time at which we switch from one symbol
to the next. What happens at the transition boundary is different for various
modulations and is quite an important thing. In the case of BPSK, at every bit
transition the signal does a 180 degree phase shift.
We worry about what the signal does at transitions because of amplifier non-
linearities. Amplifiers used in communications have a very hard time with sudden
changes in signal amplitudes and introduce distortions. Since this makes it harder
to decode the symbol, we try to control these transitions.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
11
Doing modulation in hardware
Notice that in the above description of the modulated carrier, there are no I or Q
channels. There is just one signal. Well here is an important bit of information.
We dont actually transmit I and Q. We transmit just one modulated signal over
the airwaves but this signal is the composite of I and Q channels.
So and why do we need I and Q channels? Not only I and Q representations help
us to understand and create consistent modulation schemes but they are also used
to simplify the hardware design.
Now lets see how in hardware we would create the composite modulated signal.
The figure below shows a bit stream that we would like to send using BPSK
modulation.
Figure 12 A random bit stream, 01101100010001
Now the modulator takes a look at the above bit stream and makes a decision
about which symbol to send. Below are first few symbols from this bit stream.
The Q channel for BPSK is zero so just ignore it. (BPSK is one dimensional
scheme.)The modulator assigns a certain symbol (and there are just two in case of
BPSK) to the I channel depending on the bit to be sent.
Bit Symbol I Q
0 S1 -1 0
1 S2 1 0
1 S2 1 0
0 S1 -1 0
1 S2 1 0
1 S2 1 0
0 S1 -1 0
Table 2 BPSK Symbol to I channel mapping for the bit stream in Figure 12
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
12
Now in the next step, a carrier of free-running frequency f
c
, is used to create the
packet of signal. All we have to do to create the correct modulated signal is to
multiply the carrier with the I channel amplitude values. Here is how the mapping
would take place.
0 1 1 1
1 1 1 2
1 1 1 2
cos( ) 0 1 1 1
1 1 1 2
1 1 1 2
0 1 1 1
s
s
s
t s
s
s
s
cos(2 )
c b
f T
cos(2 )
2
c b
f T
+
QPSK
Carrier
x
RRC
shaping
RRC
shaping
/ 2
b
R
Set I
amplitude
Set Q
amplitude
+
2
delay
Table
Lookup
Figure 21 Hardware implementation of the QPSK modulator
Summary: The Serial to parallel converter takes the bit stream coming in at a bit
rate of Rb and splits it into two streams, each of half the bit rate. Depending on
the dual bit pattern coming in, I and Q amplitudes are set from a table lookup
function. Each of these is then individually modulated by a sine or a cosine wave
of carrier frequency after being shaped into a root raised cosine pulse. These are
added together to get the transmitted signal.
How QPSK differs from BPSK: QPSK has two basis signals. It can be
considered to be composed of two BPSK signals each of one half the bit rate.
Constant Envelope modulation
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
20
QPSK is part of a class of signals called constant-envelope signals. There is no
rigorous definition of a constant envelope signal. One definition is; when sampled
at the symbol rate, the sampled value of the amplitude is constant. Another is that
there are no discontinuous phase changes. Yet another is that the maximum and
minimum amplitude attained by the signal over one period is constant. The sine
wave is an ideal constant envelope signal.
Constant envelope signals suffer less distortion in high power amplifiers and are
preferred for wireless applications. The reason is that amplifiers work by
changing a signal's amplitude, either increasing or decreasing it. To increase a
signals power is to increase its amplitude. A non-linear amplifier changes the
signal amplitude by differing amount depending on the instantaneous amplitude
of the signal. The more the amplitude of a signal varies, the more non-linear
amplification occurs and this results in a distorted signal. QPSK is not technically
a constant envelope because of its discontinuous phase shifts but is considered
nearly so.
1.5
1.5
fsk t ( )
s t ( )
10 0 t
0 2 4 6 8 10
0
Figure 22 FSK is definitely a constant envelope modulation.
ask t ( )
s t ( )
t
0 2 4 6 8 10
0
Figure 23 - ASK is definitely not a constant envelope modulation
Offset QPSK
Offset QPSK is a minor but important variation on QPSK. In Offset QPSK, the Q
channel is shifted by half a symbol time so that I and Q channel signals do not
transition at the same time. The result of this simple change is that phase shifts at
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
21
any one time are limited and hence offset QPSK is more constant-envelope than
straight QPSK. In high power amplifiers and for certain satellite applications,
Offset QPSK offers better performance. Although in a linear channel its bit error
rate is the same as QPSK, in non-linear applications, its BER is lower when
operating close to the saturation point of the transmitting amplifier. Offset QPSK
(OQPSK) is also called staggered QPSK (SQPSK).
2
b
s
R
R =
Serial to Parallel
Converter
Information
bitstream
/ 2
b
R
I Channel
Q Channel
x
b
T delay
Symbol time delay
for OQPSK only
cos(2 )
c b
f T
cos(2 )
2
c b
f T
+
QPSK (or OQPSK)
Carrier
x
RRC
shaping
RRC
shaping
/ 2
b
R
Set I
amplitude
Set Q
amplitude
+
2
delay
% &
=
' (
) *
+ % &
= +
' (
) *
The I and Q channels are then multiplied by the carrier, cosine for the I channel
and sine for the Q channel. Note that the period of pulse shape is twice that of the
symbol rate.
( ) ( ) sin cos
2
( .5 )
( ) ( ) sin sin
2
t t
MSKcarrI t QPSKI t
T T
t t T
MSKcarrQ t QPSKQ t
T T
% & % &
=
' ( ' (
) * ) *
+ % & % &
=
' ( ' (
) * ) *
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
25
Figure 30 MSK I and Q modulated carriers.
Adding the I and Q components gives the MSK carrier of Figure 31. Compare this
carrier to a QPSK carrier. This one has much smoother phase shifts at the symbol
boundaries. This results in lower side lobes which is an advantageous property for
wireless signals since it results in less adjacent signal interference.
Minimum Shift Keying (MSK) is also called continuous phase (CP) Frequency Shift
Keying (FSK). FSK is the digital version of analog Frequency Modulation (FM) and
MSK is a form of FSK, where modulation index is equal to .5 which results in a
minimum frequency separation such that the modulation frequencies are still
orthogonal. (See FM tutorial)
Figure 31 MSK modulated carrier
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
26
2
b
s
R
R =
Serial to Parallel
Converter
Information
bitstream
1
Bit rate in bits/sec
b
b
R
T
=
/ 2
b
R
I Channel
Q Channel
x
x
2
delay
b
T delay
cos(2 )
c b
f T
cos(2 )
2
c b
f T
+
MSK
Carrier
x
x
1
sin 2
4
b
t
T
, - % &
. / ' (
) * 0 1
b advance
T
Figure 32 MSK modulator block diagram
Figure 32 shows the modification made to the QPSK modulator to create the
MSK signal. Only the pulse shaping has been changed. The half cycle time shift
of the OQPSK stays.
How MSK differs from QPSK: MSK is generally considered a FSK modulation
but it is exactly the same as OQPSK except that it uses a half-sinusoid for pulse
shaping instead of root-raised cosine pulses.
Gaussian MSK (GMSK)
We created MSK by applying a half sinusoid to the square pulse. By using a
Gaussian pulse shape, the result can be improved even further. The modulation
obtained this way is called GMSK.
GMSK is used in several mobile systems around the world. Global Speciale
Mobile (GSM), Digital European Cordless Telephone (DECT), Cellular Digital
Packet Data (CDPD), DCS1800 (Digital Communications System in the 1800
MHz band) in Europe, and GSM-based PCS1900 (Personal Communications
Services in the 1900 MHz band) in the U.S. uses GMSK.
Recall that the root-raised cosine pulse has a roll off factor, . The roll-off factor
determines how sharply the pulse rolls off to zero energy. A Gaussian pulse
similarly has a BT factor that determines how sharply it rolls off. A BT of .3 is
used commonly.
MSK and GMSK, both being related to FM modulation, can both be created two
ways, 1. as a PSK signal and 2. as a FSK signal. Both are most commonly
implemented as a FSK technique.
All About Modulation Part I
Copyright 2002 Charan Langton www.complextoreal.com
27
The Gaussian pulse shape used instead of the half-sinusoid or the root raised
cosine is given by
2
( / 2)
1 .5 .5
( ) 2 ( 2 (
2 ln 2 ln 2
1
( )
2
b b
x
t
t T t T
g t Q B Q B
T
where
Q t e dx