ELEC3505 - Communication

Edmund Tse 2008

Communications Formula Sheet

Introduction

Shannons Information Capacity Theorem

o Maximum rate at which information can be transmitted across the channel without error
o

C B log 2 1 SNR

C
B
SNR

Information capacity
Channel bandwidth
Received SNR

Continuous Wave Modulation

Amplitude modulation

Carrier wave

Baseband signal

Modulated wave

ct Ac cos2f c t

Ka

mt

Amplitude sensitivity

Fourier transform

S f

1
1
Ac f f c f f c K a Ac M f f c M f f c
2
2

Linear modulation schemes

o

st s I t cos2f c t sQ t sin 2f c t

in phasecomponent

Modulation Scheme
DSB-SC
SSB (upper)
SSB (lower)
VSB of lower
VSB of upper

st Ac 1 K a mt cos2f c t

quadraturecomponent

sI(t)
m(t)

sQ(t)
0

1
mt
2
1
mt
2
1
mt
2
1
mt
2

1
m t - Hilbert transform
2
1
m t
2
1
mt - Output of HQ(f)
2
1
mt
2

DSB-SC

Modulated wave

st AC mt cos2f c t

Fourier transform

S f

1
Ac M f f c M f f c
2

Page 1 of 10

ELEC3505 - Communication

Edmund Tse 2008

Coherent detection (synchronous demodulation)

vo t

After filtering

1
Ac Ac mt cos
2

-- proportional to signal m(t)

Conditions:
Must have constant phase error
Maximum amplitude when = 0;
Quadrature carrier multiplexing:

SSB

Quadrature null effect when = /2

st

1
1
Ac mt cos2f c t Ac m t sin 2f c t
2
2

o
o
o

+ for upper sideband, - for lower sideband

Uses product modulator then bandpass filter
Must have an energy gap

VSB

o
o

Suppresses one sideband

Instead of using bandpass filter, uses filter H(f)
Odd symmetry around fc
Linear phase response
Angle modulation
o

Modulated wave

i(t)
Ac

st Ac cos i t

angle of a modulated carrier

carrier amplitude

Instantaneous freq

f i t lim f t t
t 0

Phase modulation
o

i t 2f c t k p mt

1 d i t
2 dt

st Ac cos 2f c t k p mt

2fc(t) angle of unmodulated signal

kp
phase sensitivity
Page 2 of 10

ELEC3505 - Communication

Edmund Tse 2008

Frequency modulation
o

f i t f c k f mt

fc
kf

t
st Ac cos 2f c t 2k f m d

frequency of unmodulated carrier

frequency sensitivity

Instantaneous frequency for sinusoidal modulating signal mt Am cos2f mt :

f i t f c k f mt f c k f Am cos2f mt f c f cos2f mt

f = kfAm
Instantaneous phase

maximum frequency deviation; proportional to amplitude of m(t)

i t 2 f i t d 2f c t
t

FM signal

Carsons Rule
o

<< 1 radian

st Ac cos2f c t Ac sin2f c t sin2f mt

narrowband FM

< 0.3 negligible residual AM and harmonic PM

>> 1 radian wideband FM

1
BT 2f 2 f m 2f 1

o Approximates bandwidth of FM with single tone modulating signal at fm

Noise AWGN
o

Power spectral density

Average noise power per unit bandwidth

Narrowband noise

Average power of noise

N0
2
N0

nt nI t cos2f c t nQ t sin2f c t

N 0 BT

SNR

average power of modulated st

average power of filtered n t
average power of demodulated mt
o SNR O
average power of noise at receiver output
average power of modulated st
o SNR C
average power of channel noise in message signal at receiver input
SNRO
o Figure of merit
FOM
SNRC
o

modulation index; maximum departure of i(t) from 2fmt

st Ac cos2f c t sin2f mt

f
sin 2f m t 2f c t sin 2f m t
fm

SNR I

Coherent detection
o Amplitude modulation with suppressed carrier

P S M f df
W

Average power of message signal

Average power of DSB-SC modulated signal

1 2 2
C Ac P
2

Page 3 of 10

ELEC3505 - Communication

Edmund Tse 2008

2

2
c

C A P
2WN 0

SNRC for DSB-SC

After coherent detector, quadrature component is rejected

Average power of message component in output

1 2 2
C Ac P
4

Envelope detection
o

Average power of carrier component in signal

Average power of message in signal

SNRC for AM

Ac2 1 k a2 P
2WN 0

1 2
Ac
2
1 2 2
Ac k a P
2

Noise in FM
o

3 Ac2 k 2f P

SNRO for FM

SNRC for FM

FOM for FM

2 N 0W 3

Ac2
2WN 0
3k 2f P
W2

Pulse Modulation

Definitions
o g(t)
an analog signal with finite energy
o Ts
sampling period (s)
o fs
sampling rate
o g(nTs)
instantaneous sample of g(t)
o g(t)
ideal sampled signal
Pulse amplitude modulation (PAM)
o

st

mnT ht nT

Quantisation noise
o Q = M V
Q
M
V

0t T
1

1
ht
t 0, t T , m t mnTs t nTs
2
n
0 otherwise

quantisation noise
continuous sample
quantised value

Step size of quantiser

Probability density function of Q

2mmax
L
1

f Q q

Page 4 of 10

q
2
2
otherwise

ELEC3505 - Communication

Edmund Tse 2008

Q2 eQ 2 2 q 2 f Q q dq

o
o

Variance of Q
SNRO of quantising

2
Q

2R
2

number of bits per sample

Average probability of symbol error in binary PCM depends on

R = log2L

3P
2

mmax
P

PCM
o

1 2 2

q dq

2
12

Eb
N0

Eb
N0

transmitted signal energy per bit

noise spectral density

DM

en mn mq n 1

Error signal

Quantised e[n]

Output of DM

Slope overload distortion occurs when slope overload:

eq n sgn en

mq n mq n 1 eq n

dmt
max
Ts
dt

DPCM

n
en mn m

Prediction error

Quantiser output

q[n]

eq n en qn

quantisation error

Quantised samples

n eq n m
n en qn
mq n m

SNRO for DPCM

M2 M2

Q2 E2

M 2
Q 2
E2
GP

E2
2 GP SNR Q

variance of input signal m[n]

variance of quantisation error q[n]
variance of prediction error
processing gain of the differential quantisation; gain in SNR due to quantisation if > 1

Baseband Pulse Transmission

Matched filter

H opt f kG f e j 2fT

for real signals, H opt f kGT t

Page 5 of 10

ELEC3505 - Communication

Edmund Tse 2008

|go(T)|2
E[n2(t)]

g o t H f G f e

Noise psd

g o T

df

H f G f e

j 2ft

df

N0
2
H f
2

N
2
E n 2 t S N f df 0 H f df

Schwarzs inequality if 2 complex functions 1(x) and 2(x) satisfy 2 conditions:

then

E n 2 t

SN f

1 x dx and
2

j 2ft

g o T

instantaneous power of output pulse signal

expected average output noise power

Peak pulse SNR

2 x dx
2

1 x 2 x 1 x dx 2 x dx
2

Use this to find the optimal filter

Output

Fourier transform

Go f H opt f G f kG f G f e j 2fT k G f e j 2fT

Inverse fourier

g o t Go t e j 2fT df k G f df

Rayleighs energy theorem

g o t kE

E n 2 t

max

kN0
2

kE

G f df

k N0 E

2
2

E g 2 t dt G f df
2

kN0 E
2

E
N0

Error rate due to noise

o

Complementary error function

Polar NRZ

erfcu

A
2
1
e z dz erfc
N 0 Tb
N T
2
0
b

e z dz
2

P10

P01

Average symbol error probability Pe P0 P10 P1 P01

Optimum threshold

A
2
1

e z dz erfc
N 0 Tb
N T
2
0
b

opt

P
N0
log 0
4 ATb
P1

Page 6 of 10

A
P0
erfc
N T
2
0
b

P1

erfc A
2
N T
0
b

ELEC3505 - Communication

Edmund Tse 2008

Eb = A2Tb
Inter-symbol interference
o

Pe

For binary symmetric channel (P0 = P1 = 0.5)

A
1
erfc
N T
2
0
b

erfc Eb

2
N0

transmitted signal energy per bit

After transmit filter

st ak g t kTb
k

Receive filter output

yt ak pt kTb nt
k

n(t)

scaling factor to account for amplitude change during transmission

noise component

Condition on P for zero ISI

P f nR T
b

Ideal Nyquist channel has

P f 2W
0

Raised cosine spectrum

1
P f
4W

Roll-off factor

Transmission bandwidth

W f W 1
f
rect

2
W
2
W

f W

1
2W

f W
1 sin

2W 2 f1
0

0 f f1
f1 f 2W f1
f 2W f1

f1
W
BT 2W f1 W 1

Signal Space Analysis

Geometric representation of signals

si t sij j t

0t T
generated by synthesizer

i 1,2,..., M

sij si t j t dt

i 1,2,..., M
calculated by analyser

j 1,2,..., N

Set of signals

j 1

Coefficients of expansion

Real-valued basis functions are orthonormal (Product of any 2 bases = Kronecker delta ij)

length of a signal vector s

Inner product of si

si

siT si sij2
j 1

Energy of signal si(t) of T s

T
T N
N

Ei si2 t dt sij j t sik k t dt si

0
0

j 1
k 1

Inner product of 2 signals

s t s t dt s

Schwarz inequality

s t 2 dt s t 2 dt

s
t
s
t
dt

1
2

1
2

Page 7 of 10

T
i k

ELEC3505 - Communication

Edmund Tse 2008

Likelihood functions

Lmi f x x | mi

Characterise AWGN

Log-likelihood function

l mi log Lmi

For AWGN channel

l mi

i 1,2,..., M

1 N
2
xi sij
N 0 j 1

Likelihood function

i 1,2,..., M

Probability of error
o

Pe pi Pe mi

Bound for average probability of error

i 1

Pe mi

Finite union of events over-bounded by sum of constituents

Pe

pi
dik

P A
ik

k 1i

d
1 M M
pi erfc ik

2 N
2 i 1 k 1i
0

i 1,2,..., M

probability of transmitting symbol mi

Euclidian distance between si and sk, = ||si sk||

d
1 M
erfc ik

2 N
2 k 1i
0

for circular symmetric signal constellation since all Pe(m) is the same

Passband Data Transmission

Passband modulation
o

Modulated passband signal

SB f

s(t) = sI(t) + jsQ(t)

sI(t), sQ(t), s(t)

st sI t cos2f c t sQ t sin2f cT Re st e j 2fct

complex envelope of s(t) passband signal

low-pass signals uniquely defined by s(t) and fc
Passband power spectral density (of s(t))

1
S B f f c S B f f c Power spectral density of s(t) frequency shifted SB(f)
4
R
B bits/s/Hz
o Bandwidth efficiency
B
o

SS f

Coherent BPSK

Representation of symbol 1

s1 t

2 Eb
cos2f c t
Tb

Representation of symbol 0

s2 t

2 Eb
2 Eb
cos2f c t
cos2f c t
Tb
Tb

Eb

fc

energy transmitted per bit

nc
Tb

Function of unit energy

nc

integer to make integer number of cycles of carrier wave

1 t

2
cos2f c t 0 t Tb
Tb

Page 8 of 10

ELEC3505 - Communication

Probability of error

Eb
1

pe p01 p10 erfc

2
N
0

Power spectral density

2 Eb sin 2 Tb f
SB f
2 Eb sinc 2 Tb f
2
Tb f

QPSK

cos 2f c 2i 1 0 t T
si t T
4

0
otherwise

si t

Trigonometric identity to break down into two orthonormal basis functions

E
Pc 1 erfc
2N0

erfc 2 E
4
2N
0

signal energy per symbol (has 2 bits)

E
Pe erfc
2N0

Average probability of error

In-phase (and ve of quadrature component)

i 1,2,3,4

2E

2E

cos 2i 1 cos2f c t
sin 2i 1 sin 2f c t
T
4
T
4

Probability of correct

Edmund Tse 2008

for large

g t T
0

E
1
N0

0t T
otherwise

S B f 2E sinc 2 Tf 4Eb sinc 2 2Tb f

Sum of in-phase and quadrature psds

Offset QPSK

1 t
o

2
cos 2f c t 0 t T
T
2
T
3T
2 t
sin 2f c t
t
T
2
2

Delayed quadrature component

/4 shifted QPSK

M-ary PSK

2E
2

i 1
cos 2f c t
T
M

si t

Probability of error

Pe erfc
sin
M
N0
Page 9 of 10

-- considers only deviation onto neighbours

ELEC3505 - Communication

Edmund Tse 2008

Power spectra

S B f 2E sinc Tf 2Eb log 2 M sinc Tb f log 2 M

Bandwidth

2 Rb
2

T log 2 M

Bandwidth efficiency

Rb 1
log 2 M
B 2

M-ary QAM

1 t
o

2
cos 2f c t
T
2
2 t
sin 2f c t
T

Transmitted M-ary QAM signal sk t

Square constellation

0t T

= 2D M-ary PAM

2 E0
2 E0
ak cos2f c t
bk sin 2f c t
T
T

0t T

E0
1

Pe 21
erfc

N
M

Probability of symbol error

Coherent FSK
o

2 Eb

cos 2f i t 0 t Tb
si t Tb

0
otherwise

Signal space for binary

E
s1 b
0

Probability of error

Eb
1

pe p10 p01 erfc

2
N
0

Baseband psd

sB f

i 1,2

Eb
2Tb

0
s2

Eb

1
1 8Eb cos 2 Tb f

2Tb
2Tb 2 4Tb2 f 2 1 2

Continuous phase FSK

2 Eb
cos2f1t 0 For symbol1

Tb
st
2 Eb cos2f t 0 For symbol 0
2
Tb

Express as angle modulated signal

ht

Tb

t 0

st

(0) = phase at t = 0

2 Eb
cos2f c t t
Tb

+ for sending 1

0 t Tb

- for sending 0

h
deviation ratio difference between f1 and f2, normalised with bit rate 1/Tb
Minimum shift keying
o

st

2 Eb
2 Eb
cos t cos2f c t
sin t sin 2f c t
Tb
Tb

Page 10 of 10

Eb
1

Pe erfc

2
N
0