Communication PW

Communication
System
Published By:
Physics Wallah
ISBN: 978-93-94342-39-2
Mobile App: Physics Wallah (Available on Play Store)
Website: www.pw.live
Email: support@pw.live
Rights
All rights will be reserved by Publisher. No part of this book may be used or reproduced in any manner
whatsoever without the written permission from author or publisher.
In the interest of student's community:
Circulation of soft copy of Book(s) in PDF or other equivalent format(s) through any social media channels,
emails, etc. or any other channels through mobiles, laptops or desktop is a criminal offence. Anybody
circulating, downloading, storing, soft copy of the book on his device(s) is in breach of Copyright Act. Further
Photocopying of this book or any of its material is also illegal. Do not download or forward in case you come
across any such soft copy material.
Disclaimer
A team of PW experts and faculties with an understanding of the subject has worked hard for the books.
While the author and publisher have used their best efforts in preparing these books. The content has been
checked for accuracy. As the book is intended for educational purposes, the author shall not be responsible for
any errors contained in the book.
The publication is designed to provide accurate and authoritative information with regard to the subject matter
covered.
This book and the individual contribution contained in it are protected under copyright by the publisher.
(This Module shall only be Used for Educational Purpose.)
Design Against Static Load
COMMUNICATION SYSTEM
INDEX
1. Amplitude Modulation ........................................................................................................ 8.1 – 8.10
2. Angle Modulation ................................................................................................................ 8.11 – 8.18
3. Random Variable and Random Process ............................................................................... 8.19 – 8.45
4. Digital Communication ........................................................................................................ 8.46 – 8.55
5. Digital Receiver .................................................................................................................... 8.56 – 8.70
6. Information Theory ............................................................................................................. 8.71 – 8.81
7. Miscellaneous ..................................................................................................................... 8.82 – 8.86
GATE-O-PEDIA ELECTRONICS & COMMUNICATION ENGINEERING

Design Against Static Load
1 AMPLITUDE
MODULATION
1.1. Introduction
Band limiting : Bandun limited to bandlimited (LPF)
Base band signal : Message signals, low cut off fre = 0 Hz or very close to 0 Hz.
Bandpass signal : By shifting baseband signal to very high freq.
fH
• Wideband signal :  1 (Base band signal)
fL
fH
• Narrowband signal : 1 (Bandpass signal)
fL
Modulated Signal:
C (t ) = Ac cos(ct + ) = Ac cos ct
Carrier signal
(carrier before modulation)
S (t ) = A(t )cos[ ct + (t )] Modulated signal
Instantaneous phase
Instantaneous Instantaneous
amplitude frequency
Amplitude Modulation:
DSB-FC (Double side band full carrier)
C (t ) = Ac cos ct carrier before modulation
S AM (t ) = Ac cos ct + m (t )cos ct  S AM (t ) = [ Ac + m (t )]cos ct carrier after modulation
[m (t )]
Modulation Index  = max
Ac
(1)  < 1 (under modulation)
Am
= 1
Ac
GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.1

Communication Systems
[ E (t )]max − [ E (t )]min
=
[ E (t )]max + [ E (t )]min
  1, A(t )  0, E(t ) = A(t )

Recovery through E, D possible.
S (t )max = E (t ) |max = Ac (1 + )
S (t )min = E (t ) |min = Ac (1 − )
(2) Critical Modulation:-  = 1, A(t )  0 , E (t ) = A (t ) , m(t) can be recovered with envelope detector .
(3) Over modulation:   1, A (t )  0, E(t ) =| A(t ) |, not possible by E.D
Frequency Related Parameters
(1)
(2) C (t ) → f max = f m
(3)
➢ PAM = PC + PSB
Pm
PUSB = PLSB =
4
Modulation efficiency
PSB P /2
= = m
PAM P + Pm
c
2
Share of sideband power in total power
➢ ka [Amplitude sensitivity of amplitude modulator]
1
ka = (per volt),
Ac
A (t ) = Ac [1 + ka m (t )]
A (t) > 0, E. D. Applicable
For single tone sinusoidal signal

Am2
fmax = fm , BW = 0Hz, Pm = → for message signal
2
Am
=
Ac
S AM (t ) = [ Ac + Am cos 2f mt ]cos 2fct
= Ac [1 +  cos 2f mt ]cos 2fct
Ac A
S AM (t ) = Ac cos2fct + cos[2( fc + fm ) t ] + c cos[2( fc − f m )t ]
 2  2 
carrier USB LSB
Pm Ac2  2  Pc2
PAM = Pc + = 1 + =
 cP + → PSB
2 2  2   2
Pc
Pc2
2 2
=  %  =  100%
Pc2 2 + 2
Pc +
2
➢ If ka given →  = ka | m (t ) |max
| m (t ) |max
If ka not given →  =
Ac
Important Points:
=0 =1 % Change
PAM = Pc PAM = 1.5Pc 50 %
1
=0  = = 33.33 % 0 % to 33.33 %
3

(2)  →  
(3) PAM → Will be constant if Pc  and  
If m (t ) is multiple single tone signal-
 Am Am 
S (t ) = Ac 1 + 1 cos 2f m1t + 2 cos 2f m2 t  cos 2fct
 Ac Ac 
Am1 Am2
1 = , 2 = − 1  2
Ac Ac
m(t) → fm1 , fm2 → fmax = fm2
fm2  fm1 BW = fm2 − fm1

S (t ) = f c BW = 2 × Max. Freq. component of m (t )
Power Related Parameters

Am21 Am2 2
Pm = +
2 2
Pm Ac2  12 22 

PAM = Pc + = 1 + + 
2 2  2 2 
Pc12 P 2
PUSB1 = PLSB1 = , PUSB2 = PLSB2 = c 2
2 2
Pc12 P 2
PUSB1 = PLSB1 = , PUSB2 = PLSB2 = c 2
4 4
PSB 2
= = T2
PAM 2 + T
T = 12 + 22 + 32 + − − − −

Important Points:
m(t) (volt) PAM Prod
 2  Pc  2 
(1) Sinusoidal Pc 1 +  1 + 

 2  R  2 
Pc
(2) Square wave Pc (1 + 2 ) (1 + 2 )
R
 2  Pc  2 
(3) Triangular wave Pc 1 +  1 + 
 3  R  3 

2 2
VAM = Vc 1 + , I AM = Ic 1 + for sinusoidal
2 2
DSB- FC [AM] Modulator

(1) Square law Modulator:
a1 Ac2 a1 Ac2
y(t ) = a0 m (t ) + a0 Ac cos 2f ct + a1m2 (t ) + + cos 4f ct + 2m (t ) Ac cos 2f ct
2 2
(1) (2) (3) (4) (5) (6)
➢ Only (2) and (6) are desirable
 2a 
Z (t ) = a0 Ac cos 2fct 1 + 1 m (t )  DSB –FC
 a0 
Z (t ) = Ac' [1 + ka m (t )]cos2fct only when fc  3 f m fc  (2 + 1) f m
2a1
Ac' = a0 Ac , ka = ,  = ka | m (t ) |max
a0
(2) Switching Modulator:

Ac  4 
Z (t ) = 1 + m (t )cos 2fct  DSB − FC
2  Ac 
Z (t ) = Ac' [1 + ka m (t )]cos2fct
Ac 4
Ac' = , ka =  = ka | m (t ) |max
2 Ac
DSB- FC Demodulator-
(1) Square law demodulator-
a1 Am2 a1 Am2
Y (t ) = a1 Ac Am cos2fmt + + cos4fmt
4 4
Y (t ) = B0 + B1 cos 0t + B2 cos 20t
B2 
➢ 2nd harmonic distortion D2 = =
B1 4
( D2 )max % = 25%
➢ Practically not used
S 2
➢ I =
 min 
Envelope Detector:
A2 + B2 cos ct → E.D → A2 + B2
(1) x (t ) = A cos 0t + B sin 0t → E(t ) = A2 + B2
(2) x (t ) = A cos(0t + ) + B sin 0t → E(t ) = A2 + B2 − 2 AB sin 

(3) x (t ) = A (t )cos ct → E (t ) =| A(t ) |
(4) x (t ) = ( Ac + m (t ))cos ct → E (t ) =| Ac + m (t ) |
Important Points:
➢ Used only when   1
1
➢ Tc = RS C , (charging time constant)
fc
1
➢ Tc = RS C  → Peaks are not detected.
fc
1
➢ Diagonal clipping → RLC =
fm
1 1 − 2
➢ To avoid diagonal clipping RLC  , RLC 
fm m
1
➢ Ta = RLC  fluctuation is output
fc
1
➢ To remove fluctuation RLC 
fc
➢ Proper choice of discharging time constant RLC -

No fluctuation No diagonal clipping
(7) m(t): Multitone f m → f max = Max. freq. component of m(t)
1.2. Synchronous Detector
   Recovery

=0 0  (2n + 1) ✓
2

=0 0 = (2n + 1) Q.N.E
2
0 =0 =0 
=0 =0 =0 ✓
DSB-SC :
S DSB−SC (t ) = m (t ) Ac (cos 2fct ) E (t ) =| A (t ) |=| Ac m (t ) |
➢ . = 2  max. freq. component of m (t )

BW
PSB Pm Pc
➢ PDSB = Pm Pc = PSB → PUSB = PLSB = +
2 2
Pc2
PDSB = = PSB
2
➢ Single tone modulation.
Ac Am AA
S (t ) = cos[2( fc + f m ) t ] + c m cos[2( fc − f m )t ]
2 2
Ac2 Am2
PDSB = Pm Pc =
4
Ac2  Am1 Am2 

2 2
Multiline PDSB = Pc Pm =  + 
2  2 2 
 
 A2 
Square wave - PDSB = Pc Pm =  c  Am2
 2 
 
 A2  A2 
Triangular wave PDSB = Pc Pm =  c  m  a
 2  3 
  
 A2  A2 
Saw-toothed wave- PDSB = Pc Pm =  c  m 
 2  3 
  
(1) Balanced Modulator- S DSB (t ) = 2 Ac ka m (t )cos fct
(2) Ring Modulator- y (t )  m (t )cos ct

 = 0,   0, y (t ) = 0 QNE
Ac AC'
  0,  = 0, y (t ) = m (t )cos(t ) → distorted m (t )
2
Ac AC'
 = 0,  = 0, y (t ) = m (t ) → Attenuated
2
Hilbert Transformation.
1
h (t ) = ,
t
1
mh (t ) = m (t ) 
t
− j 0

H () = − j sgn () = 0 =0
j  0

➢ M h ( f ) = M ( f )[− j sgn ( f )]
H .T
➢ H.T [cos (t )] = sin (t ) ⎯⎯⎯ →− cos (t )
➢ Non causal LTI system.
H .T .
➢ x (t ) ⎯⎯ → xh (t )
H .T .
xh (t ) ⎯⎯ →−x(t )
➢ Magnitude spectrum of x(t) and xh (t ) will be same
➢ If x(t): Band limited then xh (t ) is also bandlimited.
➢ If x(t) is non periodic then xh (t ) is also non periodic
➢ x(t) and xh (t ) are orthogonal signal.
Drawback of DSB-SC
➢ 2 sideband Txed.
➢ If receiver is designed in such a way that it may recover the complete message signal from single SB then DSB-SC
S/S becomes impractical.
SSB- SC (Single sideband suppressed carrier)

(1) Point to point communication
Phase derserimination
(2) Two methods of generation
Frequency discrimination
(a) Phase Discrimination:

Ac m(t ) A m (t ) +  LSB
S (t )SSB = cos 2fct  c h sin 2fct
2 2 −  USB
Problem Solving
(1) Identify the phase descrimination setup
(2) Phase descrimination setup
(3) Phase descrimination setup:
+  SDSB (t ) → LSB
−  SDSB (t ) → USB
Spectral gap in D.S.B BPF Signal

0 Hz Ideal SSB-SC
0 Hz Practical VSB-SC
 0Hz Ideal SSB -SC
VSB − SC
depends on
 0Hz Practical
practical BPF
SSB − SC
➢ SSB- SC can be demodulated by Synchronous detection.
(1)  = 0,   0, m (t ) recovery not possible → freq. synchronization
(2)   0  = 0 , m (t ) recovery not possible → Phase synchronization
(3) Perfect syne,  = 0,  = 0 can be recovered

(4)  = 0,  = → NoQNE
2
Note:
(1) When video signal is transmitted through SSB- SC modular VSB- SC is generated.
(2) Synchronous detector can not recover m (t ) video signal from the above generated VSB- SC.

Percentage Power Saved

(1) % power saved in DSB- SC as compare to DSB-FC.
Psaved
% Psaved = 100%
PTotal
Pc 2
% Psaved = = = (1 − )
 2  2 + 2
Pc 1 + 
 2 
(2) % power saved in SSB- SC as compare to DSB-FC-
4 + 2
% Psaved =
4 + 22
(3) % power saved in SSB-SC as compared to DSB-SC.
% Psaved = 50%
Modulation B.W Power Application

(1) DSB-FC 2 f max PC + PSB Broadcatting
(2) DSB- SC 2 f max PSB 
(3) PSB Point to point voice
SSB-SC f max
2 communication
PSB Point to point video
(4) VSB-SC f max  f  2 f max  PVSB  PSB
2 communication.
Pre envelope and Complex Envelope

(1) Pre Envelope calculated for both baseband and bandpass signal.
Let x (t ) is real signal.
x+ (t ) = Pre envelope of x (t )
x+ (t ) = x (t ) + j xˆ (t )
xˆ (t ) = HT [ x (t )]
x+ ( f ) = x ( f )1 + sgn( f )
Complex Envelope: For bandpass only but result in low pass only
x (t ) → Bandpass signal.
Step-1. Calculate x+ (t ) = x (t ) + j xˆ (t )
xc (t ) = x+ (t )e− jct
Step 2. left shift of pre envelope by fc
X c ( f ) = X + ( f + fc )


2 MODULATION
ANGLE
2.1. Introduction
Signal = x(t) |x (t)|max

(1) A cos 0t + B cos 0t A+ B
(2) A sin 0t + B sin 0t A+ B
(3) A sin 0t + B cos 0t A2 + B2
(4) A cos 1t + B cos 2t A+ B
(5) A sin 1t + B sin 2t A+ B
(6) A cos 1t + B sin 2t A + B if A = B
 A + B if A  B
2.1.1. Instantaneous Angle and Instantaneous frequency-

S (t ) = Ac cos[i (t )]
i (t ) → Instantaneous angle (rad)
d i (t )
= i (t ) → instantaneous angular frequency.
dt
1 d i (t )  (t )
fi (t ) = or fi (t ) = i
2 dt 2
t
i (t ) =  i (t )
−
• Angle Modulation :
➢ Frequency Modulation
➢ Phase Modulation
• Frequency Modulation :
Sangle (t ) = Ac cos[ct + (t )]

d (t )
If angle Modulation is FM,  m(t )
dt
d (t )
= K f m(t ), K f = frequency sensitivity of frequency modulator
dt
i (t ) = c + K f m(t )  i (t ) = c + (t )

frequency
deviation
t
i (t ) = c +  K f m(t) dt (t ) = K f m(t )
−
d (t )
(t ) =
dt
Few Important Results
rad Hz
For Important Results For K f : Kf :
V - sec Volt
1. Instantaneous frequency i (t ) = ct + K f m(t ) fi (t ) = fc + K f m(t )
2. Instantaneous frequency deviation (t ) = K f m(t ) f (t ) = K f m(t ) Hz
3. Frequency deviation in +ve direction [(t)]max = K f [m(t)]max [f ]max = K f [m(t )]max
4. Frequency deviation in –ve direction [(t)]min = K f [m(t)]min [f (t)]min = K f [m(t)]min
5. Maximum value of instantaneous frequency [i (t )]max = c + [(t )]max [ fi (t )]max = fc + [f (t )]max
6. Minimum value of instantaneous frequency [i (t )]min = c + [(t )]min [ fi (t )]min = fc + [f (t )]min
7. Peak to peak frequency deviation [] p− p = [i (t)]max − [i (t)]min [f ] p− p = [ f (t )]max − [ f (t )]min
8. Maximum frequency deviation
9. Modulation index or deviation ratio of FM (t ) max = K f [m(t )]max f (t ) max = K f [m(t )]max
Maximum frequency deviation K f m(t ) max K f m(t ) max

BFM = BFM = BFM =
Maximum frequency component of m(t ) max f max
 rad  Hz
Important Phase Calculation Kf   Kf :
 V-sec  Volt
1. Instantaneous phase deviation in FM t t
(t ) = K f  m() d  (t ) = 2K f  m() d 
− −
2. Maximum phase deviation in FM t t
(t ) max = K f  m() d  2K f  m() d 
− − max

General expression for FM

rad
Kf :
V-sec
 t 
Sangle (t ) = Ac cos ct +  K f m() d 
 − 
Hz
For K f :
Volt
 t

SFM = Ac cos ct + 2K f  m() d 
 − 
For m(t ) = Am cos2fmt –
fmax = fm ,[m(t)]max = + Am , [m(t )]min = − Am , m(t ) max = Am
SFM (t ) = Ac cos c (t ) + BFM sin(2fmt )
For m(t ) = Am1 cos2fm1t + Am2 cos2fm2 t –
fmax = ( fm1 , fm2 )max
SFM (t ) = Ac cos c (t ) + B1 sin 2f m1t + B2 sin 2f m2 t 
K f Am1 K A
B1 = , B2 = f m2
fm1 fm2
Phase Modulation –
(t )  m(t )
(t ) = K p m(t )
 
rad Volt
K p : Phase sensitivity of phase modulator

rad
Kp =
Volt
Phase Calculation :
K p : rad/Volt
i (t ) = ct + K pm(t )
1. Instantaneous phase deviation = (t ) = K p m(t )
2. Maximum phase deviation = (t ) max = K p m(t ) max

Frequency Calculation
i (t ) = c + t
dm(t )
➢ (t ) = K p
dt
dm(t )
➢ (t ) max = K p
dt max
dm(t )
➢ (t ) min = K p
dt min
➢ i (t ) max = c + [(t )]max
➢ i (t ) min = c + [(t )]min
➢ p− p = [i (t )]max − [i (t )]min
dm(t )
➢ (t ) max = K p
dt max
dm(t )
Kp
(t ) max dt max
➢ FM = =
max max
SFM (t ) = Ac cos c (t ) + KPM (t )
When m(t ) = Am cos2fmt
m(t ) max = Am , fmax = fm , (t ) = −K p Amm sin mt
➢ [(t )]max = K p Amm
➢ {(t )}min = −K p Amm
➢ [i (t )]max = c + K p Amm , [i (t )]min = c − K p Amm
➢ () p− p = 2K p Amm
➢ (t ) max = K p Amm
➢  = K p Am = (t ) max
SPM (t ) = Ac cos c (t ) +PM cos 2fmt 
SPM (t ) = Ac cos c (t ) + 1 cos 2f m1 t + 2 cos 2f m2 t 

Types of FM –
➢ Narrow Band (  1)
➢ Wide Band
Ac A
SFM (t ) = Ac cos 2fc (t )+ cos[2 ( fc + f m ) t ] − c cos[2 ( f c − f m ) t ]
 2  2 
Carrier USB LSB
SFM (t ) = SNBFM (t )
➢ B.W = 2 fm
 2 
➢ PNBFM = PC 1 +    1, 2  1
 2
Ac2
PNBFM  PC =
2
Relation between DSB-FC and NBFM –

Ac A
S AM (t ) = Ac cos2fc (t ) + cos[2 ( fc + fm ) t ] + c cos[2 ( fc − fm ) t]
2 2
Ac A
SNBFM (t ) = Ac cos2fc (t ) + cos[2 ( fc + fm ) t ] − c cos[2 ( fc − fm ) t ]
2 2
1.
Frequency Component Strength AM Strength NBFM
fc Ac Ac
2 2
fc + f m Ac Ac
4 4
fc − f m Ac −Ac
4 4
2. SNBFM (t ) + SAM (t ) = SSB-SC → USB-FC

SAM (t ) − SNBFM (t ) = SSB-SC → LSB-FC
3. LSB in NBFM is 1800 inverted w.r.t to LSB in AM

➢ SFM (t ) = Ac  J n ()cos[2 ( fc + nf m ) t ]
n=−
For aby value of 
➢ Jn () = (−1)n Jn ()


➢ J
n=−
2
n () = 1
➢ J0 () = 0,  = 2.4,5.5,8.6,11.8
➢ as n , → J n () 
  1: S (t ) → 1 Carrier + 2 SB NB Angle Modulation
If   1: S (t ) : 1 Carrier + Infinite SB Wide Band Angle Modulation
Ideal BW of WBFM = 
Carson’s Rule –
BW = (+1)2 fm PM for PM
FM for FM
Ac2
➢ Power of Carrier before modulation = = Pc
2
Power of Carrier after modulation P = Pc  J 0  + 2( J1 () + J 2 () + 
2 2 2
➢
➢ Power of Carrier component in modulated signal = Pc J02 ()
➢ PSB = 2Pc  J12 () + J 22 () + 
PSB
➢ =
PTotal → Pc  J () + 2( J12 () + J 22 () + ) 
2
0
If J0 () = 0 then  = 100%

➢ For Infinite sidebands PWB = Pc
For Non sinusoidal –


SFM = Ac  Cn cos  2 ( fc + nf m ) t + Cn 
n=−
m(t ) BW
Singletone sinusoidal → (+1)2 fm

⎯⎯
Non sinusoidal → (+1)2 fm , fm = fundamental frequency

⎯⎯
periodic signal
Other Cases → (+1)2 fmax
⎯⎯
BW = (1+)2 fm or 2(f + fmax )

Frequency Mixture and Multiplier
Mixture/Multiplier Input Mixture Output (Multiplied by n) Multiplier Output
Ac Ac' Ac'
fc fC − f L or fC + f L = fC' nfc
  n
fm fm fm
f f n f
BW BW (n+1)2 fm
Spactral spacing fm fm
Frequency components fc' , fc'  fm , fc'  2 fm nfc , nfc  fm , nfc  2 fm
Wideband Angle Modulation generation –

 dm(t ) 
PM [m(t )] = FM  If K p = K f = K
 dt 
t 
FM [m(t )] = PM   m() d 
− 
Wideband FM Generation Methods

1. Armstrong Method (Indirect Method)
2. Direct Method
• VCO (Voltage Controlled Oscillator is used). It is modified version of Hartley oscillator
 C
=
c 2C0
FM Demodulator
1. Theoretical method
2. Practical method. PLL (Phase Locked Loop)
Kf
(1) v(t ) = m(t )
KV
(2) Lock mode → Frequency lock
Capture mode → Phase lock
(3) L.R  C.R
Super Hetrodyne Receiver
f L = Local oscillator frequency
f S = Desired frequency
f Si = Frequency of image station
Case 1 : If relation between fl and fs is not mentioned.
Assume : fl  f s
1. fl = fs + IF
2. fSi = fl + IF
3. fSi = fs + 2IF
Case 2 : When relation between fi and fs is given
If fSi  fl  fs If f S  f s  f sl ' then Case 1
then 1. fs = fl + IF
2. fl = fSi + IF
3. fs = fSi + 2IF
Image Rejection Ratio
IRR = 1 + P2Q2
Q : Quality factor of Oscillator
f Si2 − f s2
P= fSi  fs P2Q2  1
f Si f s
IRR = PQ

3 RANDOM VARIABLE AND

RANDOM PROCESS
3.1. Introduction
Random variable → Real and complex
• R.V. is a function performing mapping from sample space of R.E. to real line.
• X () : Random variable
• Domain of R.V. →  (Sample point)
• Range of R.V. → Subset of real line
• One to one or many to one mapping
• P{X  a} → Probability of set in which all the comes satisfy x ()  a .
CDF of R.V.
Let random variable X, x →Values taken by R.V.
(1) FX ( x) = P{X  x} = 1 − P{X  x}
(2) FX (a) = P{ X  a} = 1 − P{ X  a}
(3) F| X | ( y) = P{| X | y} = P{− y  X  y}
Properties
(1) FX () = 1
(2) FX (−) = 0
(3) FX () + FX (−) = 1
(4) FX ( x) = P{ X  x}  0  FX ( x)  1
(a) CDF always non negative.
(b) Lower bound: FX ( x) = 0 , upper Bound = 1
 dF ( x) 
(5) CDF is monotonically non decreasing function of x  X  0
 dx 
(6) Graph of CDF is always amplitudes continuous from right.
• Key point :
(1) P{a  x  b} = FX (b+ ) − FX (a+ )
(2) P{a  X  b} = FX (b+ ) − FX (a− )
(3) P{a  X  b} = FX (b− ) − FX (a+ )
(4) P{a  X  b} = FX (b− ) − FX (a− )

P ( x = a)
CDF is continuous at x = a CDF is continuousat x = a

+ −
P{x = a } = FX ( a ) − FX ( a ) P{ x = a} = FX ( a + ) − f X ( a − )
P{ x = a} = 0 0
P{ x = a} = Size of Jump
Probability Density Function

Random variable X
x →Variable taken by R.V.
f X ( x) → Symbol
dFX ( x)
f X ( x) =
dx
x
f X ( x) =  f X ( x) dX
−
a
f X ( x) =  f X () d 
−
Properties :
(1) f X ( x)  0 → Non negative
(2) 0  f X ( x)   Upper bound
Lower bound

(3) FX () =  f X ( x) dx = 1
−
(4) Graph of PDF can be even or NENO but cannot be odd.

x
(5) P{−  X  x} =  f X () d 
−
b+
(6) P{a  X  b} = + f X ( x) dx
a

b+
(7) P{a  X  b} = − f X ( x) dx
a
b−
(8) P{a  X  b} = + f X ( x) dx
a
P (x = a )
Graph having impulse at x = a No impulse

+
P{ x = 0} = FX ( a ) − FX ( a ) − at x = a +
a
f
a+
P{x = a } = ( x) dx
f
Y
P{x = a } = Y ( x) dx a−
P{ x = a} = 0
−
a
= Total area
Discrete Random Variable:

(1) PDF should have impulses only.
(2) CDF should have staircase only.
(1) Probability mass function of DRV : Let X is D.R.V.
PX ( x) = P( X = x) probability such that X = x
➢ 0  PX ( x) = 1
➢  PX ( x) = 1
x
(2) PDF of a D.R.V : Let X is O.R.V.

f X ( x) =  PX ( xi )  ( x − xi ) =  P( x = xi )  ( x − xi )
i i
(3) CDF of a D.R.V. : Let X is D.R.V.

x
FX ( x) =  f X ( x) dx
−
FX ( x) =  P{X = xi }u{x − xi }
x
(4) P{X = a} may or may not be zero.
Continuous Random Variable

Maps sample point to continuous range of values on real axis.
(1) PDF of C.R.V should not contain impulses at all.
(2) CDF of C.R.V
➢ Should not contain jump type discontinuity
➢ It should be amplitude continuous every where
(3) PMF not defined for C.R.V because for CRV P{ X = a} will always be zero.
P( A  B)
P( A / B) = → Conditional probability of A given B.
P( B)
 B
P( A  B) = P(B)P( A / B) = P( A)P   = Joint probability.
 A
Expectation operator : Performs operations on R.V. only.
Linear Operator

  xf X ( x) dx X : CRV

E [C] = C, E [C 2 ] = C 2 E ( X ) =  −
xi P{X = xi } X : DRV
 i
E [aX ] = a E [ X ]
E [aX + b] = a E [ X ] + E[b]
E [ag ( X ) + bH ( y)) = a E [ g ( x)] + b E [ H ( y)]

E[ X 2 ] = x
2
f X ( x) dx
−

E  g ( x)  =  g ( x) f X ( x) dx
−
Gaussian Random Variable

CRV X is having Gaussian or random distribution.
X is having Gaussian PDF, X is called G.R.V.
E [ X ] =  X , E [( X − x )2 ] = Variance = 2X
−( x− x )2
1 22X
X N{ X , 2X } f X ( x) = e −  x  
22X
Key Point :
−( x − X )2

1 22X
(1)  22X
e dx = 1
−
−( x − X )2

1 22X
(2)  22X
e dx =  X = E [ X ]
−
−( x − X )2
 X
1 −( x −  X )2 1 22X 1
(3)  22X
e
22X
dx =  22X
e dx =
2
X −

Zero mean Gaussian distribution-

X N ( X ,2X )  X N [0, 2X ]  E [ X ] = 0
− x2
1 2
f X ( x) = e 2 X
22X
Zero Mean, unit variance :

− x2  − x2
1 1 1
X N (0.1) f y ( x) =
2
e 2 ,
 2
e 2 dx =
2
0
Q- function :

1 2
Q ( x) =  e− z /2dz as x , Q ( x) 
2 0
Q () = 0, Q(−) = 1, Q (0) = 0.5, Q( x) + Q(− x) = 1
 z − X 
P [ X  z ] = Q ( P) = Q  
 X 
 z − X 
f X ( z) = P( X  z ) = 1 − P( X  z ) = 1 − Q  
 X 
Statistical averages of a R.V.

nth order moment about origin-
 n
  x f X ( x) dx X : CRV

E[( X − 0)n ] = E [ X n ] =  −
 xn P{X = x } X : DRV
 i i i
1st order moment about origin


E[ X ] =  xf X ( x)dx =i xi P{X = xi }
−
E [ X ] = X =  X = mi → dc value, avg. value Mean value
[E [ X ]]2 → d.c. power

2nd order moment about origin-

  2
  x f X ( x) X : CRV
E [( X − 0)2 ] = E [ X 2 ] = X 2  −
 x P{ X = x } X : DRV
i i i
E [ X 2 ] = Mean square value of R.V. X = Total power of R.V. x

1st order moment about mean - E [( X −  X )] = 0
2nd order moment about mean −E[( X −  X )2 ] = E[ X 2 ] − 2X
 2X = E [ X 2 ] − u2X
A.C. Total dc
Power Power Power
Important point:
(1) 2X  0, E [ X 2 ]  2X
(2) If X is zero mean R.V.
E[ X 2 ] = 2X , MSV( X ) = Var( X )

(3) Standard deviation
Variance = 2X =  X
(4) Y = aX + b
E[Y 2 ] = a2 E [ X 2 ] + b2 + 2ab E [ X ]
Y2 = a22X
Standard Distribution of R.V.

(1) Uniform distribution X U [a, b]
 1
 a X b
f X ( x) =  (b − a)
 0
 otherwise

a+b a2 + b2 + ab (b − a)2
E[ X ] = , E[ X 2 ] = , 2X =
2 3 12
(2) Triangular distribution
X tri (a, m, b)
a+m+b
E[ X ] =
3
(3) Rayleigh Distribution

X → CRV
 − x2

 x 22X
f X ( x) =  2 e x0

 X

 0 else
− x2

x 2
 2 e 2 X dx = 1
0 X
If X and Y are two G.R.V. Then Z = X 2 + Y 2 will have reyleigh distribution.

(4) Exponential Distribution : If CRV has exponential distribution then it will have PDF
e−x 
x0 −x
f X ( x) = 
 0 x0  e dx = 1
0
Laplacian Distribution
X → CRV
f X ( x) = ae−b|x| −  x  
a
If = 1 , a  0, b  0
b


 ae
bx
x0
f X ( x) = 
−bx

ae x0
Discrete Random variable-Binomial, Position distribution
Binomial distribution necessary condition:-

(1) The no of trials n showed be finite.
(2) Trials are independent
(3) Each trials should result in 2 outcomes success or failure.
(4) Prob of success in each trial should be constant.
PMF: P{X = r success} = ncr pr qn−r
E [ X ] =  xi p{X = xi } = n p 2X = npq

i
E[ X 2 ] = npq + (np)2
Std deviation  X =  npq
Position Distribution
Specific type of binomial distribution where n →
n →very large, p → very small, np → finite  = np
r e − 
p{X = r} = probability of X = r (success)
r!
E [ X ] = , 2X = 
If Y = g ( X ) is having monotonic TX .
PDF
Given X ⎯⎯⎯
→ f X ( x) ,
 dx 
fY [ y] =  f X [ x]  function of y
 dy 
(1) case a > 0

 y −b  1  y −b 
FY [ y] = FX   , fY ( y) = f X 
 a  a  a 
(2) Y = −aX + b a  0
 y −b  1  y −b 
FY ( y) = 1 − FX   , fY ( y) = f X 
 −a  a  −a 
Monotonic linear TX:

y = aX + b
X U [m1 , m2 ] → Y U [am1 + b, am2 + b]
X  [m1, m2 , m3 ] → Y  [am1 + b, am2 + b, am3 + b]
X N [ X , 2X ] → Y N [Y , 2y ]
Y N[a X + b, a22X ]
Monotonic Non-Linear TX:

X → f X ( x)
Y → X 3 , fY ( y) = ?
 dx 
fY ( y) =  f X ( x) 
 dy 
1
fY ( y) = f ( y1/3 )
2/3 X
3y
Non - Monotonic TX:

Y = y, g ( X ) = y, X = g −1( y)
→ x1

→ x2
→ x
 3
dx1 dx
fY ( y) = f X ( x1 ) + f X ( x2 ) 2 + − − − −
dy dy
2D Random variable :
→ FX ,Y ( x, y) = Joint CDF

( X , Y ) → 2DR.V. → f X ,Y ( x, y) = Joint PDF

→ PXY ( xi , yi ) = Joint PMF
If A and B are independent
 A  B
P   = P( A), P   = P(B), P( A  B) = P( A)P(B)
 B  A

If X and Y are independent R.V.

FXY ( x, y) = f X ( x) fY ( y)
 y  x
➢ f XY ( x, y) = f X ( x) f Y   = fY ( y ) f X  
X
x Y  
y

f XY ( x, y) = f X ( x) fY ( y)
 yj   xi 
➢ PXY ( xi , y j ) = PX ( xi ) PY   = PY ( y j ) PX  
 yj 
X  i 
x  
Y
PXY ( xi , y j ) = PX ( xi )PY ( y j )
Joint CDF = Let ( X , Y ) are BIVARIATE R.V.
FXY ( x, y) = P{X  x)  (Y  y) = P{X  x; Y  y}
Properties:
(1) 0  FXY ( x, y)  1
(2) FXY (−, y) = P{( X  −)  (Y  y)} = 0
(3) FXY ( x, −) = 0
(4) FXY (−, −) = 0
(5) FXY (, ) = 1
(6) FXY ( x1, y1 ) = P{( X  x1 )  (Y  y1 )}
(7) P{( x1  X  x2 )  ( y1  Y  y2 )}
= FXY ( x1+ . y1+ ) + FXY ( x2+ , y2+ ) − FXY ( x1+ , y2+ ) − FXY ( x2+ , y1+ )
 y x
(8) FXY ( x, y) = FX ( x) FY   = FY ( y) FX  
X
x Y  
y
(9) X and Y are independent R.V.

FXY ( x, y) = FX ( x) FY ( y)
(10) FX ( x, y) = FXY ( x, ), FY ( y) = FXY (, y)

Conditional CDF
 x  F ( x, y)
FX   = XY of FY ( y)  0
Y  y FY ( y)
 x  P[( X  x)  (Y  y)]
FX   =
P [( X  )  (Y  y)]
Y  
y
Joint PDF
2 FXY ( x, y)
f XY ( x, y) =
X Y
x y
FXY ( x, y) =   f XY (u, v)du dv
− −
 
FXY ( x, y) =   f XY ( x, y)dx dy = 1
− −
Marginal PDF
 
(1) f X ( x) =  f XY ( x, y)dy, fY ( y) =  f XY ( x, y)dx
− −
If X and Y are independent f XY ( x, y) = f X ( x) fY ( y)
 y  x
f XY ( x, y) = f X ( x) f Y   = fY ( y ) f X  
X
x Y  
y
Conditional PDF
x y
f ( x, y) − −
  f XY ( x, y)dxdy
f XY ( x, y) = XY = 
fY ( y)
 f XY ( x, y)dx
−
Probability Calculation in 2-D region

Given Joint PDF
P{(a  X  b)  (c  y  d )} = ?
R1 : Region in which probability has to be calculated.
Method:
P ( X , Y  R1 ) =   f XY ( x, y)dxdy ( R = R1  R2 )
R
(1) X and Y not independent R.V.

P ( X ,Y  R1) =   f X ,Y ( x) fY (Y ) dxdy R = ( R1  R2 )
Rr
(Central Limit Theorem)

If X and Y are D.R.V
 PXY ( xi , y j ) = 1
i j
PXY ( xi , y j ) = P{( X = xi )  (Y = y j )}
Joint PMF
Marginal PMF :
PX ( xi ) =  PXY ( xi , y j )
j
PY ( y j ) =  PXY ( xi , y j )
i
Minimum of 2 independent R.V.

X, Y are two I.R.V
min ( X ,Y )  Z = ( X  Z )  (Y  Z )
P [min ( X , Y )  Z ] = P [ X  Z ]P [Y  Z ] =   f XY ( x, y)dxdy
R
P[min ( X ,Y )  Z ] = 1 − P[min ( X , Y )  Z ]
P[min( X ,Y )  Z ] =  f X ( x) fY ( y)dxdy
R R
Let Z = Max ( X , Y ) → R.V.
CDF of Z FZ (Z ) = FX (Z )  FY (Z )
PDF of Z f Z (Z ) = FX (Z ) fY (Z ) + FY (Z ) f X (Z )
Let Z = min[ X ,Y ] → R.V.
CDF of Z FZ (Z ) = f X (Z ) + gY (Z ) + fY (Z ) FX (Z )
PDF of Z f Z (Z ) = f X (Z ) + fY (Z ) − FX (Z ) fY (Z ) − FY (Z ) f X (Z )
Statistical parameters of 2D R.V.

(1) (k , r )th order joint moment about origin E [ X kY r ] (1,1)st order joint moment about origin.
E[ X 1,Y1] = E [ XY ] = RXY → Cross correlation between R.V. X and V.

➢ E [ XY ] = RXY = 0 → R.V. X and Y are orthogonal.
(2) (k , r )th order joint moment about mean-
E [( X − X )k (Y − Y )r ]
(1,1)st order joint Moment about mean-
E [( X − X )(Y − Y )] = E [ XY ] − XY = cov( X , Y )
cov( X ,Y ) = XY = E[ XY ] − E[ X ] E[Y ] = RXY −x  y

When 2 R.V. X and Y are uncorrelated-
cov( X , Y ) = 0 , E [ X ,Y ] = E [ X ] E [Y ]
➢ E [ X kY r ] = E [ X k ]E [Y r ] = X ,Y are independent.
➢ If 2 R.V. are independent then they has to be uncorrelated but converse is not necessarily true.
One function of two R.V.

W = aX + bY
(1) E [W ] = a E [ X ] + bE [Y ]
(2) E[W 2 ] = a2 E [ X 2 ] + b2 E [Y 2 ] + 2ab RXY
(3) W
2
= a22X + b2Y2 + 2ab cov( X ,Y )
One function of Three R.V. W = aX1 + bX 2 + cX 3

(1) E [W ] = a X1 + b X 2 + a X3
(2) E[W 2 ] = a2 X12 + bX 22 + c2 X32 + 2abX1X 2 + 2bcX 2 X3 + 2caX1X3
(3) W
2
= a22X1 + b22X 2 + c22X 3 + 2ab cov( X1, X 2 ) + 2bc cov( X 2 , X 3 ) + 2ca cov( X1, X 3 )
Var (X + Y) = Var (X – Y)
Only when X, Y are → uncorrelated and independent
Correlation coefficient
 XY cov( X ,Y )
( X ,Y ) = =
 X Y (Std.dev.of X )  (std.dev.of Y )
➢ −1    1
➢ ( X , X ) = 1, ( X , − X ) = −1
➢ X, Y are independent  ( X , Y ) = 0
Let X, Y are two R.V.


 g (Y )   y
E =  g ( y) f Y  dy
 X = x  − X
 x

 g( X )   x
E  =  g ( x) f X   dx
 Y = y  − Y  
y
Calculation of probability in n-D region
Theorem -1
If X1, X 2 , X 3 − − − X n are statistically independent random variables.
Let Z = X1 + X 2 − − − − X n
  
f X1 ( z) f X 2 ( z) f X n ( z)
f Z ( z) = f X1 ( z)* f X 2 ( z) − − − − * f X n ( z)
When and only when all the R.V. are statistically independent.
➢ R.V. are linearly combined.
Theorem-2
X1, X 2 , X 3 − − − X n are statically independent non Gaussian R.V.
Z = X1 + X 2 + X 3 − − − − X n
f Z ( z) = f X1 ( z)* f X 2 ( z) − − − − * f X n ( z)
If n → f Z ( z ) = Gaussian irrespective of nature of ( X i )in=1
Theorem-3
X1, X 2 , X 3 − − − X n are statistically independent G.R.V.
Z = X1 + X 2 − − − + X n
fZ ( z) = f X1 ( z) + f X 2 ( z) + − − − − f X n ( z)
n → Finite| infinite, → Z: GRV
Problem Solving Technique :

Case 1 : X1, X 2 − − − − X n are statistically independent G.R.V.
 a
P[ X1 + X 2 + X 3  a] = P(Z  a) =  f Z ( z )dz = 1 −  f Z ( z )dz
 a −
Non GRV
Where Z = X1 + X 2 + X 3

f Z ( z) = f X1 ( z)  f X 2 ( z)  f X3 (z)
Case 2 : If X1, X 2 , X3 are statistically independent G.R.V.

 a − z 
P( X1 + X 2 + X 3  a) = P [Z  a] = Q  
 z 
Z = X1 + X 2 + X 3
2 =  X1 +  X 2 +  X 3 , 2z = 2X1 + 2X 2 + 2X 3
Note : If X1, X 2 , X 3 − − − − X n are I.I.D. random variables

1
P (one of them is largest) =
n
1
P (one of them is smallest) =
n
Random Process
X ( , t ) = {X (1, t), X ( 2 , t )} Collection of sample function
of
Euemble of sample function
Random process or Random signal
or stochastic signal
X (1 , t1 ) → sample value, values taken by R.V. When R.P. is observed at t = t1
C.T.R.P → It maps the sample points onto continuous time sample function, collection of continuous time sample function.
t =t1
X (t ) = Acos(0t + ) ⎯⎯⎯ → X (t1) = Acos(0t1 + )
C.T.R.P R.V., D.R.V.
t → Continuous time, A → Constant, 0 → constant
 U [−, ] → CRV
Any typical R.P can be understood as x(t ) = f (t , )
(Function of time and R.V.) x(n) = f (n, )
Statistical parameter of R.P.
Case 1 : X (t ) X (t0 ) − CRV

t = t0

E [ X (t0 )] =  x f X (t0 ) ( x) dx
−

E [ x2 (t0 )] = x 2X (t0 ) = E [ X 2 (t0 )] − 9E ( X (t0 ))2 ]
2
f X (t0 ) ( x) dx ,
−
X (t ) X (t 0 ) − DRV
Case 2 : t = t0
CTRP

E[ X (t0 )] =  xi PX (t0 ) ( xi ) =  xi P{X (t0 ) = xi )

i i
E[ x 2
(t0 )] =  xi2 PX (t0 ) ( xi ) =  xi2 P{X (t0 ) = xi )
i i
2X (t0 ) = E [ X 2 (t0 )] − ( E [ X [t0 ])2

Case 3 : DTRP → CRV
E [ X (n0 )], E [ x2 (n0 )], 2X (n0 ) → Some as case 1, replace to by n0
Case 4 : DTRP → DRV
E [ X (n0 )], E [ x2 (n0 )], 2X (n0 ) → Same as case-2, Replace to by n0
CTR.P
t = t1 X (t1 )
X (t )
t = t2 X ( t2 )
E[ X (t1) X (t2 )] = RX (t1) X (t2 ) = RXX (t1, t2 )
Auto correlation of RP X(t)
X (t )
X (t )
X (t + )
Then E[ X (t1 ) X (t + )] = RXX (t , t + )
Cov ( X (t1 ) X (t2 )) = E [ X (t1 ) X (t2 )] − E[ X (t1)]E[ X (t2 )]]
XX (t1, t2 ) = RXX (t1, t2 ) −X (t1)X (t2 )
Auto covariance of R.P. X (t )
XX (t, t + ) = RXX (t, t + ) −X (t )X (t +)
Cross Correlation
X (t ) ⎯⎯⎯
t =t
→ X (t1), Y (t ) ⎯⎯⎯
t =t
→Y (t2 )
R.P 1 R.V R.P 2 R.V
E [ X (t1 )Y (t2 )] = RXY (t1, t2 )
(1) If RXY (t1, t2 ) = 0 V t1  TR X (t ) and Y (t ) R.P. will
t2  TR Become orthogonal.
(2) Cov [ X (t1),Y (t2 )] = RXY (t1, t2 ) − X (t1) X (t2 )
= 0 V t1  TR
t2  TR

RP X (t ) and Y (t ) are uncorrelated.
If X (t1 ) and X (t2 ) are independent
 E[ X (t1 ) E[ X (t2 )] t1  t2
E[ X (t1 ) X (t2 )]  2
 E[ X (t1 )] t1 = t2
Same for DRV, replace t by n.
Types of R.P.
(1) Strict sense stationary R.P. → R.P. should be independent of time shift
X (t1 ) X (t2 ) − − − X (tk )
X (t ) →
kRV . .
Kth order Joint PDF-

( x1, x2 − − − xk ) …(i)
f X (t1 ) X (t2 ) − − − X (tk )

X (t1 + ), X (t2 + ) − − − X (tk + )
X (t ) :
k R.V.
Kth order Joint PDF-

( x1, x2 − − − xk ) …(ii)
f X (t1+) X (t2 +)−−− X (tk +)
(i) = (ii) → X (t ) is solid to be SSSRP.
f X (t1) ( x) = f X (t2 +) ( x) independent of time

2nd order joint PDF is independent of time shift.
f X (t1) X (t2 ) ( x1, x2 ) → f X (0) X (t2 −t1) ( x1, x2 )
➢ Does not depend on individual sampling instances t1 and t2
➢ Depends on time difference between sampling instances t1 and t2
E[ X (t1 ) X (t2 )] = E[ X (0) X (t2 − t1 ) = RXX (t1, t2 )

E[ X (0) X (t2 − t1 )] = RXX (0, t1 − t2 ) = RXX (t1 t2 )
XX (t1 t2 ) = RXX (t1 t2 ) − 2X

WSSRP → There are stationary RP which are stationary at least upto 2nd order.

(1) E[ X (t )] =  X Constant
(2) E [ X 2 (t )] = Constant
(3) 2X (t ) = Constant
E[RP] = E[RV]
MSV(RP) = MSV(RV)
Var (RP)=Var (RV)
(4) E [ X (t1 ) X (t2 )] = RXX (t1 t2 )
E [ X (t + ) X (t )] = RXX (−)
E [ X (t ) X (t + )] = RXX ()
RX X () = RXX (−)
2X (t ) = RXX (0) − 2X
Cov [ X (t ) X (t + )] = RX X () − 2X
(5) E[ X 2 (t )] = RXX (0) =0

E[ X (t )]E[ X (t + )] =  X
2
(  0)
(6) E[ X (t ) X (t + )] = RXX ()ACF = 
 E[ X (t )] = RXX (0) ( = 0)
2

(7) cov[ X (t ) X (t + )] = RXX (t , t + ) −  X (t ) X (t + )
 0 0
C XX () = RXX () − 2X = 
 RXX (0) −  X =0
2

 
2
0
RX X () =  X
RX X (0)  = 0

Important point :
(1) If X(t) is zero mean WSSRP.
E [X (t) = 0

2X (t ) = E [ X 2 (t )]
Var [ X (t )] = MSV{X (t )}
Var { X (t = t1 )} = MSV{ X (t = t1 )}
(2) If X (k) is zero mean WSSRP
E[ X (k )] = 0 2X (k ) = E[ X 2 (k )]
Var [ X (k )] = MSV[X (k )]
➢ X (t ) : WSSRP+IIDRP
E [ X (t )] =  X , E[ X (t + )] =  X , E[ X 2 (t )] = RXX (0) = Constant
2X (t ) = Constant
Let Y (t ) = X (at + b)
E [Y (t ) =  X , E[ y2 (t )] = RXX (0)
Y2 (t ) = RXX (0) −  X 2
E[Y (t )Y (t + )] = RXX (a) = RYY ()
Cov [Y (t )Y (t + )] = CYY () = RXX (a) − 2X
→ Y =  X = Constant
For Y (t ) →  ,Y (t ) → WSSRP
 RYY () = RXX (a)
➢ Time shift, Time reversal, time scaling does not affect stationary nature of R.P.
Let Y (t ) = aX (t ) + b, X (t ) is WSSRP
E [Y (t )] = a X + b = Constant
E[Y 2 (t )] = a2 RXX (0) + b2 + 2ab X = Constant
Y2 (t ) = a22X (t )
Cov [Y (t )Y (t + )] = RYY () − Y2
E[Y (t )Y (t + )] = a2 RXX () + 2ab X + b2 = RYY ()
y(t ) → WSSRP
➢ Linear transformation of WSSRP does not change its stationarity.
➢ If WSSRP passed through LTI system, output is also a WSSRP.

E [Y (t )] = (a + b) X
E [Y 2 (t )] = a2 RXX (0) + b2 RXX (0) + 2abRXX (0 )
Y2 (t ) = (a2 + b2 )2X (t ) + 2ab[ RXX (T0 ) − 2X ]
RYY () = (a2 + b2 )RXX () + abRXX ( − T0 ) + abRXX ( + T0 )
CYY () = a2CXX () + b2CXX () + abRXX ( − T0 ) + abRXX ( + T0 ) − 2ab2X
E[ X (n) X (n + k )] = [k ] = RXX (k ) IIDRP
E [ X (n) X (n + k )] = E [ X 2 (n))(k = 0)
A, 0 → constant,  U (0,2)OR  U [−, ]
X (t ) = A cos(0t + )
E [ A cos(0t + )] = 0
E [ A cos(0t + + )] = 0
A2 2 A2
E [ X 2 (t )] = ,  X (t ) =
2 2
A2
E [ X (t ) X (t + )] = cos 0 = RXX ()
2
A2
Cov [ X (t ) X (t + )] = cos 0
2
X (t ) : WSSRP + periodic with 0 → RXX () will also be periodic with same T.P.
ERGODIC Random Process :

Time Avg = Statistical Aug.

1
T
X (t ) dt = E [ X (t )]
Auto Correlation and its properties

Similarity between 2 Samples
Let X (t ) is WSSRP, X (t ) is observed  duration apart
(1) E [ X (t ) X (t + ) = RXX ()
(2) RXX (−) = RXX () : Even
(3) RXX ()  RXX (0)
(4) {RXX ()}max = RXX (0) = Maximum similarity

  0
(5)  RXX ()d  = 2  RXX ()d  = 2  RXX ()d 
− 0 −
(6)
(7)
(8) X (t ) is power signal
RXX (0) = E[ X 2 (t )] = 2X ( t ) + 2X ( t )
Total power A.C power D.C power

of R.P of R.P of R.P
(9) If X (t ) is ergodic and WSSRP, it has no periodic component
E [ X (t ) =  X  0
2X = lim RXX () = lim RXX ()

→ ||→
If not ergodic but WSSRP then RXX (0) = E[ X 2 (t )] RXX ()  2X
,

Important point:
X (t )  X () X (t )  X ( f )
 
X () =  x(t )e− jt dt X( f ) =  x(t )e
− j 2ft
dt
− −
 
1 jt j 2ft
x(t ) =  X ()e dt
2 −
x(t ) =  X ( f )e dt
−
 
X (0) =  x(t )dt x(0) =  X ( f )df
− −
 
1
x(0) =  X ()d 
2 −
X (0) =  x(t)dt
−
3.2. Parserval Theorem

  
1
Ex(t ) =  x2 (t )dt =  | X () | d  =  | X ( f ) | df
2 2
−
2 − −
Energy spectral density X (t ) → WSSRP, Engery
ESD
x(t ) ⎯⎯→
F.T
X ( )  | X () | 2 = G XX ( )
x( t) ⎯⎯
→ X ( f )  | X ( f ) | 2 = G XX ( f )
ESD of x(t)
F .T .
E[ X (t ) X (t + )] = RXX () ⎯⎯⎯ →GXX ()
F.T .
RXX () ⎯⎯⎯ →GXX ( f )
F .T
ACF( X (t )) ⎯⎯ ⎯ ESD[ x(t )]
 
GXX (0) =  RXX () d  = 2  RXX () d 
− 0
Zero freq. value of ESD = Area under ACF


1 Area under ESDG XX ()
RXX (0) =


2 −
GXX ()d  =
2
E[ X 2 (t )]

 GXX ( f ) df = Area under ESD G XX ( f )
−
Energy Calculation :
  
1
EX (t ) =  | X (t ) |2dt =  | X () | d  =  | X ( f ) | df
2 2
−
2 − −
 
1
=  GXX () d =  GXX ( f ) df
2 − −
GXX () = GXX (−)
Power spectral density – (PSD)

X (t ) → Power signal, WSSRP
PSD 1
X (t ) ⎯⎯ → S XX () = lim | X T () |2
T → T
PSD 1
X (t ) ⎯⎯ → S XX ( f ) = lim | X T ( f ) |2
T → T

− j
(1) E [ X (t ) X (t + )] = RXX () S XX () =  RXX ()e d
−
F .T .
(2) ACF [ X (t )]⎯⎯ →PSD[ X (t )]

− j 2f 
 RXX ()e
F .T .
RXX ⎯⎯ → XX () S XX ( f ) = d
−
 
(3) S XX (0) =  RXX () d  = 2  RXX () d 
− 0
Zero freq. value of = Area under ACF

PSD
 
1
(4) RXX () =  S XX ()e jd  =  S XX ( f )e jdf
2 − −
1 
  S XX ()d 
 2 −
RXX (0) = 


  S XX ( f )df
 −

(5) RXX () = RXX (−), S XX () = S XX (−)

(6) Calculation of power
1  Area under PSD

 2
 S XX ()d  =
2
E[ X 2 (t )] = RXX (0) =  −


  S XX ( f )df = Area under PSD
 −
 
1
 0
E [ X 2 (t )] = RXX (0) = S XX ()d  = 2  S XX ( f )df
0
Total power
A.C. Power = 2 X (t ),D.C.Power = 2 X (t )
Mean or Aug value

0+
1
E[ X (t )] =
2 − S XX ()d 
0
 0+
 1 S XX ()d 
 2 −

E[ X (t )] = 2X (t ) = +0
0
  S XX ( f )df →
0−
Is non-zero only when impulse is not prelent at zero frequency

 0− 
 1 S XX () d  + 1 S XX () d 
 2  2 +
 − 0
2X (t ) = −
 0 

 −
 XX
S ( f ) df + + S XX ( f ) df
0
• If X (t) is real, PSD is also real.

• PSD is even.
• PSD is non-negative, S XX ()  0; S XX ( f )  0
ESD of Modulated Signal (Band Pass Signal)

ESD
X (t ) ⎯⎯ → GXX ( f )

Base band R.P.
Ac
Y (t ) = X (t ). Ac cos2fct Y( f ) = [ X ( f + f c ) + X ( f − f c )]
 2
Band pass R.P.

ESD
Or ⎯⎯ →GYY ( f ) =| Y ( f ) |2
ESD
X (t ).Ac sin 2fct ⎯⎯ →GYY ( f ) =| Y ( f ) |2
Ac2
GYY ( f ) = [GXX ( f − fc ) + GXX ( f + fc )] fc  f m
4
Ac2
RY () = RX ()cos2fcT
2
PSD of Modulated Signal (Bandpass Signal)

X (t ) → power single
Y (t ) = Ac cos(2 fct ). X (t ) → SYY ( f ) → PSD
| YT ( f ) |2
SYY ( f ) = lim
T → T

Ac2  | XT ( f − fc ) |2 
| X ( f + fc ) |2 
SYY ( f ) =  lim + lim T 
4 T → T T → T 

Ac2
SYY ( f ) = [S XX ( f − fc ) + S XX ( f + fc )]
4
Ac2
RYY () = RXX ()cos2fc 
4
Ac2 Ac2
RY () = RX ()cos c  RY () = RX ()cos c 
2 2
Ac2 Ac2
SY ( f ) = [S X ( f − f c ) + S X ( f + fc )] SY ( f ) = [S X ( f − f c ) + S X ( f + fc )]
2 4
3.3. Cross Correlation

X (t );WSSRP,Y (t ) : WSSRP
E [ X (t )Y (t + )] = RXY (t , t + ) = RXY ()
E [Y (t + ) X (t )] = RYX (−)
RXY () = RYX (−)
•
RXY ()  RYX (−)

•
RXY () = RXY ( f ) → May/maynot be
• RXY ()  RXX (0)RYY (0)
1
• | RXY () |= RXX (0) + RYY (0)
2
Cross Covariance
CXY () = RXY () −  X Y
CYX () = RYX () − Y  X
Cross Spectral Density

F .T
RXY () ⎯⎯ → S XY ( f )
F .T
RYX () ⎯⎯ → SYX ( f )
If RP X (t) and Y (t) are orthogonal-
E[ X (t )Y (t + )] = RXY () = 0 = RYX (−)
E[Y (t ) X (t + )] = RYX () = 0 = RXY (−)
➢ X(t) and Y(t) are uncorrelated and atleast one of them have zero mean-
E [ X (t )] = 0or E [Y (t )] = 0
cov [ X (t )Y (t + )] = 0
C XY () = RXY () −  X Y = 0
RXY () = 0
RXY () = RYX () = 0
➢ X(t) and Y(t) are independent R.P. and atleast one of them have zero mean.
Cov [ X (t )Y (t + )] = 0
C XY () = RXY () −  X Y = 0
RXY () = 0
RYX () = 0 = RXY ()
Combination of WSSRP
Z (t ) = X (t )  Y (t )

RZZ () = RXX () + RXY ()  RYX ()  RYY ()
SZZ ( f ) = S XX ( f ) + SYY ( f )  S XY ( f )  SYX ( f )
If orthogonal [X (t) and Y (t)] then
RYX () = RXY () = 0
SYX ( f ) = S XY ( f ) = 0
Transmission of WSSRP through in LTI system
E  X (t ) = X E Y (t ) =  X  H ()=0
Y =  X H (0)
If x(t) is zero mean WSSRP then y(t) is also zero mean WSSRP
RXY () = RXX ()  h () and S XY ( f ) = S XX ( f ) H ( f )
RYX () = RXX ()  h (−) and SYX ( f ) = S XX ( f ) H (− f )
RYY () = RXX ()  h()  h(−)
SYY (t ) = S XX ( f ) H ( f ) H (− f ) if h(t) = real
SYY ( f ) = S XX ( f ) | H ( f ) |2 H ( f ) = H (− f )
 
PSDof O/P PSDof i/p
Power of Y(t)
 
E[Y 2 (t )] =  SYY ( f )df =  | H ( f ) |2 S XX ( f )df
− −


4 DIGITAL
COMMUNICATION
4.1. Sampling
Sampling converts C.T.S into D.T.S, it retains analog or digital nature of signal.
C (t ) : Impulse Train – Instantaneous or Ideal sampling

C (t ) : Rectangular Pulse Train : Natural sampling or Flat Top sampling.
Ideal instantaneous sampling

ms (t ) =  m (nTs )(t − nTs )
n=−
m(t ) = M ( f ) | M ()

M s () = f s  M ( − ns )
n=−

M s ( f ) = fs  M ( f − nfs )
n=−
1. If a low pass signal is sampled at f s  2 f m then it can be recovered from its samples, when
( f s  2 f m )  ( f m  fc  f s − f m )
( PBG = Ts )
f c = cut off free of ideal LPF at RX
2. f s = 2 f m , the sampled signal ms (t ) can be recovered into m(t ) if,
( f s = 2 f m )  ( fc = f m ) Ideal LPF
( PBG = Ts )

3. f s  2 f m , under sampling,
TX : Replace generation with ALIASING
R X : Recovery not possible
➢ ALIASING is overlapping of adjacent replica’s in sampled signal.
4.1.1. Low pass Sampling Theorem

A low pass sampling signal band limited to f max Hz, can be sampled and recovered from its samples when and only when
f s  2 f max at TX Proper LPF at RX

No Aliasing Recovery
Nyquist Rate and Nyquist Internal

Let m(t ) is lowpass signal bandlimited to f max Hz.
1 1
f NY = 2 fmax TNY = =
f NY 2 fmax
( f s )min = 2 fmax min → sampling rate which ensure to aliasing
S (t ) = m(t )cos c (t ) = ( fc + f m )

f s (max)
N R = 2( fc + f m ) = f NY
1
NJ = = TNY
2( fc + fm )
Combination of Two signals –

x1 (t ) → f max → f1, x2 (t ) → f max → f 2
m(t ) : Am cos mt , Am sin mt → f m


c(t ) :   (t − nTs ) → 0, f s ,2 f s ,3 f s
n=−
ms (t ) = m(t ) c(t ) = 0  fm , f s  fm , 2 f s  fm , 3 f s  fm

Sampling of signal by using general carrier pulse train

m(t ) ⎯⎯
→M ( f ) or M ()
 
c(t ) ⎯⎯
→ c( f ) =  Cn ( f − nf s ), C () =  2Cn ( − ns )
n=− n=−
 
M s (t ) = M s ( f ) =  Cn M ( f − nf s ), M s () =  Cn M ( − ns )
n=− n=−
Area in Ts
C0 =
Ts
Recovery – TX : f s  2 f m RX : LPF = Proper fc

1
y(t ) = m(t ) PBG =
C0
If c(t ) is rectangular pulse
➢ m(t ) is lowpass –

2A  2n 
ms (t ) = c(t ) m(t ) → M s ( f ) =  sin C   M ( f − nf s )
 a 
n=− a
Recovery- ( f s  2 f m )  ( f m  fc  f s − f m )
PBG of LPF y( t )
1 C0m(t)
1 M(t)
C0
K KC0 m(t )
Area in Ts
C0 =
Ts
➢ m(t) is sinusoidal –

m(t ) = Am cos2fmt ⎯⎯
→ fm
Ka
c(t ) = 0, f s ,2 f s , except n = K I, K  0
2
ms (t ) = 0  f m , f s  f m ,2 f s  f m ,
If c(t) is Triangular-Frequency absent n = Ka K  0, K  I
➢ Rest of the things same as Rectangular pulse
Bandpass Sampling :
m(t) is lowpass signal.
2 fH fH
fs = K= NR = 2 f H
K fH − fL
Previous Integer
f H = Maximum frequency component of Bandpass signal
Natural Sampling
m(t): Low pass sampling
c(t): Train of finite duration pulse or rectangular pulse

Ms ( f ) =  Cn M ( f − nfs )
n=−
Flat Top Sampling

Instantaneous sampling followed by a filter.

y(t ) =  m (nTs ) h(t − nTs )
n=−
Quantizer
Discretizes amplitude axis, analog to digital signal.

Dynamic Range of x(t ) = ( xmax − xmin )
Q input : Q : output : xˆ (t )

1.25 0  x(t )  2.5

xˆ (t ) = 
3.25 2.5  x(t )  4
➢ Many to one circuit.
Uniform Quantizer
DR of Q mL+1 − m1
= =
L L
1 = 2 = 3
L = Number of quantization level of Q.
Non uniform Quantizer

1. 1 = 2  3 i = i +1  L
2. 1  2  3 i  i +1  L
Quantizer Error –
X ⎯⎯ → Xˆ = Q( X )
→Q ( ) ⎯⎯

Q.E.P E  X QE
2 
=  xQE fQE [qE]dqE
2
 
−
When PDF of QE is given.


E  X QE
2 
= E ( X − Xˆ )2  =  ( x − xˆ)
2
f X ( x) dx
   
−
When PDF of R.V at input of Quantizer (X) is given.

(a) PDF is uniform
2  1
2
2 2
E  X QE =  A1 + 2  A2 + 3  A3 +
  12 12 12
12 2
= [Area of region in which step size is 1] + 2 [Area of region in which step size is 2] +
12 12

2
If quantization is uniform E  X QE
2 
=
  12
12 2
(b) PDF is stair case E  X QE
2 
=  A1 + 2  A2 +
  12 12
12 2
(c) PDF is non uniform E  X QE
2 
= A1 + 2 A2 +
  12 12
Signal power E[ X 2 ]
SQNR = =
Q E  P E  X QE
2 
 
(SQNR)dB = 10log10 SQNR
4.2. Pulse Transmission

1. Rectangular Pulse
➢ Bandwidth of wired channel for TX of single pulse of duration Tb :
(BW)  Rb
2. Since pulse
➢ BW of wired channel for TX of single sinC pulse having zero cross over or integer multiple of Tb.
Rb
(BW) 
2
Tb: Bit interval
Rb : Bit rate → Bit/sec

Rb
➢ Minimum transmission BW of a wired channel for baseband transmission is =
2
Number of levels L  2n Lmax = 2n

n is used to represent binary power quantization level.
M-ary Scheme
M = 2N
M  Number of different symbols of duration NTb each.
Ts = NTb : Symbol duration

N = Number of bits combined in binary sequence at a time
Pulse Code Modulation

Bit rate, Rb = nf s bits/sec
−   2
 Q  E  , E ( y)2 =
2 2 12
If Mid point Mapping is used.
DR of signal DR of Q
= =
L L
−   DR of signal
L  2n  Qe  Qe max = min =
2 2 2 2n
PQE = E [ X QE
2
] = E[ y 2 ] =  y 2 fY ( y) dy
When PDF of Q. Eis given.
L mi +1
PQN = PQE =   ( x − xˆi )2 f X ( x) dx
i = L mi
When PDF of X is given.
 −   2
If Qe U  ,  = PQE =
 2 2 12
12
SQNR = Ps
2
1
Bit Interval Tb =
Rb
B.W  Rb → Rectangular Pulse
Rb
B.W  → Since Pulse
2
Rb
(B.W)min = (BW) PCM =
2

Signal to Quantization Noise Power

m(t ) → Single tone sinusoidal
m(t ) = Am cos mt
Am2
1. Ps = m2 (t ) =
2
Am2
2. Ps = m (t ) =
2
3L2
2 Am
3. =
L
3 2
4. SQNR = L
2
5. (SQNR)dB = (1.76 + 20log10 L) dB
3
6. (SQNR)max = 4n
2
7. (SQNR)max dB = (1.76 + 6n) dB
x(t) is uniformly distributed [–Am, Am]
Am2
1. Ps =
3
Am2
2. PQE =
3L2
2 Am
3. =
L
4. SQNR = L2
5. (SQNR)dB = 20log10 L
6. (SQNR)dB  6n dB
7. (SQNR)max dB = 6n dB
Key point:
3 2
1. SQNR = L
2
3
2. (SQNR)max = 4n
2
If n → n  k
(SQNR)max → 4k Rb = nf s

3. (SQNR)max = (1.76 + 6n) dB

n →n  k
( SQNR)max → 6 dB
4. n given : L = 2n
5.
6.
7. n calculation : nmin
8. Default m(t ) = Sinusoidal
Drawback of PCM
nf s 2
BW = , PQE =
2 12
n  ⎯⎯
→ L  ⎯⎯
→  ⎯⎯
→ PQE  ⎯⎯
→ BW 
n  ⎯⎯
→ L  ⎯⎯
→  ⎯⎯
→ PQE  ⎯⎯
→ BW 
4.3. DPCM (Differential Pulse Code Modulation)

4.3.1. PCM vs DPCM
1.  fix for Both Q –
(BW) PCM  (BW) DPCM
(SQNR) PCM = ( SQNR) DPCM
D.R at input of Q of PCM is greater than DPCM.
2. L fix for Both Q –
(BW) PCM = (BW) DPCM
(SQNR) PCM  (SQNR) DPCM
3. In case of DPCM the difference between current sample and predicted value of current sample is Quantized, Encoded,
Line Coded and Wired TXed .

Delta Modulator
• The recovered signal is “stair-case” approximation of tXed original analog message signal.
• Stairs are added or substracted of sampling instance.
• Size of each stair is  = Step size of stairs
Tracking Error in DM.

1. Slope overload Error
dm(t ) 
 Occurance of S.O.E
dt max Ts
➢ To avoid SOE,   by keeping Ts constant such that –
d 
m(t )   For sinusoidal m(t ) = Am cos nt
dt max Ts
f s
Am 
m
2. Granular Error
It occurs when  is large.
➢ To remove it → small
If SOE , G.E  and vice versa.
SQNR in DM
2  
2
PQE = E  X QE =
  3
Ps 3P
1. SQNR = = 2s f H = cut off frequency of LPF
PQE 
3Ps f s 3Ps f s
2. (SQNR)D =  = 
 2 f m 2 f H
2
3 f 
3. (SQNR)max + m(t ) is sinusoidal + SOE avoid =  s 
80  fm 
3
3 f 
4. [(SQNR) D ]max + m(t ) is sinusoidal + SOE avoid =  s  → By default.
80  f m 


5 DIGITAL RECEIVER
5.1. Introduction
X (t ) → Deterministic signal process
N (t ) → random signal process
1. If but 1 is taxed → Receiver must recover but ‘1’.

2. If but 0 is taxed → Receiver must recover but ‘0’.
Output of Sampler:
 S + N01 1 TX
Y = S01 + N0 =  01 Channel noise is signal dependent
S02 + N02 0 TX
 S + N0 1 TX
Y = S0 + N0 =  01 Channel noise is signal independent
S02 + N0 0 TX
BER Calculation :
P(1 TX )  P(S1 (t ) : TX ) = P(S01 (t ) : Reception) = p
P(0 TX )  P(S2 (t ) : TX ) = P(S02 (t ) : Reception) = (1 − p)
At the input of decision device a condition R.V. is obtained

Pdf
N0 ⎯⎯⎯ →
Y   : Decide in favour of 1, or decides that bit 1 would have been Txed by Txer
Y   : Decide in favour of 0, or decides the bit 0 would have been Txed by Txer.
Average Bit Error Rate :
Problem Solving Technique:

Case 1: When PDF of Noise [noise R.V. at i/p of D.D] is given

(a)  is given
 1   0 
Pe = P(0 TX ) P   + P(1 TX ) P  1 T 
 0 TX   X

 1 
P  = P[Y (0 TX )  ] = P(S01 + N0  ) = P( N0   − S01) =  f N0 (no)dno
 0 TX  −S01
−S02
 0 
P
 1 TX
 = P[Y (1 TX )  ] = P(S02 + N0  ) = P( N0   − S02 ) =

 f N0 (no)dno
−
(b) opt → calculate , Pe min → calculate
Steps 1 :
1. Identify conditional PDF of conditional R.V. from the noise R.V. pdf (pdf of No.)
 S + N0 1 TX
Y =  01
S02 + N0 0 TX
2. Plot the conditional PDF one over another

3. Identify the overlapping or common region and decide range of 
(1     2 )
4. Choose any arbitrary  in the above range and calculate Pe = Pe ()
5. Pe () Vs 
(i) Pe () Vs  : Independent of 
(a) 1     2 → optimum  is every    (1 :  2 )
(b) Pe(min) = A
(ii) Pe () Vs  : Linear
1    2 →opt = 1
Pe ()min = Pe (opt )

(iii) Pe() Vs  : Non linear

 opt
d
Pe ( ) = 0 Pe ( =  opt ) = Pe
d min
Pe ( = opt ) = Pe min
(iv) If no overlapping region b/w conditional PDF
Pe () = 0 → BER is 0
Case 2: When PDF of conditional R.V. Y is gives at the input of the decision device
 → Given  opt → Calculate

Pe → Calculate Pe min → Calculate
 1   0 
(a) Pe = P(0 TX ) P   + P(1 TX ) P  1T 
 0 TX   X

 1  fY
P  = P(Y (0 TX )  ) =  0 T ( y)dy
 0 TX   X

 0  fY
P  = P[(1 TX )  ] =  1T ( y)dy
 1 TX  − X
(b) Same as case 1 (b)
MAP Analysis (Maximum A posteriori Analysis)

• MAP receiver always calculate min Pe .
• Calculation of opt  using
d
Pe () = 0
d
MAP Analysis
"1"
1 T  0T 
P  X   P  X 
 Y  "0"  Y 
"1" "1" opt = 
( y )  P(0 TX ) f y ( y )  y 
y
P(1 TX ) f  
1 TX "0" 0 TX "0"

ML Analysis : (Maximum Likelihood Analysis )

1
It is same as MAP analysis When P(0 TX ) = P(1 TX ) =
2
If noise is independent of signal then PDF of conditional R.V at input of decision device will have.
fY
( y)
0 TX Same Width
Shifted to different
+ parameters
fY Same Height
( y)
1TX
Key point : P(0 TX )  P(1 TX )
1. Noise is signal dependent/independent  min     max ,

min + max
=
2
P(0 TX )  P(1 TX ) opt  
P(1 TX )  P(0 TX ) opt  
2. If noise is signal dependent
 S + N0 1 TX
Y =  01
S02 + N0 0 TX
1
P(0 TX ) = P(1 TX ) = ; opt = 
2
Only when Y is having Non uniform PDF.
1
P(0 TX )  P(1 TX ) = ; opt  ; P(0 TX )  P(1 TX )
2
opt  ; P(0 TX )  P(1 TX )
When channel noise is Gaussian Random Process

 1   0 
Pe = P(0 TX ) P   + P(1 TX ) P  
 0 TX   1 TX 
Method 1 :
 1    −  y [0 TX ] 
P  = P[Y[0 TX ]  ] = Q   [0 T ] 
 0 TX   y X 
 0    −  y (1 TX ) 
P  = 1 − Q   [1 T ] 
 1 TX   y X 

Method 2 :
 1   ( − S02 ) −  N0 
P  = P  y[0 TX ]    = P  S02 + N0    = Q  
 0 TX   N0 
 0   ( − S01) −  N0 
P  = P  y[1 TX ]    = P  S01 + N0    = 1 − Q  
 1 TX   N0 
If PDF of noise at the input of D.D is given along with  –

 1  fY  y − 2 
P  = P  y[0 Tx ]   =  0 T ( y)dy = Q   
 0 TX   X

 0  fY  y − 2 
P  = P  y[1 Tx ]   =  1 T ( y)dy = 1 − Q   
 1 TX  − X
 optimum
d
1. Differentiation : Pe = Q() , Pe () = 0 → opt
d
d
2. Map Analysis : Pe () = 0 → opt
d
  + 2  y 2 P(0 TX )
opt =  1 + ln
 2  (1 − 2 ) P(1 TX )
Channel noise is Gaussian, signal and channel noise are independent
Pe () = Pe (opt ) = Pe min
1
3. ML Analysis : P(0 TX ) = P(1 TX ) =
2
1 + 2
opt =
2
 −  
Pe |min = Q  1 2 
 2 y 
Schwartz Inequality
 2  
 X1( f )X 2 ( f )df  
2 2
 X1 ( f ) df X 2 ( f ) df
− − −
P (S0 ) max = Es (t )  Eh ( t )
Max. signal power Signal energy Energy of

at sampling at input of filter h(t)
instance



2
Es(t ) = S ( f ) df
−


2
H ( f ) df = Eh(t )
−
W (t ) t =T
h (t )
WSSRP N 0 ( t)
ACF WSSRP
RW (t ) ACF
F.T. RN0 ()
SW ( f ) F.T.
SN 0 ( f )

N
PN0 (t ) = E  N02 (t ) = 0 
2
H ( f ) df
2 −
N0
E  N02 (t )  =  Eh(t )
2
Es(t )
(SNR)max =
( N0 / 2)
Only when H ( f ) = [S ( f )e j 2fT ]*
Energy of i/p pulse

( NSR)max at t = T =
PSD of i/p white noise
For General Noise
H ( f ) =  S ( f )e j 2fT  *
 PS 0 max Es (t )  Eh(t )
When (SNR)max = = 
PN0
 SN0 ( f )df
−
2
S N0 ( f ) = H ( f ) S N ( f )
5.2. Optimum Filter
H ( f ) = e− j 2fT S* ( f ) = e− j 2fT S (− f )
h(t ) = S (T − t )
T = Sampling instance
= Duration of incoming pulse

Unit impulse response of optimum filter

Optimum filter = matched filter  Maximizes signal power at sampling instances.
Properties of MF
S(t) is an energy pulse of duration T.
1. h(t ) = S (T − t )
2. S (t ) = h(T − t )
3. S0 (t ) = S (T ) * h(t )
4. S (t ), h(t ), S0 (t ) are energy signal
5. ES (t ) = Eh (t ) = S0 (t ) max
Es (t )
6. ( SNR)max at t = T =
( N0 / 2)
7. ES1 (t )  Es2 (t ) then Pe1  Pe2
S0 ( f ) = S ( f ) e− j 2fT
2
8.
2
S (t )  S ( f )  S ( f ) = Gs ( f )
F.T .
Rs () ⎯⎯ →Gs ( f )
ACF  S (t ) ⎯⎯
F .T .
→ PSD S (t )
2
Rs () ⎯⎯
F .T .
→ S( f )
→ S ( f ) e− j 2fT = S0 ( f )
2
S0 () = Rs ( − T ) ⎯⎯
F .T .
Matched Filter Output

 a (t ) + N0 (t ) :1 TX
y(t ) = S0 (t ) + N0 (t ) =  1
a2 (t ) + N0 (t ) : 0 TX
Channel Noise is White

1
Let P(0 TX ) = P(1 TX ) =
2
a − a   x
[ Pe]min = Q  1 2  = Q   x  Q   
x
 2 y  2  2
2
Maximization of x


2
S1 ( f ) − S2 ( f ) df
= −
2
x max
( N0 / 2)

* 
H ( f ) =  S1( f ) − S2 ( f )e j 2fT  , Ed = 
2
When S1( f ) − S2 ( f ) df
−
2Ed
xmax =
N0
2Ed
h(t ) = S1 (T − t ) − S2 (T − t ) → x : xmax =
N0
a −a 
Pe |min = Q  1 2 
 2 y 
 x x 
Pe |min = Q   M .F Pe |min min = Q  max 
 2  2 
h(t )
 Ed 
Pe |min min = Q  
 2 N0 
➢ Only when, P(0 TX ) = P(1 TX ) = 1/ 2
➢ AWGN,  →opt
➢ M.F
For K noise R.V



(a − a ) 
Pe |min = Q  K 1 2 
 2 

 Ed 
Pe |min min = Q K 
 2 N0 
1.

Ed =  d 2 (t )dt = 4 A2Tb
−
d (t ) = S1 (t ) − S2 (t ) = 2 A
0  t  Tb

 2 A2T 
Pe = Q  b
 N0 
 2( Eb )avg 
Pe = Q  
 N0 
( Eb )avg = A2Tb
Eb = A2Tb
2.
 E 
Pe = Q  b 
 N0 
Eb
( Eb )avg =
2
 A2T 
Pe = Q  b
 2 N0 
M-ary Base Bond Signaling

1. Bit rate : Rb
2. Bit interval = Tb = 1/ Rb
3. Symbol duration = Ts = NTb
1 1 R
4. Symbol rate or Baud or Baud rate Rs = = = b
Ts NTb N
Rs
5. TX Bandwidth ( BW )  Rs → Rectangular, ( BW )  → sin C
2
1 R  R
(BW )min =  b  = s
2 N  2
M-ary PAM (2-Any PAM)

 S1 (t ) = A 0  t  Tb
1. M = 2, ( N  M ) , s(t ) =  NRZ coding.
S2 (t ) = A 0  t  Tb

S2 (t ) S1 (t )
(t )
( − A Tb ,0) 0 ( A Tb ,0)
d12 = Ed
 Ed 
Pe |min min = Q 
 2 N0 
Ed = (d12 )2
 2 A2T 
Pe = Q  b
for NRZ
 N0 
( Es )avg = A2Tb
➢ Distance of each point from origin = Energy of that point
➢ Distance between two point = Difference energy = d12
➢ d12 → Q( ) → Pe 
 A2T 
Pe = Q  b
for RZ
 2 N0 
Bandpass Sampling:
(a) Binary ASK : (m-ary ASK), For ‘1’→ A, ‘0’, →0
 Ed   A2T 
Pe = Q   = Q  b
 2 N 
0  4 N 
0 
1  A2Tb 
( Eb )avg = p1E1 + p2 E2 =   
2  2 
 ( Eb )avg 
Pe = Q  
 N0 
 A2T cos2  
Pe = Q  b 
 4 N0 
Correlator Based :
2
(t ) = cos 2fC t
Ts
0  t  Ts
"1" = A cos C t 0  t  Tb
"0" = 0 0  t  Tb
 A 0  t  Tb : 1TX
(b) BPSK : p(t ) =  → Baseband
− A 0  t  Tb : 0 TX
 A cos 2fC t 0  t  Tb : 1TX

s(t ) = 
 A cos(2fC t + ) 0  t  Tb : 0 TX
 A2T 
Pe = Q  b
 N0 
 
Orthonormal Basis Function

2
(t ) = cos 2fct 0  t  Tb
Tb
 A2T cos2  
Pe = Q  b 
 N0 
 
 d2 
Pe = Q  12 
 2 N0 
 
Tb
d12 = 2 A
2
S 0 (t ) S1 ( t) 2
(t) = cos c t
O Tb
T Tb
A b A
2 2
M-Ary PSK (Quadrature PSK):

 2 
M =4 Sk (t ) = Acos  2fct + K  0  t  Ts ,(Ts = NTb )
 M 

N = 2 Sk (t ) = A cos  2fct + K  Ts = N Tb
 2 
K = 0,1,2,3  2
dmin = 2d0 sin   d0 = Es ,  =
 2 M

d12 = 2d0 sin  
 2
M-ary PSK
M = (2N )
1. Bit Interval = Tb , Bit rate = Rb , symbol duration (Ts = NTb )
1 Rb
2. Baud rate or symbol rate Rs = =
Ts N

A2
3. Bit energy  Eb = Tb
2
4. Symbol energy Es = NEb
5. Radius of constellation : d0 = Es
6. Area of constellation circle = d02 = Es
  2 
7. dmin = 2d0 sin   ,   = 
 2  M
Binary FSK
 A cos 2f1t : 0  t  Tb :1TX
SFSK (t ) =  ( f1  f 2 )
 A cos 2f 2t : 0  t  Tb : 0TX
BFSK
Same phase Phase shift keying
Coherent BFSK Non Coherent 8FSK

 S 1 (t ) = A cos( 1t + 1 ); 0  t  Tb
S (t ) = 
 S 2 (t ) = A cos ( 2t + 2 ); 0  t  Tb
 S1 ( t) = A cos 1t : 0  t  Tb  S 1 (t ) = A cos( 1t + 1 ) 0  t  Tb
S (t ) =  S (t ) = 
 S 2 ( t) = A cos 2t : 0  t  Tb  S 2 (t ) = A cos( 2t + 2 ) 0  t  Tb
Coherent BFSK
1.  = 0, Rb = HCF[mRb , nRb ] = HCF[2( f1 + f 2 ),2( f1 − f 2 )]
2. (  0) Rb = HCF[mRb , nRb ] = HCF[( f1 + f 2 ),2( f1 − f 2 )]

3. Non-Coherent – Rb = HCF[mRb , nRb ] = HCF[( f1 + f 2 ),( f1 − f 2 )]
Condition for Orthogonality

mRb nR
d = 0( f1 + f 2 ) = , ( f1 − f2 ) = b
2 2
Coherent FSK
d  0( f1 + f2 ) = mRb ,( f1 − f 2 ) = nRb
Rb = HCF[mRb , nRb ]
Non-Coherent FSK
1 , 2
( f1 + f 2 ) = mRb ,( f1 − f 2 ) = nRb
Rb = HCF(mRb , nRb )
1
➢ P(OTX ) = P(1TX ) =
2
➢ Channel Noise : White (AWGN)
➢ opt
f1 = k | Tb
➢
Filter Method
f2 = m | Tb
 Ed   A2T 
 = Q
b
Pe = Q 
 2 N0   2 N0 
 
Orthogonal FSK :
 0.5 A2T 
Pe = Q  b
 N0 
 
Non-orthogonal FSK :
 0.6 A2T 
Pe = Q  b
 N0 
 
 (d )2   2 
Pe = Q  12  = Q  A Tb 
 2 N0   2 N0 
   
M-ary FSK
N bits are grouped together so that M = 2 N symbols or sinusoids of duration Ts = NTb are generated having
➢ Same amplitude, same frequency, different frequency.
n  A2 
fk = Es0 = Es1 = = EsM −1 =   Ts 
Ts  2 
 
1 1 R
Ts = NTb Rs = = = b
Ts NTb N
Scheme Pe For K
 A2T   KA2T 
BASK ⎯⎯
→ Pe = Q  b
 Pe = Q  b 
 4 N0   4 N0 
   
 A2T   KA2T 
BPSK ⎯⎯
→ Pe = Q  b
 Pe = Q  b 
 N0   N0 
   
 A2T   KA2T 
BFSK ⎯⎯
→ Pe = Q  b
 Pe = Q  b 
 2 N0   2 N0 
   
  − 2    1 − 2  
Pe = Q  1  Q K  
 2 N0    2 N0  
For K AWGN identical independent.
Amplitude phase shift keying (APSK)

Si (t ) = ri cos[2fct + i ] (0  t  Ts ) i = 0 to M −1
Ts = NTb
Case 1: ri = constant
i = variable
Case 2: ri = variable
i = constant
 Si (t ) = ri cos(2fct + i ) → M-Ary PSK


Si (t ) = ri cos(2fct + i ) → M-Ary ASK
Fig. ASK Fig. PSK
8 point APSK = 8 point QAM


6 INFORMATION
THEORY
6.1. Introduction
Information in Event ( X = xi ) Base Unit
I [ X = xi ] = − logb p{X = xi } 2 Bits
10 Decit
e Nat
6.1.1. Properties of Digital Information
1. I [ X = xi ] = − log2 P[ X = xi ]
2. P[ X = xi ]  P[ X = x2 ]  I [ X = x2 ]  I [ X = x2 ]
3. P[ X = 1] = log2 1 = 0 bits
4. P [ X = 0] = − log 2 0 =  bits
5. 0  P [ X = xi ]  1  0 bits  I [ X = xi ]   bits
6. For any event [ X = xi ], I [ X = xi ]  0
7. I ( X = xi )  ( X = x2 ) = I [ X = x1] + I[ X = x2 ]
Average information of source X = Entropy of source X
M
H [ X ] = − P [ X = xi ]log 2 P [ X = xi ] bits/symbol
i =1
M
H [ X ] = − Pi log 2 Pi
i =1
Case 1. All M events are equiprobable –

H [ X ] = log 2 M bits/symbol  Maximum entropy
Case 2. Out of M events only 1 event is certain.
H[ X ] = 0
1
0  H [ X ]  log 2 M M=
P

Information Rate – Symbol rate = r symbols/sec

Entropy = H(X) bits/symbol
Information Rate R = r H (X) bits/sec
1. If r = f s and all event equiprobable, L = 2n , H ( X ) = log2 L
R = nf s
Source Coding
1. Reduces the redundancy of bits.
2. Two types of source coding
(a) Fixed length source coding
(b) Variable length source coding
(i) Shannon Fano coding
(ii) Huffman coding
Key Point :
K
(a) Average code length = Lavg =  ni pi
i =1
(b) Entropy of source = H ( X )
H(X )
(c) Code efficiency  =
Lavg
 should be as high as possible.

(d) Code redundancy  = (1 – )
Discrete channel : A channel is called as discrete if X and Y are having finite size.
Memoryless channel : Each present output symbol depends on present input symbol.
x y 
x =  1 → P( y j / xi ) → 1  = y
 x2 y2 
Binary Channel : (2 input & 2 output)

Non Binary Channel :
Binary Channel :
*Sum of elements of row in channel matrix is always ‘1’.

Joint Channel Matrix
y1 ym
x1  P ( x1  y1 ) P ( x1  ym ) 
P (x  y ) P ( x2  ym ) 
[ P ( x; y)] =  2 1
 
 
xn  P ( xn  y1 ) P ( xn  ym )  nm
Y 
[ P ( x  y)] = P( X ) P  
X
Condition Channel Matrix
  y1   y 
P  P  m 
x1   x1   x1  
 y  
[P   =  
 x  
xn   y1  y 

P  m 
  xn 
P
  xn   nm
Y 
[ P ( y)] = P( X ) P  
X
P( y1 ) = P ( x1  y1 ) + P ( x2  y1 ) + + P ( xn  y1 )
[P( y)]1m =  P( x1), P( x2 ) P( xn )1n
  Y 
[ P( y)]1m = [ P( x)]1n  P  
  X  nm

  y1   y 
P  P  m 
  x1   x1  
 
 
  y1   ym  
P   P 
  xn   xn   nm
Binary Non-symmetrical channel
Cross over probabilities are different

y1 y2
Y  1 − q q  x1
P  =  p 1 − p x
X    2
  y1   y2  
 P   P  
 x1   x1  
=  
 P  y1  P  y2  
  x2   x2   22
• P( x1 ) + P( x2 ) = 1
• P( y1 ) + P( y2 ) = 1
y  y 
• P 1  + P 2  =1
 x1   x2 
y  y 
• P 1  + P 2  =1
 x2   x2 
• P( y1 ) = P( x1  y1 ) + P( x2  y1 )
• P( y2 ) = P( x1  y2 ) + P( x2  y2 )
Aposteriori Probabilities
y 
P( x1 ) P  1 
x   x1 
P 1  =
 y1  P( y1 )

y 
P( x2 ) P  2 
x   x2 
P 2  =
 y2  P( y2 )
Map Analysis
(a) At r0 :
m0
 m0   m 
P   P 1 
 0  m1  r0 
r
m0
 r0   r 
P( m0 ) P    P( m1 ) P  0 
 m0  m1  m1 
(b) At r1:
m0
 m0   m 
P   P 1 
 r1  m1  r1 
m0
 r1    r 
P( m0 ) P    P( m1 ) P  1 
 m0  m1  m1 
➢ After MAP application receiver will be optimum.
Probability of Correctness
Pc = P(m0  r0 ) + P(m0  r1 )
r   r 
Pc = P(m0 ) P  0  + P(m0 ) P  1 
 m0   m0 
Pe = 1 − Pc
Binary Symmetrical Channel

Cross over probabilities are same.
x 
➢ P  1  = Probability that x1 was transmitted given than y2 received
 y2 

x  x 
P 1  + P 2  =1
 y2   y2 
x  x 
P 1  + P 2  =1
 y1   y1 
Joint Entropy
n m
H ( XY ) = − P( xi , y j )log 2 P( xi , y j )
i =1 j =1
n m
  
H ( XY ) = − P ( X = xi )  (Y = y j ) log 2 P ( X = xi )  (Y = y j ) 
i =1 j =1
H ( XY ) = H (YX )
Conditional Entropy
X  n m  X = xi 
H
Y


= − 
i =1 j =1
P{( X = xi )  (Y = y j )}log 2 P  
 Y = y j 
Conditional entropy of
X given Y
Y 
• Similarly can write H  
X
Important point :
Y 
1. H ( XY ) = H ( X ) + H  
X
X
2. H ( XY ) = H (Y ) + H  
Y 
3. If X and Y statically, independent H ( XY ) = H ( X ) + H (Y )
Y  X
H   = H (Y ), H   = H ( X )
X Y 
For B.S.C – Cs = Channel capacity
Cs = 1 + P log2 P + (1 − P)log2 (1 − P)
Cs = {I ( X ;Y )}max
Y 
I ( X ;Y ) = H (Y ) − H  
X
X
I ( X ;Y ) = H ( X ) − H  
Y 

Y   yi 
H  = − P ( xi , y j )log2 P  x 
X   i
Cs = log 2 n
I ( X ;Y ) = H ( X ) loss less channel
Lossless Channel :
1. Single non zero element in each column.
2. Channel matric should be DMC type
3. Summation of each now must be 1.
X  I ( X ;Y ) = H ( X ), Cs = I [ X ;Y ]max = [ H ( X )]max = log2 n
4. H =0
Y 
n = number of input symbol.
6.2. Average Mutual Information
X
I ( X ;Y ) = I ( X ) − I  
Y 
X
I ( X ;Y ) Avg = H ( X ) − H   bit/symbol
Y 
Y 
I ( X ;Y ) Avg = H (Y ) − H  
X
I ( X ;Y ) = I (Y ; X )
n m (
 P xi | y j
I ( X ;Y ) =  P{xi , y j }log 2 
) 
i =1 j =1  P( xi ) 
 
  x 
   f X  
 y 
I ( X ;Y ) =   f XY ( x, y)log 2  Y  dxdy
f ( x )
− −  X 
 
x 
PXY ( xi , y j ) = PX ( y j ) P  i 
 yj 
 
 yj 
PXY ( xi , y j ) = P( xi ) P  
 xi 
If R.VS are Independent then I ( x; y) = 0

6.2.1. Channel Capacity

Maximum Average Mutual Information
Cs = {I ( x; y)}max
I ( x; y) = H (Y ) + P log 2 P + (1 − P)log 2 (1 − P)
Cs = 1 + P log2 P + (1 − P)log 2 (1 − P)
B.S.C
P → Cross over probability
➢ Input are equiprobable.
Determine Channel :
1. Number of rows in each row must be single.
• In each row angle element must be 1.
• Summation of each row will become 1.
Y  Y 
• H   = 0, I ( X ;Y ) = H (Y ) − H  
X X
I ( X ;Y ) = H (Y )
Cs = [ H (Y )]max = log2 m bit/symbol
Noise Less Channel :

1. Deterministic + Lossless
Each row → Single element
Each column → Single element
X Y 
4. H   = 0, H   = 0
Y  X
5. I ( X ;Y ) = H ( X ) = H (Y )
6. Cs = [ I ( X ;Y )]max = [ H ( X )]max = [ H ( y)]max = log 2 m = log 2 n bits/symbol
m=n
Binary Erasure Channel

I ( X ;Y ) = (1 − P) H ( X )
Cs = I [( X ;Y )]max
= (1 − P)log2 n n = 2 for BEC
Cs = (1 − P)
1   2 
➢ Cs = I ( X ;Y ) = log 2 1 +  2x  Bits/symbol
2   N 
2N = N0 B
(i) X is zero mean R.V E[ X 2 ] = 2x = S
(ii) Noise is zero mean R.V E[ N 2 ] = 2N = N
Channel Capacity of AWGN Channel
1  S
Cs = log2 1 +  Bit/symbol
2  N
 S
C = B log2 1 +  bit/sec
 N
Channel capacity for AWGN
Cs  R
Channel Information
Capacity rate
• For X = zero mean R.V, N = N0 B

B →
S
Cs = 1.44 Finite value
N0
For loss less transmission.
6.3. Continuous Source and Differential Entropy

X : DRV, H(X) = − p[ x = xi ]log2 p{x = xi }
i
+
X : CRV, H(X) = − Fx( x)log2 f x ( x)dx → Differential entropy
−
Y : DRV, H(Y) = −  p{y = y j }log2 p{y = y j }

j
−
Y : CRV, H(Y) = −  f y ( y)log2 f y ( y)dy → Differential entropy
−

6.3.1. Channel capacity

(1) For error less | distortion less transmission
(i) If all quantization level are not eauiprobable :
CR
C  rH ( X )
C  fS H ( X )
(ii) If all quantization level are eauiprobable :
CR
C  rH ( X )
C  f s log 2 L
C  nf s
C  Rb
(2) For AWGN channel- Y: GRV,
Y = X + N, Let X and N are independent
2y = 2x + 2N
1
H ( y) = log2 [22y e]
2
1
H (y) = log2[2 e[2N + 2X ] Maximum
2
PX = Power of signal X
1   2  bits
CS = log 2  1 + X2 
2  N  symbol
P X = Noise Power
C = C2  f s
2N = N0 B
 2  Bits N0
C = B log 2 1 + 2X  → PSDof white Noire
 
 N  sec 2
B → B.Wof channel
C = B log 2 (1 + SNR)
 PX 2X
 NR = =
Not in dB PN 2N
 ER 
C = B log2 1 + b b 
 N0 B 
PX
C = 1.44
N0
For infinite Bandwidth B → 
Information in bits | symbol

y y  y  y  y 
H   = I  0  P  0  P( x0 ) + I  1  P  1  P( x0 )
 x0   x0   x0   x0   x0 


7 MISCELLANEOUS
7.1. FDMA (Frequency Division Multiplexing)

➢ Multiple signals are multiplexed and simultaneously transmitted through channel.
K = Number of signals are multiplexed
B.W  K [B.W of modulation scheme] + (K – 1) [BW of guard Band]
TDMA (Time division Multiplexing)
Ts = Frame rate or sampling interval or time taken by commentator to completer its 1 rotation
(Band limited to same freq.)
Ts = nTb  N
Ts = NnTb N = Number of signals being multiplexed
n = of bits/sample
Tb = 1 Bit duration
Rb = Nnf s
rotation
Speed of commentator = f s = f s  60rpm
seocnd
Rb Nnf s
(BW)min = =
2 2
bits
➢ When x number of synchronization are added – (Band limited to same freq.)
frame
Ts = ( Nn + x)Tb Rb = ( Nn + x) f s
➢ x bit/frame :Ts = ( Nn + x)Tb
 x
x bit/2frame :Ts =  Nn + Tb
 2
➢ y% (Total of y%) synchronization bits are added –(Band limited to same freq.)
 Nn  y% 
Ts =  Nn + Tb
 100 

 Nn  y% 
Rb =  Nn + fs
 100 
➢ When N signals are band limited to different freq.
Rb = nf s1 + nf s2 + ......nf sn
COMA (Code division Multiple Aces)
R 
Processing gain of CDMA  G =  c 
 Rh 
➢ Each user is assigned with unique code
Noise
(1) PSD of thermal noise is Gaussian in nature. Also known as Johnson noise
(2) Thermal Noise power Pn = 4KTBR = Vn2 = (Vn )2 rms
Thermal Noise voltage
(Vn )rms = 4 KTBR
(Vn )rms
( I n )rms =
R
➢ Max. Power which could be deliver to amplifier = KTB
➢ Noise figure (F) or Noise factor
F (dB) = 10log10 F
N1 = kTB
N o = Ni G + N a = KTBG + Na
( Ni G + N a )
F=
GNi
Output Noiseincluding Noisyamplifier

F=
Output NoiseConsideringnoiselessamplifier
Na (SNR)i /p Te
Te = F= =1+
KBG (SNR)o/ p T
Te = ( f − 1)T N0 (output Noise power) = KTBGF
N0 = K BG(T + Te )

Cascaded Amplifier
Output Noise with noisy ampr = [KTB(F2 −1) + KTBF1G1]G2
F2 − 1
F = F1 +
G1
F2 − 1 (F3 − 1)
F = F1 + + + ..........
G1 G1G2
Te2 Te3
Te (equivalent Temp.) = Te1 + + + ..........
G1 G1G2
xdBW = ( x + 30) dBm
Noise performance of Analog Signal

(SNR)0 SNR at theoutput of RX
FOM = = =
(SNR)i (SNR)in m
Pm P
For DSB-Se  = 1 , (SNR)0 = ,(SNR)i = m
2N0 B 2N0 B
pm
For DCB-FC  = =  → efficiency
Ac2 + pm
2
3 K f pm 3 2
For F.M  = 2 = FM (For sinusoidal)
4 B2 2
2PM
For PM  = K 2p Pm = (for sinusoidal)
2
Channel Coding

Linear Block Code

 modulo 2 sum → EXOR
(1)
(2)
(3) Different data words (message word) with K bits = 2K
(4) Each data word will have m parity bits attached to generate 2K code words.
(5) Total no of arrangements with n bits at output of encoder will be →2n out of which only 2K code words are valid.
(6) Rate efficiency = code efficiency = code rate = K / n
Hamming Weight
Number of 1’s present in L, B,C
C (7.4)
Example: C : 1110001, H.W = 4
Hamming distance
It represent bit change at respective position
X = 11 0 1 0 111
  
Y = 0 111 0 1 0 1 d ( x, y ) = 3
Minimum Hamming Distance (dmin):

Method 1 dmin = Min hamming weight of 2K codes except codes having 0 weight.
Combination: 2KC2crosscheck
Method 2 d min  n − K + 1
Method 3 “Minimum no of columns in parity check matrix [H] Which makes zero sum (module 2).”
Error detection L.B.C
dmin  t + 1 can detect t errors
Error correction dmin  2t + 1
Code Generation at TX − [C]1n = [ D]1K [G]Kn C = DG

[G]K n → Generator Matrix
[G]Kn = [ I K ; p]Kn or[ p; I K ]Kn I k = Identity Matrix of order K.
Parity check Matrix – [ H ] = [ PT ; I n − K ](n− K )n

Or
[ H ] = [ I n− K ; PT ](n− K )n
Note – [C] [HT] = 0
Correction at Receiver
c →r
r =C (No error)
r  C (Error)
➢ r will given
➢ Calculate syndrome : S = r[ H ]T
➢ Observe the syndrome: S matches with ith row of [ H ]T which Means ith bit from left has error.
Non systematic L.B.C
Hamming Code
(1) It is a L.B.C
(2) dmin = 3
(3) Detect upto 2 bit error
(4) Correct upto 1 bit error
(5) K bit data, m bits parity  n = (m + K ) bits code
(6) Parity bit no. is calculated 2m  (m + K + 1)
m=?
0 1 2
(7) Placing of parity bits ate at 2 ,2 ,2 ,........ locations
Cyclic redundancy check code (CRC-Code)
Problem solving Technique:

(i) d = K bits msg
(ii) divisor polynomial
x3 + x + 1 = x3 + 0 x 2 + x + 1 → (1011)
Step 1. K msg bits are given
From (K + m) message bits
 m → addition of m zeros(append)
➢ Highest order of divisor polynomial (or) (number of bits in divisor polynomial)-1
Step 2. Modulo 2 division

Library:- https://smart.link/sdfez8ejd80if
PW Mobile APP:- https://smart.link/7wwosivoicgd4

Communication PW

Uploaded by

Copyright:

Available Formats

Communication PW

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Communication PW

Uploaded by

Copyright:

Available Formats

Communication

Mobile App: Physics Wallah (Available on Play Store)

1. Amplitude Modulation ........................................................................................................ 8.1 – 8.10

2. Angle Modulation ................................................................................................................ 8.11 – 8.18

3. Random Variable and Random Process ............................................................................... 8.19 – 8.45

4. Digital Communication ........................................................................................................ 8.46 – 8.55

5. Digital Receiver .................................................................................................................... 8.56 – 8.70

6. Information Theory ............................................................................................................. 8.71 – 8.81

7. Miscellaneous ..................................................................................................................... 8.82 – 8.86

GATE-O-PEDIA ELECTRONICS & COMMUNICATION ENGINEERING

S AM (t ) = Ac cos ct + m (t )cos ct  S AM (t ) = [ Ac + m (t )]cos ct carrier after modulation

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.1

  1, A(t )  0, E(t ) = A(t )

Frequency Related Parameters

For single tone sinusoidal signal

= Ac [1 +  cos 2f mt ]cos 2fct

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.3

(3) PAM → Will be constant if Pc  and  

If m (t ) is multiple single tone signal-

m(t) → fm1 , fm2 → fmax = fm2

fm2  fm1 BW = fm2 − fm1

Power Related Parameters

Pm Ac2  12 22 

T = 12 + 22 + 32 + − − − −

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.4

DSB- FC [AM] Modulator

➢ Only (2) and (6) are desirable

Z (t ) = Ac' [1 + ka m (t )]cos2fct only when fc  3 f m fc  (2 + 1) f m

(2) Switching Modulator:

A2 + B2 cos ct → E.D → A2 + B2

(1) x (t ) = A cos 0t + B sin 0t → E(t ) = A2 + B2

(2) x (t ) = A cos(0t + ) + B sin 0t → E(t ) = A2 + B2 − 2 AB sin 

➢ Proper choice of discharging time constant RLC -

No fluctuation No diagonal clipping

(7) m(t): Multitone f m → f max = Max. freq. component of m(t)

1.2. Synchronous Detector

➢ . = 2  max. freq. component of m (t )

Ac2  Am1 Am2 

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.7

SSB- SC (Single sideband suppressed carrier)

(a) Phase Discrimination:

Spectral gap in D.S.B BPF Signal

 0Hz Ideal SSB -SC

➢ SSB- SC can be demodulated by Synchronous detection.

(1)  = 0,   0, m (t ) recovery not possible → freq. synchronization

(2)   0  = 0 , m (t ) recovery not possible → Phase synchronization

(3) Perfect syne,  = 0,  = 0 can be recovered

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.9

Percentage Power Saved

(2) % power saved in SSB- SC as compare to DSB-FC-

Modulation B.W Power Application

Pre envelope and Complex Envelope

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.10

Signal = x(t) |x (t)|max

2.1.1. Instantaneous Angle and Instantaneous frequency-

i (t ) → Instantaneous angle (rad)

GATE WALLAH ELECTRONICS & COMMUNICATION HANDBOOK 8.11

2. Instantaneous frequency deviation (t ) = K f m(t ) f (t ) = K f m(t ) Hz

4. Frequency deviation in –ve direction [(t)]min = K f [m(t)]min [f (t)]min = K f [m(t)]min