Telecommunications B (EE2T21)

Lecture 4 overview:

Signal-to-Noise Ratio after detection of analog signals:

* Detection of PM and FM signals
* Maximum achievable output SNR

Modulation techniques for digital signals - binary schemes:

* Amplitude Shift Keying (ASK)
* Binary Phase Shift Keying (BPSK)
* Frequency Shift Keying (FSK)

EE2T12 Telecommunications B
Dr.ir. Gerard J.M. Janssen
March 2, 2022

1
Lectures & Work-instructions Telecommunications B
Telecommunication Techniques

Lectures:
Monday 2-5, 9-5 1st+2nd hour
Tuesday 3-5 1st+2nd hour
Wednesday 11-5 3rd+4th hour

Q&A/Working lectures:
Wednesday 4-5 3rd+4th hour
Thursday 12-5 5th+6th hour

2
 Phase- and Frequency modulation
The transmitted signal for angle modulation:
s (t )  Re{g (t )e jct }  Re{ Ac e j ( t ) e jct }  Ac cos[c t   (t )]

Phase Modulation (PM):

 (t )  D p m(t )  so  (t ) is proportional with the

information signal m(t).
 D p = phase deviation constant [rad/V]

Frequency Modulation (FM):

t
 (t )  D f  m( ) d   so  (t ) is proportional with the integral of
 information signal m(t).
 D f = frequency deviation constant [rad/V.s]
3
SNR after detection: angle modulation

4
 SNR: PM modulation (1)
PM signal: g s (t )  Ac e j s ( t ) with  s (t )  D p m(t ),  p    D p max{| m(t ) |}

Received signal + noise: Phase deviation Modulation

constant index

gT (t )  [ g s (t )  g n (t )]  gT (t ) e jT ( t )
 Ac e j s ( t )  Rn (t )e jn ( t )

Phase detector with gain K: r0 (t )  K arg{gT (t )}  KT (t )

For SNRin  1  Ac  Rn (t ) :
r0 (t )  KT (t )
 Rn (t )sin[ n (t )   s (t )] 
 K  s (t )  
 Ac 
5
SNR: PM modulation (2)

 yn (t )

yn (t ) is the noise in
quadrature (perpendicular)
to the signal phasor gs (t ) .

6
 SNR: PM modulation (3)

Without modulation:  s (t )  0 and

Rn (t ) yn (t )
r0 (t )  K sin  n (t )  K  K n'  0 for Ac  
Ac Ac

This is the "quieting–effect" or noise suppression effect when there is

a strong carrier ( SNRin  1) at the input of the PM/FM detector
(also without modulation: silence!).

At a certain time t,  s (t ) is deterministic (but unknown), but  n (t )   s (t )

is random and uniformly distributed at any moment.  In the
stochastic noise model, we don't need to take  s (t ) into account.
For large SNR, the signal and noise terms are independent.

7
 SNR: PM modulation (3)

For SNRin  1, the detector output signal can be approximated by:

r0 (t )  s0 (t )  n0 (t ) For SNRin 
/ 1 , the phase of the
noise becomes dominant.

K
with s0 (t )  K s (t )  KD p m(t ) and n0 (t )  yn (t )
Ac
yn (t ) is the noise in quadrature
(perpendicular) to the signal
vector g s (t ).

The noise PSD of the (two-sided) baseband spectrum is:

2
 K
K2  A2 N 0 |f |  B2T
Pn0 ( f )  2 Pyn ( f )   c  a white noise spectrum
Ac 0
 |f |  B2T
8
SNR: PM modulation (4)

9
 SNR: PM modulation (5)
In the receiver baseband bandwidth Bbb  B (B = signal bandwidth
in Couch)
m (t )  s0 (t )  n0 (t )

with signal power: s02 (t )  K 2 D p2 m 2 (t )

Bbb 2
2 K N 0 Bbb
and noise power: n02 (t )   Pn0 ( f )df  2
 Bbb
A c

Now the SNR at the detector output is given by:

2
s (t )
0
Ac2 D p2 m 2
SNRout  2

n (t ) 2 N 0 Bbb
0

10
 SNR: PM modulation (6)

Starting with
2
s (t )
0
Ac2 D p2 m 2
SNRout  2

n (t ) 2 N 0 Bbb
0

p
and using:  p  D pV p  D p 
Vp
with D p the phase deviation constant,  p the phase modulation
index and V p the peak-value of | m(t ) | , we find for matched
receiver and signal bandwidth Bbb  B :

Ac2  p2 (m 2 / V p2 )
SNRout 
2 N0 B

11
 SNR: FM modulation (1)
j s ( t )
For an FM-modulated signal: g s (t )  Ac e

t
with  s (t )  D f  m( )d 
 Frequency deviation
constant

The complex envelope of signal + noise:

gT (t )  [ g s (t )  g n (t )]  gT (t ) e jT ( t ) yn (t ) is the noise in quadrature
y (t ) (perpendicular) to the signal
with T (t )   s (t )  n for SNRin >>1. vector g s (t ).
Ac
FM-detector: output signal is proportional to the derivative of the phase
of the total signal:
K d arg{gT (t )} K dT (t )
r0 (t )  
2 dt 2 dt
where K = FM-detector gain
12
 SNR: FM modulation (2)
For SNRin  1 , the output signal can be approximated by:

r0 (t )  s0 (t )  n0 (t ) thus s0 (t ) and n0 (t ) become independent.

K d s (t ) KD f
with s0 (t )   m (t )
2 dt 2
K dyn (t )
and n0 (t ) 
2 Ac dt
The noise PSD of the (two-sided) baseband spectrum is:
2
2
 K 
 K  2 A  N 0 f 2
|f |  B2T
Pn0 ( f )    2 jf Pyn ( f )   c
 2 Ac  
0 |f |  B2T
 a quadratic noise spectrum
13
SNR: FM modulation (3)

14
 SNR: FM modulation (4)
In the receiver baseband bandwidth Bbb  B (B = signal bandwidth
in Couch)
m (t )  s0 (t )  n0 (t )
2
2  KD f  2
with signal power: s (t )  
0  m (t )
 2 
Bbb 2
2 K 
and noise power: n02 (t )  P ( f ) df  N B 3

 Bbb
n0
3  A  0 bb
 c

Now the SNR at the detector output is given by:

2 2 2
2
s (t ) 3 A [ D /(2  B )] m B 2
SNRout  0
2
 c f  2
n (t )
0
2 N 0 Bbb Bbb
15
 SNR: FM modulation (5)

Starting with
2 2 2
2
s (t )
0
3 Ac [ D f /(2 B )] m B2
SNRout    2
n (t )
2
0
2 N 0 Bbb Bbb

D f Vp F Df f
and using:   f  
2 B B 2 B V p
with D f the frequency deviation constant,  f the frequency
modulation index and V p the peak-value of | m(t ) | , we find
for matched receiver and signal bandwidth Bbb  B :

3 Ac2  f2 (m 2 / V p2 )
SNRout 
2 N0 B Ideal case, when the
receiver is matched
to the signal.

16
 SNR: PM- and FM modulation (1)
With Carson's bandwidth BT  2(   1) B, SNRin becomes:

Ac2 / 2 Ac2
SNRin  
N 0 BT 4 N 0 (   1) B

and we find the following relation between SNRout and SNRin

2
SNRout m
PM:  2 p2 (  p  1) 2
SNRin Vp
The factor 3 difference is due
to the different baseband noise
PSD: flat versus quadratic.
2
SNRout m
FM:  6 f2 (  f  1) 2
SNRin Vp

17
 SNR: PM- and FM modulation (2)
Comparing SNRout with the SNR in baseband:

Ps Ac2 / 2
SNRbaseband  
N0 B N0 B
2
SNRout m
PM:   p2 2
SNRbaseband Vp Note: here the assumption is
made that the receiver
2 bandwidth is matched to the
SNRout 2 m signal bandwidth: Bbb  B.
FM:  3 f 2
SNRbaseband Vp

By choosing a higher phase/frequency deviation, we can obtain a

SNRout higher than the SNR in baseband:

Trade-off between transmission power and bandwidth!!!

18
 SNR: PM- and FM modulation (3)
Gain PM: due to the maximum value of

 p m(t ) / V p  D p m(t )     p  D pV p  
the maximum achievable gain compared to baseband for sine-wave
modulation is limited to:
2
m
 p2 2   2 / 2  6.9 dB
Vp
Gain FM: for FM there is no such limitation on f . For sine-wave
modulation with: m 2 (t ) / V p2  0.5 this gain is:

2 With increasing f the SNRin

2 m 3 2
3 f 2   f will decrease due to increasing
Vp 2 transmission bandwidth BT .

Iff SNRin is above the detection threshold!

19
SNR: FM detection threshold
Ac2 / 2
SNRbaseband 
N0 B
Ac2 / 2
SNRin 
N 0 BT
Ac2

4 N 0 (  f  1) B

Thus we find:

SNRbaseband  2(  f  1) SNRin
and
SNRbaseband _ thr
SNRin _ thr 
2(  f  1)
20
 Exercise output SNR PM-modulation (1)

A PM signal with sine wave modulation m(t )  sin 2000 t is

received with a power S R  Ac2 / 2  40 dBm , N 0  90 dBm/Hz ,
the baseband bandwidth of the receiver is Bbb  2 kHz and the
phase deviation constant is D p   / 2 rad/V.

Determine:
- SNRout
- SNRout for the receiver baseband bandwidth matched to
the information signal bandwidth
- maximum achievable SNRout

21
 Exercise output SNR PM-modulation (2)
2 2 2 2 2
Ac2 D p m Ac2  p (m / V p )
1. SNRout  
2 N 0 Bbb 2 N 0 Bbb
 40 dBm  90 dBm/Hz  10log10 Bbb dBHz  10log10  p2 ( m 2 / V p2 )
2
 40  90  10log10 2000  10log10 8  17.9 dB

2. With optimal baseband bandwidth Bbb = B, we obtain a gain

of 3 dB  SNRout = 20.9 dB

3. Maximum SNR is obtained with  p  V p D p   . Then:

2
SNRout  40  90  10log10 1000  10log10 2  26.9 dB

22
 Exercise output SNR FM-modulation (1)

An FM signal with sine wave modulation m(t )  sin 2000 t is

received with a power S R  Ac2 / 2  40 dBm , N 0  90 dBm/Hz ,
the baseband bandwidth of the receiver is Bbb  2 kHz and the
modulation index is  f   / 2.

Determine:
- SNRout
- SNRout for the receiver baseband bandwidth matched to
the information signal bandwidth
- maximum achievable SNRout

23
 Exercise output SNR FM-modulation (2)
2
B2
3 Ac2 (V p D f / 2 B ) 2 Vm2 2
Bbb
p
1. SNRout 
2 N 0 Bbb
2
B2
3 Ac2  f2 Vm2 2
Bbb
p

2 N 0 Bbb
2
 10log10 3 dB  40 dBm  90 dBm/Hz  10log10  2 
2
 3 dB  10log10  12  dB  10log10 Bbb dBHz
 4.77  40  90  3.92  3  6  33  16.7 dB

2. With optimal baseband bandwidth Bbb = B, we obtain a gain

of 3x3 dB = 9dB  SNRout = 25.7 dB

3.  f can be increased to much larger values than , however,

SNRin  SNRthr . For  f    SNRout = 31.7 dB
24
 Output SNR for ideal demodulation (1)

In an ideal detection system:

no loss of capacity by detection.
Cin  Cout

Shannon: BT log 2 1  SNRin   B log 2 1  SNRout 

BT / B
 SNRout  1  SNRin  1

B
With: SNRin  SNRbaseband
BT
BT / B
  B  
we find: SNRout  1    SNRbaseband  1
  BT  

25
Ideal output SNR for analog modulations (2)

26
 Where do we find digital modulation?

Digital modulation techniques are applied in:

- computer modems: analog as well as in ADSL, WiFi

- mobile communications: GSM, GSM-EDGE, UMTS,
LTE, WiMAX, 5G
- digital audio and video broadcasting
(DAB, DVB: e.g. digitenne)
- digital cable television: DVB-C
- etc, etc.

27
Binary modulation schemes

28
 On-Off Keying (1)

On-Off Keying (OOK) or Amplitude Shift Keying (ASK)

Carrier is switched on or off depending on the bit to be send

 Ac m(t ) cos c t with m(t )  {0,1} Unipolar

OOK-signal: s (t )  
 2 Ac [1  m '(t )]cos c t with m '(t )  {1,1}
1
Polar

 Ac m(t ) with m(t )  {0,1}

Complex envelope: g (t )  
1
 2 Ac [1  m '(t )] with m '(t )  {1,1}

m(t ) For an average power m 2 (t )  1 ,

m(t ) {0, 2} with equal probability.  Ac


0 Tb Rb  1/ Tb

29
 On-Off Keying (2)
Power spectral density (PSD) of OOK

1. For rectangular pulses:

Ac2
Pg ( f )  [ ( f )  Tb sinc 2 fTb ] for m 2 (t )  1
2
Ac2
Ps ( f )  [ ( f  f c )  Tb sinc 2 ( f  f c )Tb ]
8
Ac2
+ [ ( f  f c )  Tb sinc 2 ( f  f c )Tb ]
8
What is the transmission bandwidth BT ?

The 0-0 bandwidth: BT _ 00  2 Rb  B  Rb , BT  2 B

30
Power spectral density OOK, BPSK

31
 Power spectral density RC-pulses

2. For raised-cosine pulses:

 in baseband: B  12 (1  r ) Rb  (1  r ) f 0
f
f 0  Rb / 2, r  : roll-off factor
f0
 in bandpass: BT  (1  r ) Rb
32
Detection of OOK signals

33
Binary modulation schemes

34
 Binary Phase Shift Keying (1)

In Binary Phase Shift Keying (BPSK) the carrier phase is switched

 
between   ,  depending of the bit to be sent.
2 2 Digital phase modulation

In general: s (t )  Ac cos[c t  D p m(t )]

m(t ) {1,1} is a
polar baseband signal
Phase deviation constant

 Ac cos[ D p m(t )]cos c t  Ac sin[ D p m(t )]sin c t

 Ac cos D p cos c t  Ac m(t )sin D p sin c t
BPSK

Carrier component Data component 2-PSK

  2 Ac


jD p m ( t )
Complex envelope: g (t )  Ac e

35
 Binary Phase Shift Keying (2)

2 2 Dp
The digital modulation index is defined as: h  
 
Where 2 [rad] is the maximum (peak-peak) phase deviation per symbol time.

For maximum power in the data component (i.e. no power in the

carrier component) :

  D p   h 1
2
and s (t )   Ac m(t )sin c t 

or by adding a phase shift of  / 2 : s (t )  Ac m(t ) cos c t

 Ac 


36
 Binary Phase Shift Keying (3)
By changing D p we can choose for a kind of pure DSB-SC with a carrier
component, which however is 90o out of phase with the data component.

For BPSK, D p   / 2 and the complex envelope is:

g (t )  jAc m(t ) (or g (t )  Ac m(t ) for a phase shift of  / 2)

and the power spectral density is

Pg ( f )  Ac2Tb sinc 2 ( fTb )

Ac2Tb
Ps ( f )  sinc 2 ( f  f c )Tb  sinc 2 ( f  f c )Tb 
4
For data detection a synchronous detector is required. When a carrier
component is available, a Phase Locked Loop (PLL) can be used.
Otherwise a Costas-loop or a squaring-loop is needed.
37
Power spectral density OOK, BPSK

38
Detection of BPSK signals

39
 Binary Phase Shift Keying (4)

Because BPSK is a constant amplitude signal, envelope detection is not

possible. Product detection has to be used, which first requires carrier
phase recovery for phase synchronization.

1. Squaring-loop: see par. 5.4

40
 Binary Phase Shift Keying (5)
2. Costas-loop

VCO = voltage controlled oscillator

LPF = low-pass filter

41
 Differential Phase Shift Keying
In Differential Phase Shift Keying (DPSK), the data signal is first
differentially encoded and successively transmitted using BPSK.
In the detector no carrier recovery is required, since we can use
the phase of the previous symbol as a reference.
The DPSK-detector combines:
1. “coherent” detection
2. differential decoding

42
Binary modulation schemes

43
 Frequency Shift Keying (1)

In Frequency Shift Keying (FSK) the carrier frequency is switched

between { f1 , f 2 } dependent of the bit to be sent.
Digital frequency modulation

Two cases:
1. Discontinuous phase FSK:

 Ac cos[2 f1t  1 ] for a "1"

s (t )  
 Ac cos[2 f 2t   2 ] for a "0"

44
Frequency Shift Keying (3)

45
 Frequency Shift Keying (2)
2. Continuous phase FSK (CP-FSK):
t
s (t )  Re{g (t )e jct }  Ac cos[c t  D f  m( ) d  ]


with
t
g (t )  Ac e j ( t ) where  (t )  D f  m( ) d 


Digital baseband signal {-1, 1}

Continuous, also when m(t) discontinuous.

46
Example FSK: computer modem

47
 Frequency Shift Keying (3)
For FSK with Rb  1/ Tb we find: 2  2FTb

Digital modulation index:

2 2F
 h  2FTb 
 Rb

48
 Bandwidth of FSK signals (1)

The transmission bandwidth of FSK signals, based on the

“null-to-null” bandwidth of the signal spectrum, is now found as:

- for rectangular pulses and B  Rb is the 1st null bandwidth

in baseband:
BT  2F  2 Rb

- for raised cosine pulses, with absolute baseband bandwidth

B  12 (1  r ) Rb , the absolute bandpass bandwidth is given by:

BT  2F  Rb (1  r )

49
Bandwidth of FSK signals (2)

50
Power spectrum of FSK signals (1)

The power spectral density of random

data FSK is difficult to derive. Except
for some special cases, a closed form
expression cannot be obtained and
we have to rely on simulation results.

51
Power spectrum of FSK signals (2)
 '  2  F  Tb   for h  1
i.e. a  rad rotation during a
symbol time, either clockwise or
counter clockwise depending on
the data symbol.

52
Detection of FSK signals

Polar signal

53
 Basic techniques binary modulation

We have seen that the basic modulation techniques

for binary signals have a direct analog counterpart:

1. OOK, ASK  AM

2. BPSK  DSB, PM

3. FSK  FM

54