Modulation and Detection

 Analog and digital modulation

 Modulation is the conversion from baseband to passband
 Vary the amplitude or/and phase of the carrier : x(t)  a(t) cos [ωc t  θ(t)]
 Demodulation is the reverse process

Note : ωc t  θ (t) is the total phase and θ (t) the excess phase.
dθ (t) dθ (t)
ωc  is the total frequency and is the excess frequency.
dt dt 1
Important Attributes of Modulation
Modem (Modulation  Demodulation) are compared on the basis of :

Signal Quality (SNR)

 Channel loss
 Tolerance to noise
 Tolerance to interferer s

Metrics for assessing a modulation scheme

 Spectral efficiency : bandwith of modulated signal compared to baseband
 Power efficiency : linear power amplifier are less efficient

Channel will be assumed to be corrupted by additive white Gaussian noise (AWGN).2

Amplitude Modulation

Types of AM modulation :
 Single sideband suppressed carrier (SSB)
 Double sideband suppressed carrier (DSB - SC)
 Double sideband large carrier (DSB - LC)

DSB - LC AM modulation :
x AM  Ac [ 1  m xBB(t)] cos ωc t with m the amplitude modulation index

Carrier not suppressed hence large carrier AM

3
-ω1 ω1

 exp  j1t   exp  j1t   exp  jc t   exp  jc t  exp j  1  c  t   exp j  1  c  t   exp j 1  c  t   exp j 1  c  t 
cos1t   cosc t        
 2 2 4

444
AM Detectors
Type of AM Dectectors:
 Coherent d etection: (requires a phase lock of the LO on the carrier)
SNR out 2 m 2 x BB
2
2
 
SNR in 1  2 m 2 x BB
2
3
Ac2 2
SNR out  SNR in 2 m x2 2
BB  2 m 2 x BB
2
for mx BB  1
2N 0B
with B  2 f BB, max , f BB, max the bandwidth of x BB (t) .
 Non-cohere nt detecti on: peak d etector
- Peak detector does not work for 1  mx BB  0
- Coherent detection and non - coherent detection have the same SNR out for SNR in  15
dB. Typically 25 dB is targeted for quality AM signals.

Attributes of AM mod ulation:

 Succeptibl e to noise.
 Requires a linear amp in the transmitt er. 5
Coherent Peak detector

6
Peak Detector Waveforms

baseband signal

modulated RF signal

rectified RF signal after diode

peak detector output

7
Phase and Frequency Modulation

Phase modulation :
xPM  Ac cos [ωc t  mxBB(t)] with m the phase modulation coefficient

Frequency modulation :

xFM  Ac cos ωc t  m  xBB(τ )dτ  with m the frequency modulation coefficient

 t

 - 
dθ (t )
Note : is proportional to xBB .
dt

8
9
FM Modulation and Demodulation
Modulation :
A varactor (voltage controlled capacitor) can be used in a LC resonator to tune
the frequency realizing a VCO.

Demodulation :
A high-pass RC filter can be used as differentiator to convert FM to AM

 t

v out   Ac R1C1 c  mxBB (t )   sin ct   xBB ( )d 
  
If c  mxBB (t ) always remains positive an envelope detector can then be used to
recover the amplitude modulation.

A limiter is required before the differentiator to prevent variation in A c to corrupt the demodulation.
10
11
Bandwidth and Frequency Dependence
Let the baseband be : xBB(t)  Am cos ωmt
For mAm  ωm (Narrow band FM) we have
mAm Ac mAm Ac
x FM (t)  Ac cos c t  cos(c  m )t  cos(c  m )t
2m 2m
Bandwidth : c  m , c  m  (if m is the maximum frequency of the baseband xBB ).
Note the frequency dependence (in m ) of the sidebands' amplitude.

For mA m  m (Broad band FM) we have

n 
mAm
xFM (t)  Ac  J n(β ) cosωc  nωm  t with β  modulation index 
n  - ωm
ωm
98% Power Bandwidth : (2   1)BBB with BBB  the baseband bandwidth.
2π 12
13
SNR for FM Modulation
SNR out 3 2
 
SNR in 2
SNR out
In book given as
SNR in
2
x BB (t )
6    1
2
2
depending on the detector
VP

with VP the peak value of x BB (t). Threshold Effect

See reference in textbook for further (10 - 15 dB) improvemen t using pre - emphasis
and de - emphasis filter to compensate for the decrease of the high - frequency
base - band components .
FM Modulat ion Attrib utes:
 The performanc e strongly depends on  (trade - off between signal quality
and bandwidth efficiency )
 FM signal can be processed by a non - linear amplier
Jayanta Mukherjee
(power efficient) 14
 mean
  
squared noise power

n(t) 2
p(  (t ))  p ( A(t ))
 4A 4 
PSD of phase PSD of amplitude
var iation
 var iation

Amplitude of signal
Jayanta Mukherjee 15
"Amplitude to phase noise conversion in electronic circuits" Edoardo Milotti, PHYSICAL REVIEW E, Jan 1998
Digital Modulation
The digital counter parts of AM, PM and FM are
 Amplitude shift keying (ASK)
 Phase shift keying (PSK)
 Frequency shift keying (FSK)
Initial observations :
 PSK and FSK nd wider usage than ASK due to higher noise immunity.
 Comparison of digital modems will again be based upon :
- signal quality
- spectral efficiency
- power efficiency
 Signal quality is measured by the bit error rate (BER) dened as the average
erroneous bits received at the output divided by the total number of bit received
in a unit time. 16
Performance Advantages of Digital
Modulation
1. Digital transmission produces fewer data errors than analog transmission:

A. Data integrity & noise immunity: Easier to detect and correct information-bearing data errors, since transmitted data is binary (1’s & 0’s :
only two distinct values) .

B. Error coding is used to detect and correct digital transmission errors.

C. Regenerative capability: Regenerative digital repeaters placed along the transmission channel can detect
a distorted digital signal and retransmit a new, clean digital data signal. These repeaters minimize the
accumulation of noise and signal distortion along the transmission channel.

2. Permits higher transmission data rates: Economical to build transmission links of very high bandwidth. Optical fiber designed for digital transmission.

3. Better spectral efficiency: Effective use of limited frequency resources (narrow bandwidth) to send a large amount of data.

4. Security & privacy: Enables encryption algorithms in information-bearing digital bit stream signals. Deters phone cloning and eavesdropping.

5. Easy to multiplex multiple sources of information: Voice, video and data in a single transmission channel, since all signals are made up of 1’s and 0’s.

6. Easy to integrate computer/communication systems.

7. Digital equipment consumes less DC power in a smaller physical size. 17

Disadvantages
1. More Bandwidth Needed:

A. Transmission of digitally encoded analog signals requires significantly

more bandwidth than simply transmitting the original analog signal.

2. Circuit complexity:

A. Analog signals must be converted to digital pulses prior to transmission

and converted back to their original analog form at the receiver:
Additional encoding/decoding circuitry needed.

3. Synchronization:

A. Requires precise time synchronization between the clocks in the

transmitter and in the receiver.
18
19
Binary and M-ary Signaling
A binary digital baseband signal xBB is synthesized from the binary data bn using :
xBB   bn p(t-nTb )
n

where b n can take either the value 0 and 1 or 1 and - 1 in the interval [nTb , (n  1 )Tb ]
p(t) is the pulse shape (e.g, rectangular pulse).

A M - ary digital baseband signal xBB is synthesized by first converting the binary
sequence in M different levels using a D/A converter.

 require higher amplitude resolution in detector

 symbol rate is decreased slightly due to D/A conversion.

20
21
Basis Function for Modulation
Each symbol in a digitally modulated waveform is presented by a linear
superposition of orthogonal basis functions :
N TS 0 if m  n
xRF,mod   αk φk (t ) with  φm (t ) φn (t ) dt  δnm 
k 1
0
1 if m  n

where TS is the symbol period and N the basis dimension.

Example # 1 : binary FSK signal
 Ac cos ω1t if bn  0
xFSK(t)  
 Ac cos ω2t if bn  1
can be represented by :

φ   cos(ω1t ) 
x FSK (t)  α1 α2   1   α1 α2   
φ2  cos(ω2 t )

( A 0) if b n  0
with α1 α2  c
 (0 Ac ) if b n  1 22
Signal Constellation
The signal constellation is a plot of all possible values of

Example # 2: Binary ASK signal Example # 3: Binary PSK signal

 Ac cos 1t if bn  1  Ac cos 1t if bn  1

x ASK (t )   xPSK (t )  
0 if bn  0  Ac cos 1t if bn  0

23
Impact of Noise and Decision Boundary
Decision
boundary

2
0 1

PSK

The distance between the points on the constellation is an indication of the robustness of the modulated signal
to noise.

24
Pulse Detection

25
26
27
Optimal Detection with a Matched Filter

x(t)

Tb

Sampling

28
Optimal Detection for Square Pulses

x(t)
p(t)

Tb

Sampling

29
Example with a Train of Square Pulses

30
Implementation of Matched Filter with a Correlator

y(Tb)

31
Correlation Receiver for 2D Signal Space

TS TS
y1 (TS )   x (t ) φ1 (t ) dt and y2 (TS )   x (t ) φ2 (t ) dt
0 0

Jayanta Mukherjee 32
Example: Coherent FSK Detector

33
Carrier Synchronization for Coherent Receiver
Typically phase error of less than 45 degrees required since 90 degrees gives incorrect reading.

An envelop detector can be used for ASK and FSK with a penalty in BER.

Method used for synchronization:

• Transmit a low level carrier for phase locking LO.

• Carrier recovery from received signal using a phase locked loop circuit.

Jayanta Mukherjee 34
Non-Coherent FSK Detector

Bit error rate is higher 35

99% Bandwidth and ACPR Definition

36
Coherent Binary Receiver for Modulated Signals

37
Integrator Output

38
 1   x  A1 2 
exp  

2 2
 2  2


1  x  A2 2 
exp  

2 2
 2 2


of bit ‘0’ being detected as ‘1’ is given by,


1  u2 
2 
A1  A 2
exp  du
 2
2

39
Variance of Output Noise

Average value of noise at integrator o/p is always zero (in time domain)

 y(Tb )   n(t ) p1 (t )  p2 (t )dt   n

(t )  p1 (t )  p2 (t ) dt  0
Tb Tb

0 0
0

Variance  2 of the noise at the integrator o/p :

 2  y 2 (Tb )  y (Tb )  y 2 (Tb )   n(t ) p1 (t )  p2 (t ) dt  n(t ' ) p1 (t ' )  p2 (t ' ) dt '

2 Tb Tb

0 0

1
n(t) n(t')  p1 (t )  p2 (t ) p1 (t ' )  p2 (t ' ) dt dt '   N 0  (t  t ' )  p1 (t )  p2 (t ) p1 (t ' )  p2 (t ' ) dt dt '
Tb Tb Tb Tb
  
0 0 0 0 2
1 1
N 0   p1 (t )  p2 (t ) 2dt  N 0 Ed
Tb

2 0 2
where we used the autocorrelation property of white noise at temperature T0 :
1 1
R(t-t')  n(t) n(t')  k T0 (t  t')  N 0  (t  t')
2 2 Jayanta Mukherjee 40
Variance of Output Noise (…Contd)
The variance of the output noise is given by,
1
 2  N 0 Ed (N 0  kT for white noise)
2
BER is calculated as follows,
2
 Ed  Ed2 Ed2 Ed
    
 2  4 2 4  1 N E 2N0
0 d
2
 Ed   Ed 
Q   Q    BER

 2   2N0 
 p1 (t )  p2 (t ) 2dt
Tb
Ed  A1  A2  
0

Jayanta Mukherjee 41
Peak and Average Bit Energy

 p1 (t )  p2 (t )2 dt
Tb
Ed  A1  A2  
0

1 2
 2 Ac Tb BASK

Ed   2 Ac2Tb BPSK
 A2T BFSK
 c b


S
Eb  Signal Power  Tb  S  Tb 
Rb

Jayanta Mukherjee 42
BER for Coherent Binary Digital Modulation

BER

Bandwidth
efficiency

43
Additional Attributes of BPSK

44
Additional Attributes of BFSK

f1=2f2

45
Comparison of Coherent and Non-Coherent BFSK
Demodulation

46
M-ary Digital Modulation
Principle:
Transmit M = 2n symbols over each symbol interval TS.

• Each symbol is realized with n bits.

• The symbol interval is TS = nTb with Tb the bit interval.
• The symbol rate is RS = 1/TS = Rb/n with Rb the bit rate

Examples:

• M-ary PSK: The bandwidth efficiency Rb/BT (Bit/sec/Hz) increases as log2(M) with increasing M but at the cost of a
degradation in BER.

• M-ary FSK: The BER decreases with increasing M but at the cost of increase in bandwidth.

Jayanta Mukherjee 47
BER for MFSK and MPSK

48
Jayanta Mukherjee
Quadrature Modulation

b1 b2 x(t)
-1 -1  3 
2 cos ct  
 4 

-1 1  3 
2 cos ct  
 4 

1 -1  
2 cos ct  
 4
1 1
 
2 cos c t  
 4

Jayanta Mukherjee 49
QPSK Constellation and BER
Zero crossing possible when one
constellation changes to another

Can cause PLL to lose synchronization

 2 Eb 
Pe  Q  

 N0 
• Bit rate is increased by 2 for same BW. Alternatively BW is reduced by 2 for same bit rate

Jayanta Mukherjee 50
Offset QPSK Modulation
• A delay of Tb reduces the phase shifts to 90o at the most. First bit needs
To be explicitly
• Cannot be used with differential encoding and therefore requires a coherent demodulator.
signaled
• BW is double of QPSK.

b1

b2

x(t)

 2 Eb 
Pe  Q  

 N0 
Jayanta Mukherjee 51
π/4-QPSK Modulation

Jayanta Mukherjee 52
Non coherent PSK detection
LPF
Q-arm m(t ) cosi  0 
m(t ) when i  0 

Input signal
2 cosct  0  Keep adjusting 0
till 0=i
m (t ) cosc t  i  ΔVctrl

m 2 (t )
sin 2i  0 
2
sin 2i  0 
+π/2 
2
 0 (when i  0 )
 2 sin c t  0 

I-arm m(t ) sin  i0 

LPF
Costa’s Loop for BPSK LPF

Similar architecture is possible for QPSK 53

Differential Phase Shift Keying

Jayanta Mukherjee 54
 To modulator

From demodulator
after non coherent
detection

Jayanta Mukherjee 55
Minimum Shift Keying
b

b1

Variable phase shift also added b2 ωc

xBB

Tb
MSK is similar to Offset PSK but with square pulses replaced by sine function

The sine function has frequency ω1 = (2π/4Tb )

56
Jayanta Mukherjee
The smooth phase transition lowers the signal power in the side lobes of the spectrum
 Flatter Gaussian
with decreasing BT

Rappaport - Wireless Communications, Principles and Practice, 2002 edition

Jayanta Mukherjee 57
MSK Representation in terms of FSK
N-1 N-1
S MSK (t)   mI (t) p(t- 2nTb ) cos 2πf c t   mQ(t) p(t- 2nTb-Tb ) sin 2πf c t ,
n 0 n 0

where,
  πt 
sin   0  t  2Tb
p(t)    2Tb 
 0 elsewhere

 The phase is continuous.
 With rectangula r mI (t) and mQ (t) pulses we obtain MSK
 With Gaussian mI (t) and mQ (t) pulses we obtain GMSK.
 MSK and GMSK can be viewed as frequency modulation, hence a constant envelop.

Jayanta Mukherjee 58
Generation of GMSK

p(t) xBB

A Gaussian filter is used

 Impulse response h(t)  K exp (- α t 2 ) . A type of Low Pass filter.
 Impulse response is usually truncated to [-Tb Tb ]
 99% bandwidth is less than 1.2 /Tb 

Jayanta Mukherjee 59
GMSK

Jayanta Mukherjee 60
Raised Cosine Filtering of Baseband for
rectangular pulse

Jayanta Mukherjee 61
Spectrum Regrowth in QPSK when using Raised
Cosine Filtering

Jayanta Mukherjee 62
Power Efficiency
x (t )  Ac cosc t   (t )   narrowband signal centered around c

No additional mixing and hence no spreading.

However gain compression will be present

There will be still some

spreading due to IM distortion

x (t )  x I t  cosc t   xQ t  sin c t 

Mixing of BW of xI and xQ with ωc  additional BW or spreading in addition to gain compression & IM distortion
63
Jayanta Mukherjee
Spectral Regrowth Advantages of GMSK

Jayanta Mukherjee 64