Modulation

The objective of modulation technique is to

transport the message signal through a radio
channel with the best possible quality while
occupying the least amount of radio spectrum.
Signal characteristics that can be modified
signal x(t) = A cos(2πft + Φ)
• A – amplitude

• f – frequency

• Φ – phase (initial angle of the sinusoidal function

at its origin
Modulation
Modulation (cont’d)
Digital modulation techniques
ASK
ASK
PSK
Example: PSK
PSK Detection
Example
FSK
FSK (cont’d)
1
2 DPSK
2
 DPSK → Differential Phase Shift Keying
 Non-coherent Rx can be used

 easy & cheap to build

 no need for coherent reference signal from Tx

 Bit information determined by transition between two

phase states
 incoming bit = 1 → signal phase stays the same as

previous bit
 incoming bit = 0 → phase switches state

d k  mk  d k 1
 DPSK (cont’d)
 If {mk} is the message, the output {dk} is as shown below.

d k  mk  d k 1

 can also be described in modulo-2 arithmetic

 Same BW properties as BPSK, uses same amount of spectrum
 Non-coherent detection → all that is needed is to compare
phases between successive bits, not in reference to a Tx phase.
 power efficiency is 3 dB worse than coherent BPSK (higher
power in Eb / No is required for the same BER)
6-
DPSK (Modulator)
Facts
Examples:
QPSK
QAM
QAM ( cont’d)
QAM (cont’d)
 QPSK Transmission and Detection Techniques
Constellation
 Energy in simple language
 What we just saw says that the energy of a signal is
simply the square of the length of its corresponding
constellation vector

E
2
E=9+4=13

3
 Constrained energy signals
 Let’s say you are under peak energy Ep constraint in
your application. Just make sure all your signals are
inside a circle of radius sqrt(Ep )

Ep
 Correlation of two signals
 A very desirable situation in is to have signals that
are mutually orthogonal. How do we test this? Find
the angle between them

transpose
s1
s1T s2
s2 cos12  
s1  s2

 Find the angle between s1 and s2
 Given that s1=(1,2)T and s2=(2,1)T, what is the angle
between the two?

2 
s s  1 2  2  2  4
T
4 4
1 2

1 
 cos12   
5 5 5
s1  1  4  5  12  36.9o
s2  4  1  5
 Distance between two signals
 The closer signals are together the more chances of
detection error. Here is how we can find their
separation

N 2
  s1 j  s2 j 
2 2
d 12  s1  s2
2

j 1
1
 (1)  (1)  2
2 2
1 2

 d12  2
 Example: BPSK Constellation Diagram

  2 Eb   2 Eb  
S BPSK    s1  t   cos  2 f c t   ,  s2  t    cos  2 f c t  ;   ; 0  t  Tb
  Tb   Tb  
Eb  energy per bit; Tb  bit period
For this signal set, there is a single basic signal
Q
2
1  t   cos  2 f c t  ; 0  t  Tb
Tb

 
S BPSK   Eb 1  t   ,   Eb 1  t  
-Eb Eb
I

Constellation diagram
QAM constellation
 Distortions

Perfect channel White noise Phase jitter

QPSK decision regions

01 00

11 10

Decision regions re color-coded

 BPSK Error
 We know from the previous derivation that
 a1  a 0 
PB  Q  where  0 2  N 0
 2N  2
 0 

where Q u   
1

exp  z 2 / 2 dz
u 2
Constellation distance = 2 Eb

 2 Eb 
PBPSK  Q  

 N0 
DPSK error

1  Eb 
PDPSK  exp   
2  N0 
 QPSK Error
Es

Distance Between Adjacent points = 2Es

Since each symbol corresponds to two bits E s  2 E b

 2 E s  2 Eb

 2 Eb 
PQPSK  Q  

 N0 
 How does BER performance compare to BPSK?

 Why? same # of states per number of basis

functions for both BPSK and QPSK (2 states per
one function or 4 states per 2 functions)
 same power efficiency
(same BER at specified Eb / No)
 twice the bandwidth efficiency
(sending 2 bits instead of 1)
 Types of QPSK

Q Q Q

I I I

Conventional QPSK Offset QPSK /4 QPSK

 Conventional QPSK has transitions through zero (i.e. 1800

phase transition). Highly linear amplifiers required.
 In Offset QPSK, the phase transitions are limited to 900, the
transitions on the I and Q channels are staggered.
 In /4 QPSK the set of constellation points are toggled each
symbol, so transitions through zero cannot occur. This scheme
produces the lowest envelope variations.
 All QPSK schemes require linear power amplifiers
 OQPSK
 Offset QPSK
 The occasional phase shift of π radians can cause

the signal envelope to pass through zero for just

in instant
 Any kind of hard limiting or nonlinear amplification

of the zero-crossings brings back the filtered

sidelobes
 since the fidelity of the signal at small voltage

levels is lost in transmission.

 OQPSK ensures there are fewer baseband signal

transitions applied to the RF amplifier,

 helps eliminate spectrum regrowth after

amplification.
 Example above: First symbol (00) at 0º, and the next
symbol (11) is at 180º. Notice the signal going through zero
at 2 microseconds.
 This causes problems.
 Using an offset approach: First symbol (00) at 0º,
then an intermediate symbol at (10) at 90º, then the
next full symbol (11) at 180º.
 The intermediate symbol is used halfway through

the symbol period.

 It corresponds to allowing the first bit of the

symbol to change halfway through the symbol

period.
 The figure below does have phase changes more

often, but no extra transitions through zero.

 IS-95 uses OQPSK, so it is one of the major

modulation schemes used.

 In QPSK signaling, the bit transitions of the even and
odd bit streams occur at the same time instants.

 but in OQPSK signaling, the even and odd bit

Streams, mI(t) and mQ(t), are offset in their relative
alignment by one bit period (half-symbol period)
 the maximum phase shift of the transmitted signal at any
given time is limited to ± 90o
 The spectrum of an OQPSK signal is identical to that
of a QPSK signal, hence both signals occupy the
same bandwidth
 π/4 QPSK
 π/4 QPSK
 The π/4 shifted QPSK modulation is a quadrature

phase shift keying technique

 offers a compromise between OQPSK and

QPSK in terms of the allowed maximum phase

transitions.
 It may be demodulated in a coherent or
noncoherent fashion.
 greatly simplifies receiver design.

 In π/4 QPSK, the maximum phase change is

limited to ± 135o
 in the presence of multipath spread and fading,

π/4 QPSK performs better than OQPSK

DQPSK
DQPSK (cont’d)
 Symbol error vs. bit error
 When a symbol error occurs, we might suffer more
than one bit error such as mistaking 00 for 11.
 It is however unlikely to have more than one bit error
when a symbol error occurs

00 10 10 11 10
Sym.error=1/10
00 11 10 11 10 Bit error=1/20

10 symbols = 20 bits
 Interpreting symbol error
 Numerically, symbol error is larger than bit error but
in fact they are describing the same situation; 1 error
in 20 bits
 In general, if Pe is symbol error

Pe
 BER  Pe
log M
 Symbol error and bit error for QPSK
 We saw that symbol error for QPSK was
E
Pe  erfc( )
2 No
 Assuming no more than 1 bit error for each symbol
error, BER is half of symbol error
1 E
BER  erfc( )
2 2 No
 Remember symbol energy E=2Eb

1 x
where Q( x)  erfc( )
2 2
 QPSK vs. BPSK
 Let’s compare the two based on BER and bandwidth
BER Bandwidth
 BPSK QPSK BPSK QPSK

1  Eb  1
erfc  erfc

Eb 
 Rb Rb/2
2  N o  2  N o 

EQUAL
 M-phase PSK (MPSK)
 If you combine 3 bits into one symbol, we have to
realize 23=8 states. We can accomplish this with a
single RF pulse taking 8 different phases 45o apart

 2E cos2fc t  (i  1)  ,i  1,...,8


si (t)   T  4  ,0  t  T

0
 8-PSK constellation
 Distribute 8 phasors uniformly around a circle of
radius √E

45o

Decision region
 Symbol error for MPSK
 We can have M phases around the circle separated
by 2π/M radians.
 It can be shown that symbol error probability is
approximately given by

 E   
Pe  erfc sin , M  4
 No  
M 
 Quadrature Amplitude Modulation (QAM)
 MPSK was a phase modulation scheme. All
amplitudes are the same

 QAM is described by a constellation consisting of

combination of phase and amplitudes

 The rule governing bits-to-symbols are the same, i.e.

n bits are mapped to M=2n symbols
 16-QAM constellation using Gray coding
 16-QAM has the following constellation
 Note gray coding where adjacent symbols differ by
only 1 bit
0000 0001 0011 0010

1000 1001 1011 1010

1100 1101 1111 1110

0100 0101 0111 0110

 Vector representation of 16-QAM
 There are 16 vectors, each defined by a pair of
coordinates. The following 4x4 matrix describes the
16-QAM constellation

 3,3 1,3 1,3 3,3 

 3,1 1,1 1,1 3,1 
[ai ,bi ]  
3,1 1, 1 1, 1 3,1
 
3, 3 1,3 1,3 3, 3
 What is energy per symbol in QAM?
 We had no trouble defining energy per symbol E for
MPSK. For QAM, there is no single symbol energy.
There are many
 We therefore need to define average symbol energy
Eavg

1 M 2
Eavg   ai  bi2 
M i 1
 Eavg for 16-QAM
 Using the [ai,bi] matrix and using E=ai^2+bi^2 we get
one energy per signal

18 10 10 18 
10 2 2 10 
E  
10 2 2 10 
 
18 10 10 18 

Eavg=10
 Symbol error for M-ary QAM
 With the definition of energy in mind, symbol error is
approximated by

 1   2Eavg 
Pe  2 1  erfc 
 M   2 M  1No 
 M-ary bandwidth
 Summarizing, M-ary and binary bandwidth are
related by
BM-ary=Bbinary/logM

 Clearly , M-ary bandwidth is reduced by a factor of

logM compared to the binary bandwidth

C S
 B max   log 2 (1  )
B N
 8-ary bandwidth
 Let the bit rate be 9600 bits/sec. Binary bandwidth is
nominally equal to the bit rate, 9600 Hz

 We then go to 8-level modulation (3 bits/symbol) M-

ary bandwidth is given by
BM-ary=Bbinary/logM=9600/log8=3200 Hz
 M-ary FSK
 Using M tones, instead of M phases/amplitudes is a
fundamentally different way of M-ary modulation

 The idea is to use M RF pulses. The frequencies

chosen must be orthogonal

2E
si t   cos2fi t ,0  t  T
T
i  1,..., M
 MFSK constellation: 3-dimensions
 MFSK is different from MPSK in that each signal sits
on an orthogonal axis(basis)

2 
i t   cos2fi t , 3 s3
T √E s1=[√E ,0, 0]
0t T s2=[0,√E, 0]
s3=[0,0,√E]
i  1,..., M
√E
s1 1
√E
s2
2
 Orthogonal signals : How many
dimensions, how many signals?
 We just saw that in a 3 dimensional space, we can
have no more than 3 orthogonal signals
 Equivalently, 3 orthogonal signals don’t need more
than 3 dimensions because each can sit on one
dimension
 Therefore, number of dimensions is always less than
or equal to number of signals
 How to pick the tones?
 Orthogonal FSK requires tones that are orthogonal.

 Two carrier frequencies separated by integer

multiples of period are orthogonal
 Example
 Take two tones one at f1 the other at f2. T must cover
one or more periods for the integral to be zero
T T

 2 cos2f t cos2f t dt   cos 2  f

1 2  f2 dt
1
0 0


averages to zero
T

  cos2  f1  f2 dt
0
   
averages to zero if T=i/(f1-f2)
 ; i =integer
Take f1=1000 and T=1/1000. Then
if f2=2000 , the two are orthogonal
so will f2=3000,4000 etc
 MFSK symbol error
 Here is the error expression with the usual notations

1  E 
Pe   M  1erfc 
2  2N o 
 Tradeoff between BW Efficiency and
Power Efficiency
 There is a tradeoff between bandwidth efficiency and
power efficiency
 Adding error control codes

 Improves the power efficiency

 Reduces the requires received power for a

particular bit error rate

 Decreases the bandwidth efficiency
 Increases the bandwidth occupancy

 M-ary keying modulation

 Increases the bandwidth efficiency
 Decreases the power efficiency
 More power is requires at the receiver
 End for 2018 course
 Sunde’s FSK
 We might have to pick tones f1 and f2 that are not
orthogonal. In such a case there will be a finite
correlation between the tones

T
2 b
   cos(2f1t) cos(2f2 t)dt
Tb 0 Good points,zero correlatio

1 2 3 2(f2-f1)Tb
 Picking the 2nd zero crossing:
Sunde’s FSK
 If we pick the second zc term (the first term puts the
tones too close) we get
2(f2-f1)Tb=2--> f=1/2Tb=Rb/2
remember f is (f2-f1)/2
 Sunde’s FSK bandwidth is then given by

BT=2f+2Rb=Rb+2Rb=3Rb
 The practical bandwidth is a lot smaller
 Sunde’s FSK bandwidth
 Due to sidelobe cancellation, practical bandwidth is
just BT=2f=Rb

1/Tb=Rb
f f
BT=2 f+2Rb
f= (f2-f1)/2
f1 fc f2
fc=(f1+f2)/2
 BFSK example
 A BFSK system operates at the 3rd zero crossing of
-Tb plane. If the bit rate is 1 Mbps, what is the
frequency separation of the tones?

 The 3rd zc is for 2(f2-f1)Tb=3. Recalling that f=(f2-

f1)/2 then f =0.75/Tb

 Then f =0.75/Tb=0.75x106=750 KHz

 And BT=2(f +Rb)=2(0.75+1)106=3.5 MHz

 Point to remember
 FSK is not a particularly bandwidth-friendly
modulation. In this example, to transmit 1 Mbps, we
needed 3.5 MHz.

 Of course, it is working at the 3rd zero crossing that

is responsible

 Original Sunde’s FSK requires BT=Rb=1 MHz

 Spectrum of M-ary signals
 So far Eb/No, i.e. power, has been our main concern.
The flip side of the coin is bandwidth.

 Frequently the two move in opposite directions

 Let’s first look at binary modulation bandwidth

 BPSK bandwidth
 Remember BPSK was obtained from a polar signal
by carrier modulation

 We know the bandwidth of polar NRZ using square

pulses was BT=Rb.

 It doesn’t take much to realize that carrier modulation

doubles this bandwidth
 Illustrating BPSK bandwidth
 The expression for baseband BPSK (polar)
bandwidth is
SB(f)=2Ebsinc2(Tbf)

2/Tb=2Rb

BPSK BT=2Rb

1/Tb f fc-/Tb fc fc+/Tb

 BFSK as a sum of two RF streams
 BFSK can be thought of superposition of two
unipolar signals, one at f1 and the other at f2

0.5
BFSK for 1 0 0 1 0 1 1
1

0.8 0

0.6
-0.5
0.4

-1

+
0.2 0 1000 2000 3000 4000 5000 6000 7000 8000

-0.2 1

-0.4
0.5
-0.6

0
-0.8

-1 -0.5
0 1000 2000 3000 4000 5000 6000 7000 8000

-1
0 1000 2000 3000 4000 5000 6000 7000 8000
 Modeling of BFSK bandwidth
 Each stream is just a carrier modulated unipolar
signal. Each has a sinc spectrum

1/Tb=Rb
f
BT=2 f+2Rb
f= (f2-f1)/2
f1 fc f2
fc=(f1+f2)/2
 Example: 1200 bps bandwidth
 The old 1200 bps standard used BFSK modulation
using 1200 Hz for mark and 2200 Hz for space. What
is the bandwidth?

 Use
BT=2f+2Rb
f=(f2-f1)/2=(2200-1200)/2=500 Hz
BT=2x500+2x1200=3400 Hz

 This is more than BPSK of 2Rb=2400 Hz

 MPSK bandwidth review
 In MPSK we used pulses that are log2M times wider
tan binary hence bandwidth goes down by the same
factor.
T=symbol width=Tblog2M
 For example, in a 16-phase modulation, M=16,
T=4Tb.
Bqpsk=Bbpsk/log2M= Bbpsk/4
 MPSK bandwidth
 MPSK spectrum is given by

SB(f)=(2Eblog2M)sinc2(Tbflog2M)

Set to 1 for zero crossing BW

Tbflog2M=1
-->f=1/ Tbflog2M
=Rb/log2M

BT= Rb/lo g2M

1/logM f/Rb
Notice normalized frequency
 Bandwidth after carrier modulation
 What we just saw is MPSK bandwidth in baseband

 A true MPSK is carrier modulated. This will only

double the bandwidth. Therefore,

Bmpsk=2Rb/log2M
 QPSK bandwidth
 QPSK is a special case of MPSK with M=4 phases.
It’s baseband spectrum is given by
SB(f)=2Esinc2(2Tbf)

B=0.5Rb-->
half of BPSK After modulation:
0.5 1 f/Rb Bqpsk=Rb
 Some numbers
 Take a 9600 bits/sec data stream
 Using BPSK: B=2Rb=19,200 Hz (too much for 4KHz
analog phone lines)
 QPSK: B=19200/log24=9600Hz, still high
 Use 8PSK:B= 19200/log28=6400Hz
 Use 16PSK:B=19200/ log216=4800 Hz. This may
barely fit
 MPSK vs.BPSK
 Let’s say we fix BER at some level. How do
bandwidth and power levels compare?
M Bm-ary/Bbinary (Avg.power)M/(Avg.power)bin
4 0.5 0.34 dB
8 1/3 3.91 dB
16 1/4 8.52 dB
32 1/5 13.52 dB
 Lesson: By going to multiphase modulation, we save bandwidth but
have to pay in increased power, But why?
 Power-bandwidth tradeoff
 The goal is to keep BER fixed as we increase M.
Consider an 8PSK set.

 What happens if you go to 16PSK? Signals get closer

hence higher BER
 Solution: go to a larger circle-->higher energy
 Additional comparisons
 Take a 28.8 Kb/sec data rate and let’s compare the
required bandwidths
 BPSK: BT=2(Rb)=57.6 KHz
 BFSK: BT = Rb =28.8 KHz ...Sunde’s FSK
 QPSK: BT=half of BPSK=28.8 KHz
 16-PSK: BT=quarter of BPSK=14.4 KHz
 64-PSK: BT=1/6 of BPSK=9.6 KHz
 Power-limited systems
 Modulations that are power-limited achieve their
goals with minimum expenditure of power at the
expense of bandwidth. Examples are MFSK and
other orthogonal signaling
 Bandwidth-limited systems
 Modulations that achieve error rates at a minimum
expenditure of bandwidth but possibly at the expense
of too high a power are bandwidth-limited

 Examples are variations of MPSK and many QAM

 Check BER rate curves for BFSK and BPSK/QAM

cases
 Bandwidth efficiency index
 A while back we defined the following ratio as a
bandwidth efficiency measure in bits/sec/HZ
=Rb/BT bits/sec/Hz

 Every digital modulation has its own 

 Bandwidth Efficiency of Modulation
 Ability of a modulation scheme to accommodate
data within a limited bandwidth.
 Bandwidth efficiency reflect how efficiently the
allocated bandwidth is utilized

R
Bandwidth Efficiency :  B  bps/Hz
B

R: the data rate (bps)

B: bandwidth occupied by the modulated RF signal

202
 Shannon’s Bound
There is a fundamental upper bound on achievable
bandwidth efficiency. Shannon’s theorem gives the
relationship between the channel bandwidth and the
maximum data rate that can be transmitted over this
channel considering also the noise present in the channel.

Shannon’s Theorem
C S
 B max   log 2 (1  )
B N

C: channel capacity (maximum data-rate) (bps)

B: RF bandwidth
S/N: signal-to-noise ratio (no unit)
 Tradeoff between BW Efficiency and
Power Efficiency
 There is a tradeoff between bandwidth efficiency and
power efficiency
 Adding error control codes

 Improves the power efficiency

 Reduces the requires received power for a

particular bit error rate

 Decreases the bandwidth efficiency
 Increases the bandwidth occupancy

 M-ary keying modulation

 Increases the bandwidth efficiency
 Decreases the power efficiency
 More power is requires at the receiver
 Example:
 SNR for a wireless channel is 30dB and RF
bandwidth is 200kHz. Compute the theoretical
maximum data rate that can be transmitted over
this channel?
 Answer:  30 dB 
S  10 
 10 
N
S
C  B log 2 (1  )  2 x105 log 2 (1  1000)  1.99Mbps
N

205
  for MPSK
 At a bit rate of Rb, BPSK bandwidth is 2Rb

 When we go to MPSK, bandwidth goes down by a

factor of log2M BT=2Rb/ log2M

 Then =Rb/BT= log2M/2 bits/sec/Hz

 Some numbers
 Let’s evaluate  vs. M for MPSK
M 2 4 8 16 32 64
 .5 1 1.5 2 2.5 3

 Notice that bits/sec/Hz goes up by a factor of 6 from

M=2 and M=64

 The price we pay is that if power level is fixed

(constellation radius fixed) BER will go up. We need
more power to keep BER the same
 Defining MFSK:
 In MFSK we transmit one of M frequencies for every
symbol duration T

 These frequencies must be orthogonal. One way to

do that is to space them 1/2T apart. They could also
be spaced 1/T apart. Following The textbook we
choose the former (this corresponds to using the first
zero crossing of correlation curve)
 MFSK bandwidth
 Symbol duration in MFSK is M times longer than
binary T=Tblog2M symbol length

 Each pair of tones are separated by 1/2T. If there are

M of them,
BT=M/2T=M/2Tblog2M
-->BT=MRb/2log2M
 Contrast with MPSK
 Variation of bandwidth with M differs drastically
compared to MPSK

MPSK MFSK
BT=2Rb/log2M BT=MRb/2log2M

 As M goes up, MFSK eats up more bandwidth but

MPSK save bandwidth
 MFSK bandwidth efficiency
 Let’s compute ’s for MFSK
=Rb/M=2log2M/M bits/sec/Hz…MFSK
M 2 4 8 16 32 64
 1 1 .75 .5 .3 .18

 Notice bandwidth efficiency drop. We are sending

fewer and fewer bits per 1 Hz of bandwidth
 Notations
 Bandwidth efficiency measure
M  2 m # of symbols
m = log2 M bits/symbol
m log2 M
R=  bits/ sec
Ts Ts
Ts  symbol duration
Rs  symbol rate
1 Ts 1
Tb    bit length
R m mRs

R log2 M 1
 
W WTs WTb
 Bandwidth-limited Systems
 There are situations where bandwidth is at a
premium, therefore, we need modulations with large
R/W.

 Hence we need standards with large time-bandwidth

product

 The GSM standard uses Gaussian minimum shift

keying(GMSK) with WTb=0.3
 Case of MPSK
 In MPSK, symbols are m times as wide as binary.

 Nyquist bandwidth is W=Rs/2=1/2Ts. However, the

bandpass bandwidth is twice that, W=1/Ts

 Then

R log 2 M
  log 2 M bits/sec/H z
W WT s
 Cost of Bandwidth Efficiency
 As M increases, modulation becomes more
bandwidth efficient.

 Let’s fix BER. To maintain this BER while increasing

M requires an increase in Eb/No.
 Power-Limited Systems
 There are cases that bandwidth is available but
power is limited

 In these cases as M goes up, the bandwidth

increases but required power levels to meet a
specified BER remains stable
 Case of MFSK
 MFSK is an orthogonal modulation scheme.

 Nyquist bandwidth is M-times the binary case

because of using M orthogonal frequencies,
W=M/Ts=MRs

 Then

R log 2 M log 2 M
  bits/sec/Hz
W WTs M
 Select an Appropriate Modulation
 We have a channel of 4KHz with an available
S/No=53 dB-Hz

 Required data rate R=9600 bits/sec.

 Required BER=10-5.

 Choose a modulation scheme to meet these

requirements
 Minimum Number of Phases
 To conserve power, we should pick the minimum
number of phases that still meets the 4KHz
bandwidth

 A 9600 bits/sec if encoded as 8-PSK results in 3200

symbols/sec needing 3200Hz

 So, M=8
What is the required Eb/No?

S Eb R Eb
  R
No No No
Eb S
(dB)  (dB  Hz)  R(dB  bits / sec
No No
 13.2dB
 Is BER met? Yes
 The symbol error probability in 8-PSK is

 2Es  

Solve for Es/No PE  M   2Q  sin  
M 

 No 

 Solve for PE
Es Eb
 log 2 M   3  20.89  62.67
No N0

PE 2.2 10 5
BER    7.3  10 6
log 2 M 3
 Power-limited uncoded system
 Same bit rate and BER

 Available bandwidth W=45 KHz

 Available S/No=48-dBHz

 Choose a modulation scheme that yields the required

performance
Binary vs. M-ary Model

R bits/s M-ary Modulator

R
Rs  symbols / s
log 2 M

M-ary demodulator

S Eb Es
 R Rs
N o No No
 Choice of Modulation
 With R=9600 bits/sec and W=45 KHz, the channel is
not bandwidth limited
 Let’s find the available Eb/No

Eb S
(dB)  dB  Hz   R(dB  bit / s) 
No No
Eb
(dB)  48dB  Hz
No
 (10 log 9600)dB  bits / s
 8.2dB
 Choose MFSK
 We have a lot of bandwidth but little power -
>orthogonal modulation(MFSK)

 The larger the M, the more power efficiency but more

bandwidth is needed

 Pick the largest M without going beyond the 45 KHz

bandwidth.
 MFSK Parameters
 From Table 1, M=16 for an MFSK modulation
requires a bandwidth of 38.4 KHz for 9600 bits/sec
data rate

 We also wanted to have a BER<10^-5. Question is if

this is met for a 16FSK modulation.
 16-FSK
 Again from Table 1, to achieve BER of 10^-5 we need
Eb/No of 8.1dB.

 We solved for the available Eb/No and that came to

8.2dB
 Symbol error for MFSK
 For noncoherent orthogonal MFSK, symbol error
probability is

M 1  Es 
PE  M   exp 
2  2 No 
Es  Eb log 2 M
 BER for MFSK
 We found out that Eb/No=8.2dB or 6.61
 Relating Es/No and Eb/No

Es E
 log 2 M  b
No No

 BER and symbol error are related by

2 m 1
PB  m PE
2 1
 Example
 Let’s look at the 16FSK case. With 16 levels, we are
talking about m=4 bits per symbol. Therefore,

 With Es/No=26.44, symbol error prob. PE=1.4x10^-5--

>PB=7.3x10^-6 23 8
PB  4 PE  PE
2 1 15
 Summary
 Given:  Solution
 R=9600 bits/s  16-FSK
 BER=10^-5  required bw=38.4khz
 Channel bandwith=45 KHz  required Eb/No=8.1dB
 Eb/No=8.2dB
 Minimum Shift Keying (MSK)

MSK is a continuous phase-frequency shift keying;

Why MSK?
-- Exploitation of Phase Information besides frequency.
MSK (cont’d)
Representation of a MSK signal
MSK Transmitter
MSK Receiver
 Minimum Shift Keying (MSK)

2 +1 Time

Data
-1

 2

Time 
Phase

0
0 2Tb 4Tb 6Tb 8Tb Time

Phase
0
0 2Tb 4Tb 6Tb 8Tb
-
-

-2 -2

MSK possible phase transitions MSK phase transitions for data:

(00111000...)

 In MSK phase ramps up through 90 degrees for a binary one, and down
90 degrees for a binary zero.
 For GMSK transmission, a Gaussian pre-modulation baseband filter is used
to suppress the high frequency components in the data. The degree of
out-of-band suppression is controlled by the BT product.

237
 Gaussian Minimum Shift Keying (GMSK)
 Gaussian Minimum Shift Keying (GMSK) is a form of
continuous-phase FSK in which the phase change is
changed between symbols to provide a constant
envelope. Consequently it is a popular alternative to
QPSK
 The RF bandwidth is controlled by the Gaussian low-
pass filter bandwidth. The degree of filtering is
expressed by multiplying the filter 3dB bandwidth (B)
by the bit period of the transmission (T), i.e. by BT
 GMSK allows efficient class C non-linear amplifiers to
be used
 GMSK (cont’d)

 GMSK is a form of continuous-phase FSK, in which the phase is

changed between symbols to provide a constant envelope.
 The RF bandwidth is controlled by the Gaussian low-pass filter
bandwidth.
 The degree of filtering is expressed by multiplying the filter 3dB
bandwidth by the bit period of the transmission, ie. by BT.
 As bandwidth of this filter is lowered the amount of intersymbol-
interference introduced increases.
 GMSK allows efficient class C non-linear amplifiers to be used,
however even with a low BT value its bandwidth efficiency is less
than filtered QPSK.
- GMSK generally achieves a bandwidth efficiency less than 0.7
bits per second per Hz (QPSK can be as high as 1.6 bits per
second per Hz).
 GMSK Signals

GMSK conceptual transmitter

 In MSK , the BT is infinity
and this allows the square
bit transients to directly
modulate
Tx
the VCO.
 In GMSK, low values of BT
GMSK, MSK create significant
BT=0.5
NRZ intersymbol interference
GMSK DATA
1-2 GLPF (ISI). In the diagram, the
BT=0.3 portion of the symbol energy
VCO  acts as ISI for adjacent
  symbols.
 If BT is less than 0.3, some
Time
form of combating the ISI is
2T T 0 T 2T
required.

GMSK Pulse Shapes and ISI

240
 GMSK Spectra

0
Data Rate: 8192 bps

-10
QPSK
MSK
-20
Power (dB)

-30

-40

GMSK BT=0.3
-50

-60
GMSK BT=0.5
-70
0 16384 32768 49152 65536
Frequency (Hz)
GMSK has a main lobe 1.5 times that of QPSK.
GMSK generally achieves a bandwidth efficiency less than 0.7 bits
per second per Hz (QPSK can be as high as 1.6 bits per second per
Hz).

241
 Shannon-Hartley Capacity Theorem

For error free communication, it is possible to define the capacity

which can be supported in an additive white gaussian noise
(AWGN) channel.
fb/W = log2(1 + Eb fb /W)

where fb = Capacity (bits per second)

W = bandwidth of the modulating baseband signal (Hz)
Eb = energy per bit
 = noise power density (watts/Hz)

thus Ebfb = total signal power

W = total noise power
fb/W = bandwidth efficiency (bits per second per Hz)
 Comparison of Modulation Schemes

This graph shows that bandwidth efficiency is traded off against power
efficiency.
 MFSK is power efficient, but not bandwidth efficient.
 MPSK and QAM are bandwidth efficient but not power efficient.

 Mobile radio systems are bandwidth limited, therefore PSK is more

suited.

bits/s/Hz vs. Eb/ for Probability of Error = 10-5

taken from “Principle of Communication Systems”
Taub & Schilling, page 482

6-
 Comparison of Modulation Types
Modulation Bandwidth Log2(C/B) Error-free Eb/No
Format efficiency C/B
16 PSK 4 2 18dB
16 QAM 4 2 15dB
8 PSK 3 1.6 14.5dB
4 PSK 2 1 10dB
4 QAM 2 1 10dB
BFSK 1 0 13dB
BPSK 1 0 10.5dB

244
 Spectral Efficiencies in practical radios
 GSM- Digital Cellular
 Data Rate = 270kb/s, bandwidth = 200kHz

 Bandwidth Efficiency = 270/200

=1.35bits/sec/Hz
 Modulation: Gaussian Minimum Shift Keying (FSK

with orthogonal frequencies).

 “Gaussian” refers to filter response.

 IS-54 North American Digital Cellular

 Data Rate = 48kb/s, bandwidth = 30kHz

 Bandwidth Efficiency = 48/30

=1.6bits/sec/Hz
 Modulation: p/4 DPSK
 Coherent Reception

An estimate of the channel phase and attenuation is recovered. It is then

possible to reproduce the transmitted signal, and demodulate. It is
necessary to have an accurate version of the carrier, otherwise errors are
introduced. Carrier recovery methods include:

 Pilot Tone (such as Transparent Tone in Band)

 Less power in information bearing signal
 High peak-to-mean power ratio
 Pilot Symbol Assisted Modulation
 Less power in information bearing signal
 Carrier Recovery (such as Costas loop)
 The carrier is recovered from the information signal
 Differential Reception

 In the transmitter, each symbol is modulated relative to the previous

symbol, for example in differential BPSK:
 0 = no change 1 = +180o
 In the receiver, the current symbol is demodulated using the previous
symbol as a reference. The previous symbol acts as an estimate of the
channel.
 Differential reception is theoretical 3dB poorer than coherent. This is
because the differential system has two sources of error: a corrupted
symbol, and a corrupted reference (the previous symbol).
 Non-coherent reception is often easier to implement.
 Modulation Summary

 Phase Shift Keying is often used, as it provides a highly

bandwidth efficient modulation scheme.
 QPSK, modulation is very robust, but requires some form of
linear amplification. Alternatives (e.g. Offset QPSK and /4-
QPSK) can be implemented, and reduce the envelope variations
of the signal.
 High level M-ary schemes (such as 64-QAM) are very
bandwidth-efficient, but more susceptible to noise and require
linear amplification.
 Constant envelope schemes (such as GMSK) can be employed
since an efficient, non-linear amplifier can be used.
 Coherent reception provides better performance than
differential, but requires a more complex receiver.
Thanks