Wireless Communications

II. Advanced Modulations

 Architecture of a Wireless Communications System

From other sources

Information
sources
Source bits Channel bits

Channel Spread Multiple

Formatting
Source Encryption coding Multiplexer Modulator Spectrum access
T
coding X
s i (t)

Digital input
m i

Channel
Digital signal
Bit sequence Synchronism
Digital output
<

mi

^s (t)
i
Source Channel
Demultiplexer Demodulator Dispreading Multiple R
Formatting decoding Decryption decoding access X

Destination Source bits

To other destinations
Optional

Essential

2
Transmission: Modulation

Modulation

Information Numbers Physical

Radio
(usually bits)
(complex, no signal Channel
restriction) (bandwidth)

Symbol Value
“00” 1+j
“01” 1-j Fingerprint of the
modulation
“10” -1+j
“11” -1-j
3
 BASEBAND SIGNAL GENERATION (real part)

dk-2 dk-1 dk dk+1

∞ ∞

p(t) s (t ) = p (t ) ∗ ∑ d k δ (t − kTs ) = ∑d k p (t − kTs )

k = −∞ k = −∞

Ts t

∞
d (t ) = ∑ d δ (t − kT )
k = −∞
k s
Shaping pulse/filter
Ex: p(t) p(t)

Ts/2
Ts
Rectangular Raised Cosine
BASEBAND SIGNAL GENERATION
Rectangular

Raised Cosine
FREQUENCY DOMAIN (LIMITED BANDWIDTH)

1
Gs ( f ) = Gd ( f ) P( f )
2

Ts
Ex: p(t)

Ts/2

Rectangular

• Time limited pulses generate unlimited bandwidth signals.

• Nyquist pulses (as raised cosine) solve the trade off
Raised cosine

Raised cosine (time domain) Raised cosine (frequency domain)

hN(t)
1

HN( f ) / T

β=0
β=1 β = 0,2
β = 0,7 β = 0,5
-3 -2 -1 0 1 2 3 t/T

β=0 β = 0 ,7
β = 0,2 β=1
β = 0 ,5

fT
 BASEBAND DIGITAL TRANSMISSION SYSTEM
n(t) Sampling at t=t0+nT

d(t) s(t) r(t) z(t) {dˆk }

p(t) hc(t) + hr(t) Decision

CHANNEL

∞ ∞ ∞
d (t ) = ∑ d δ (t − kT )
k = −∞
k s s (t ) = p (t ) ∗ ∑ d δ (t − kT ) = ∑ d
k = −∞
k s
k = −∞
k p (t − kTs )

∞
r (t ) = hc (t ) ∗ ∑d
k = −∞
k p(t − kTs ) + n(t )

∞ ∞
z (t ) = hr (t ) ∗ hc (t ) ∗ p(t ) ∗ ∑ d δ (t − kT ) + h (t ) ∗ n(t ) = h(t ) ∗ ∑ d δ (t − kT ) + n
k = −∞
k s r
k = −∞
k s f (t )

h(t ) ≡ p (t ) ∗ hc (t ) ∗ hr (t )
n f (t ) ≡ n(t ) ∗ hr (t ) 8
Noise

𝑑𝑑(𝑡𝑡) s(t) y(t) y(t) + n(t) ̂

Source Channel + Receiver 𝑑𝑑(𝑡𝑡)

Hc(f) n(t)

•Binary Signal (Eb) 1  Eb  Errors in the

•AWGN (No/2) Pb = erfc 

2  N o  estimated symbols
•Matched Filter may appear

Matched Filter maximizes

S Eb
  =
 N T max
No
SER, BER, Pb
2

9
Ways to reduce the error probability
 Eb 
In BPSK: 1
Pe = erfc  
2  N0 
Eb=PR/rb
No ↓
Eb/No ↑ PR=PT/L

rb ↓
Pe ↓ Eb ↑ L↓
PR ↑
PT ↑

Channel coding

10
Error probability in M-PAM modulation
For the general case of a M-PAM modulation, the symbol error probability can be
determined as:
 1   3 ⋅log 2 M Eb 
Ps = 1 − ⋅ erfc ⋅ 
 M (
 M −1 No 
2
) 
Ps (probabilidad de error en el símbolo)

11
Eb / No (dB)
BAND-PASS SIGNALS

• What is a band-pass signal?

– Signal is frequency shifted
– IQ decomposition: TWO signals in ONE carrier
𝑠𝑠 𝑡𝑡 = 𝑠𝑠𝐼𝐼 𝑡𝑡 cos 2𝜋𝜋𝑓𝑓𝑐𝑐 + 𝜃𝜃 − 𝑠𝑠𝑄𝑄 𝑡𝑡 sin 2𝜋𝜋𝑓𝑓𝑐𝑐 + 𝜃𝜃

B B

f f carrier f (Hz)
base

12
 BASEBAND DIGITAL TRANSMISSION SYSTEM
n(t) Sampling at t=t0+nT

d(t) s(t) r(t) z(t) {dˆk }

p(t) hc(t) + hr(t) Decision

Complex
signals CHANNEL

∞ ∞ ∞
d (t ) = ∑ d δ (t − kT )
k = −∞
k s s (t ) = p (t ) ∗ ∑ d δ (t − kT ) = ∑ d
k = −∞
k s
k = −∞
k p (t − kTs )

∞
r (t ) = hc (t ) ∗ ∑d
k = −∞
k p(t − kTs ) + n(t )

∞ ∞
z (t ) = hr (t ) ∗ hc (t ) ∗ p(t ) ∗ ∑ d δ (t − kT ) + h (t ) ∗ n(t ) = h(t ) ∗ ∑ d δ (t − kT ) + n
k = −∞
k s r
k = −∞
k s f (t )

h(t ) ≡ p (t ) ∗ hc (t ) ∗ hr (t )
n f (t ) ≡ n(t ) ∗ hr (t ) 13
Example of Band-pass modulator
Delay
I(t)
Pulse shaping
Filter

{Bits} A·cos (2πf0t)

v(t)
S/P π/2
A·sin (2πf0t)
Pulse shaping Q(t)
Filter

14
Constellation
– Representation of the Symbol Value
complex values of the “00” 1+j
symbols in a 2D plot “01” 1-j
• In-phase component: x axis “10” -1+j
• Quadrature component : y axis “11” -1-j

– Theoretical constellation of Q

the modulation only

present at the transmitter
I

15
NOISE: Symbol Error Rate
• Noise also can be decomposed in
IQ components
– It “moves” received symbols from
theoretical constellation
– Decision regions
• Receiver estimates the received symbol
as the closest to the theoretical one
– Higher distance between symbols ⇒
Lower error probability
– Higher SNR ⇒ Lower error
probability
– Higher constellation order ⇒ Lower
distance between symbols ⇒ Higher
spectrum efficiency (bits/s/Hz) ⇒
Higher error probability 16
BAND-PASS DIGITAL MODULATIONS
Digital Modulations
Frequency Keying:
• Each symbol corresponds to a real value related
to the frequency of the transmitted signal.
• Symbol detection in the frequency domain
• M-FSK, MSK

Amplitude and Phase Keying:

• Each symbol corresponds to a complex value
that represents the in-phase and quadrature
components of the IQ decomposition of the
transmitted signal.
• Symbol detection in the time domain,
separated for each complex component.
• M-ASK, BPSK, QPSK, O-QPSK, M-PSK, M-QAM. 17
FSK

• Frequency Shift Keying

– Information in the variation of the Frequency
– Example: Binary FSK
− V → "0" 2∆f
m(t ) = 
 V → "1"

f i (t ) = f c ± k f V = f c ± ∆f f0 f1 freq

– M-FSK
2∆f

...

f00 f01 f11 freq

18
ASK

• Amplitude Shift Keying

– Information in the variation of the Amplitude
s (t ) = Cm(t ) cos(2πf c t )
– Examples
Component en
quadratura

OOK 4-ASK 4-ASK

Component
en fase cos(ωct)
C cos(ωct) C 2C 3C

Binary ASK OOK (On-Off Keying) M-ASK (M =4)

19
PSK

• Phase Shift Keying

– Information in the variation of the Phase
θ 
– Example: BPSK s(t ) = C cos[2πf t +  ] 0

θ 
c
1

- Signal for “1”: C cos[2πf ct ] BPSK

θ 0 = 0º θ1 = 180º
- Signal for “0”: − C cos[2πf ct ]
C cos(ωct)

– Example: QPSK (M=4)

QPSK QPSK
C

C cos(ωct) cos( ωc t)

20
M-PSK

21
QAM

• Quadrature Amplitude Modulation

– Information in the variation of the Amplitude and
Phase
• Combination of ASK and PSK
– Example: 16-QAM
• M=16 levels coded with 4 bits/symbol

16-QAM 16-QAM

22
M-QAM

Constellation
Q

23
Signal Amplification
• Radiofrequency Amplifiers  linear zone followed by saturation.
• Operating in the non-linear zone involves deformation of the shape of
wave ⇒ Pb increase, and occurrence of intermodulation products ⇒ ACI
(Adjacent Channel Interference)
• Amplitude modulations ⇒ require lower working point (higher Back-off) to
avoid saturation and distortion. Higher Back-off ⇒ Lower power efficiency
• Phase modulations are more interesting although linearization techniques of
the amplifiers should also be applied as the signal may not have constant
envelope and still suffer distortion.
Pout

Lineal zone
Pin 24
PAPR
• Peak-to-Average Power Ratio
𝑃𝑃𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝
PAPR=𝟏𝟏𝟏𝟏 log (𝒅𝒅𝒅𝒅)
𝑃𝑃𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
Modulus
 The higher the PAPR, the higher the Q Error
required back-off, the lower the
efficiency, the higher the EVM
 Best value: PAPR=0 dB (1 in linear scale)
• The higher the PAPR, the higher the EVM
(Error Vector Magnitude, Distortion)
• Possibilities to reduce EVM:
− Amplifier Linearization Techniques I
 Higher consumption
− Reduce PAPR
− Modulation
− Shaping Pulse 25
Efficiency
• Lower working point forces amplifiers to work with low
efficiency. PAPR fixes working point
– Higher PAPR  Higher Back-Off  Lower efficiency
• Example: Average TX power 24 dBm (250 mW) and PAPR 10 dB
 amplifier with saturation power 34 dBm (2.5W)  only 10%
efficiency  Higher consumption, higher cost
Pout

5 dB Backoff

10 dB Backoff

Pin 26
Vector Diagram

QPSK
Rectangular shaping pulse Q

PAPR>1
Offset-QPSK
7.2: OQPSK (Offset QPSK)

•OQPSK: An offset of Tb between the in-phase and quadrature

branches is introduced so they can not change simultaneously

•Delay Tb signal Q(t)

v (t ) = I (t ) cos( w0t ) + Q (t ) sin( w0t )

I (t ) = A∑ d 2 k • rectT (t − 2kTb );Q (t ) = A∑ d 2 k +1 • rectT (t − ( 2k + 1)Tb )

k k

28
QPSK / OQPSK

Ts=0.2
QPSK Tb=0.1 OQPSK

29
Offset-QPSK

Vector diagram

180° jumps removed I

PAPR improved

30
 7.3: MSK (Minimum Shift Keying)
MSK: Minimum Shift Keying

• O-QPSK with sinusoidal pulse shaping

v (t ) = I (t ) cos( w0 t ) + Q(t ) sen( w0 t )
π t
I (t ) = A ∑
k
d 2 k ⋅ rect T (t − 2kTb ) ⋅ cos   ;
 T
π t
Q( t ) = A ∑
k
d 2 k +1 ⋅ rect T ( t − ( 2k + 1) Tb ) ⋅ sen  
 T

p(t)

Pulse shaping
sin(x)
-T/2 t
T/2

31
MSK: Minimum Shift Keying

Vector diagram
Circle Constant envelop
Allows the use of non
PAPR=1 linear power amplifiers

Spectrum
10
rejection to secondary lobes
0 higher than with QPSK
-10 23dB
-20 GENERALLY NOT
ENOUGH!!!
-30

-40 1.5/Tb
-50
-60
f0-0.75Tb f0 f0+0.75Tb f
32
MSK: Minimum Shift Keying
• MSK is also a particular case of FSK with frequency deviation: fd=1/(4Tb)

 1
 f + si d k ="1"

0
4Tb
f i (t ) = 
1
 f0 − si d k ="0"

 4Tb

Bit Sequence “1” “0” “0” “1” “1” “1” “0” “0” “1” “0”
f0+(1/4Tb)
f0-(1/4Tb)
1

0
V(t)

-1

Normalized Time (t/T)

dk FREQUENCY v(t)
MODULATOR LEVEL
ADAPTER MODULATOR
33
 7.4: GMSK (Gaussian MSK)
GMSK: Gaussian Minimum Shift Keying
MSK with symbol pre-distortion to improve spectral properties.
freq. MSK freq.
GMSK

f0+fd f0+fd

t
t
f0-fd f0-fd

BÁSIC MODULATOR

dk g(t) FREQUENCY s(t)

GAUSSIAN
MODULATOR
FILTER

34
GMSK
SPECTRUM

BT=0.5
MSK
BT=1.0

BT=0.35

35
PSK EDGE
(Enhanced Data Rates for GSM Evolution)

8-PSK Rotated Modulation

Reduction of PAPR with respect to a classic 8-PSK

36
Differential Coding/Modulations
Although distance
Transmitted symbols Systematic error
between points has not
Received symbols been altered detection

(“0”,”1”) (“1”,”1”) (“0”,”1”) (“1”,”1”)

(“1”,”0”) (“0’”,”0”) (“1”,”0”)

(“0’”,”0”)

Common in Unknown channel

Wireless Channels phase shift
37
Differential Coding/Modulations

Conventional Coding Differential Coding

Symbol Value Symbol ∆ PHASE
00 1+j 00 90º
01 1-j 01 180º
10 -1+j 10 270º
11 -1-j 11 0º

Information in the
DIFFERENCE of phase
between consecutive
symbols 38
Differential Coding/Modulations

Transmitted symbols
(N+1)-th symbol
Received symbols

Phase DIFFERENCE between

consecutive symbols not N-th symbol
changed for ANY channel shift

Position of one received symbol

means NOTHING
39
Encoder/Decoder

dk=complex number with the phase

equal to the allocated value

dk=ak+jbk ck=dkck-1
ENCODER
z-1

ck dk = ck ck* −1

DECODER
( )* z-1
BIT ERROR RATE (BER)

-2
10

•COHERENT DEMODULATOR -4
10

PbM − DPSK ≈ 2 PbM − PSK

BER
(high SNR) -6
10

BPSK
-8 BPSK (diferencial)
10
QPSK
QPSK (diferencial)
-10
10
0 2 4 6 8 10 12
Eb/No (dB)

41
π/4-DQPSK

a 2k-1 a 2k ∆φk

0 0 + π/ 4

0 1 + 3π/ 4

1 0 - π/ 4

1 1 - 3π/ 4

Constellation

• Bit error rate same as DQPSK for coherent demodulation

• Simpler symbol synchronism mechanisms

42