Digital Communication

Dr. A. Ananth
 IEC 222 Digital
Communication [3-0-3-4]
Course Objectives:
• Understand the mathematical representation of signal, symbol, and noise

• Understand the concept of signal processing of digital data and signal

conversion to symbols at the transmitter and receiver

• Compute performance metrics and parameters for symbol processing and

recovery in ideal and corrupted channel conditions

• Compute performance parameters and mitigate channel induced 2

impediments in corrupted channel conditions.
 IEC 222 Digital
Communication [3-0-3-4]
Course Outcomes:

• Associate and apply the concepts of Bandpass sampling to well specified signals and
channels

• Analyze and compute performance parameters and transfer rates for low pass and
bandpass symbol under ideal and corrupted non band limited channels

• Test and validate symbol processing and performance parameters at the receiver under
ideal and corrupted band-limited channels

• Demonstrate that bandpass signals subjected to corruption and distortion in a band-

limited channel can be processed at the receiver to meet specified performance criteria.
3
Topics to be Covered
Bandpass Signal to Equivalent Low pass: Hilbert Transform, Pre-envelopes, Complex envelopes,
Canonical representation of bandpass signals, Complex low pass representation of bandpass systems,
Complex representation of band pass signals and systems.

Unipolar, Polar, Bipolar (AMI) and Manchester code and their power spectral densities. Overview of
HDB3, B3ZS, B6ZS.

Signaling over AWGN Channels- Introduction, Geometric representation of signals, GramSchmidt

Orthogonalization procedure, Conversion of the continuous AWGN channel into a vector channel,
Optimum receivers using coherent detection: ML Decoding, Correlation receiver, matched filter
receiver.

Digital Modulation Techniques: Phase shift Keying techniques using coherent detection: generation,
4
detection and error probabilities of BPSK and QPSK, M–ary PSK, M–ary QAM.
Topics to be Covered
Frequency shift keying techniques using Coherent detection: BFSK generation, detection and error
probability. Non coherent orthogonal modulation techniques: BFSK, DPSK Symbol representation, Block
diagrams treatment of Transmitter and Receiver, Probability of error.

Communication through Band Limited Channels: Digital Transmission through Band limited channels:
Digital PAM Transmission through Band limited Channels, Signal design for Band limited Channels:
Design of band limited signals for zero ISI–The Nyquist Criterion, Design of band limited signals with
controlled ISI-Partial Response signals, Probability of error for detection of Digital PAM: Probability of
error for detection of Digital PAM with Zero ISI, Symbol–by– Symbol detection of data with controlled
ISI.

Principles of Spread Spectrum: Spread Spectrum Communication Systems: Model of a Spread Spectrum
Digital Communication System, Direct Sequence Spread Spectrum Systems, Effect of Despreading on a
narrowband Interference, Probability of error (statement only), Some applications of DS Spread Spectrum
Signals, Generation of PN Sequences, Frequency Hopped Spread Spectrum, CDMA based on IS-95.5
 IEC 222 Digital
Communication [3-0-3-4]
Text Books/ References:

1. Simon Haykin, “Digital Communication Systems”, John Wiley & sons, First
Edition, 2014, ISBN 978-0-471-64735-5.
2. John G Proakis and MasoudSalehi, “Fundamentals of Communication
Systems”, 2014 Edition, Pearson Education, ISBN 978-8- 131-70573-5.
3. B. P. Lathi and Zhi Ding, “Modern Digital and Analog communication Systems”,
Oxford University Press, 4th Edition, 2010, ISBN: 978- 0-198-07380-2.
4. Ian A Glover and Peter M Grant, “Digital Communications”, Pearson Education,
Third Edition, 2010,ISBN 978-0-273-71830-7.
5. Bernard Sklar and Ray, "Digital Communications - Fundamentals and
Applications", Pearson Education, Third Edition, 2014, ISBN: 978-81-317-2092-
6
9.
 Today’s Topic of Discussion
Communication Systems

Analog Communication

Digital Communication

Source Coding

Line Coding

Error control coding/ Channel coding 7

Historical Background
• Contributors to Digital Communication

• Information Theory and Coding

• Wireless Communication

• Internet

8
Block Diagram of Digital Communication

9
Line Coding
• Unipolar Non return to zero (NRZ)

• Polar NRZ

• Unipolar return to zero (RZ)

• Bipolar RZ

• Split phase or Manchester code

10
Amplitude Shift Keying
• Definition

• BASK Block Diagram

• Signal Representation

• Constellation Diagram

• Coherent Receiver

11
Phase Shift Keying
• Definition

• BPSK Block Diagram

• Signal Representation

• Constellation Diagram

• Coherent Receiver

12
Phase Shift Keying
Generation of BPSK Signals

Signal Representation

13
Phase Shift Keying
Coherent Detection of BPSK Signals

14
Phase Shift Keying
Signal Space/ Constellation Diagram

15
Frequency Shift Keying
• Definition

• BFSK Block Diagram

• Signal Representation

• Constellation Diagram

• Coherent Receiver

16
Phase Shift Keying
Generation of BFSK Signals

Signal Representation

17
Phase Shift Keying
Coherent Detection of BFSK Signals

18
Phase Shift Keying
Signal Space/ Constellation Diagram

19
BASK, BFSK, BPSK Waveforms

20
Differential Phase Shift Keying
DPSK Transmitter

21
Differential Phase Shift Keying
DPSK Transmitter

• Non coherent version of BPSK

• Eliminates the need for synchronization

DPSK Receiver

22
Quadrature Phase Shift Keying
(QPSK)
QPSK Transmitter

23
Quadrature Phase Shift Keying
(QPSK)
QPSK Receiver

24
Quadrature Phase Shift Keying
(QPSK)
QPSK Signal Representation

Four phases: /4, 3 /4, 5/4, 7/4

E  energy per symbol
T  symbol duration
Gray coding: 00, 01, 11, 10
25
Quadrature Phase Shift Keying
(QPSK)
QPSK Signal Space Diagram

26
Quadrature Phase Shift Keying
(QPSK)
QPSK Signal Space Diagram

27
Quadrature Phase Shift Keying
(QPSK)
QPSK Waveform

28
Quadrature Phase Shift Keying
(QPSK)
Offset QPSK
• Possible phase switching in QPSK  180o  two bits are changes (00  11, 01
 10)

• 90o phase shift  one bit changes (00  10, 01)

• 0o phase shift  no change in successive bits

29
Quadrature Phase Shift Keying
(QPSK)
Offset QPSK
• 180o and 90o phase shift are a big concern while filtering at the receiver

• They result in change in the amplitude of the carrier signal while signal
transmission

• Change in the amplitude causes symbol errors at the receiver

• To restrict the amplitude variations/ limit the amplitude fluctuations 

offset QPSK

30
Quadrature Phase Shift Keying
(QPSK)
Offset QPSK
• The bit stream sent to the quadrature component is delayed by half of the
symbol interval

• The phase change is restricted to 90 o in OQPSK

• Amplitude fluctuations will be smaller in OQPSK than QPSK

31
Quadrature Phase Shift Keying
(QPSK)
Offset QPSK

https://www.everythingrf.com/community/what-is-oqpsk-modulation
32
M- ary Phase Shift Keying
(MPSK)
• Generic form of PSK  M-ary PSK

• Phase of the carrier signal takes M distinct values

• Modulated signal

• Uses two orthogonal basis function, signal constellation is 2-D 33

Quadrature Phase Shift Keying
(QPSK)
M-ary PSK Signal Space Diagram

34
Quadrature Phase Shift Keying
(QPSK)
M-ary PSK

• M = 8, octa phase shift keying

• Signal space diagram is circularly symmetric

• Euclidean distance in 8 PSK

35
Quadrature Amplitude
Modulation (QAM)
M-ary QAM
• M-ary PSK, amplitude is constant

• QAM  hybrid modulation scheme  amplitude and phase modulation

• M-ary PAM  one dimensional

• M-ary QAM  two dimensional realization of M-ary PAM

36
Quadrature Amplitude
Modulation (QAM)
M-ary QAM
• Transmitted signal

• Modulated signal has two-phase quadrature carriers multiplied with

discrete amplitude  QAM

• Square constellation  even number of bits per symbol

37
Quadrature Amplitude
Modulation (QAM)
Square QAM Constellations
• Even number of bits per symbol

• Cartesian product of one dimensional L-PAM with itself

• Ordered pair of coordinates

38
Quadrature Amplitude
Modulation (QAM)
Construction of 16 - QAM Constellation from PAM
• Construct two 4 – ary PAM vertically (2) and horizontally (1) which are
orthogonal to each other

39
Quadrature Amplitude
Modulation (QAM)
Construction of 16 - QAM Constellation from PAM
• First quadrant constellation  positive part of 1 & 2

Second quadrant constellation  positive part of 2 & negative part of 1

40
Quadrature Amplitude
Modulation (QAM)
Construction of 16 - QAM Constellation from PAM
• Third quadrant constellation  negative part of 2 & positive part of 1

• Fourth quadrant constellation  & negative part of 2 & 1

41
Quadrature Amplitude
Modulation (QAM)
Construction of 16 - QAM Constellation from PAM
• Gray encoding principle

42
Power Spectral Density of Line
Codes
• PSD of Line codes
Unipolar NRZ Line Code Polar NRZ Line Code

43
Power Spectral Density of Line
Codes
• PSD of Line codes
Bipolar RZ Line Code Unipolar RZ Line Code

44
Signalling over AWGN Channels
AWGN Channel Model
• Output from a computer 0s and 1s for every Tb seconds

• Transmitter converts 1s and 0s by distinct signals s1(t) and s2(t)

• Signals s1(t) and s2(t) are real valued energy signals, send through analog
channel

• Analog channel is AWGN channel

45
Signalling over AWGN Channels
AWGN Channel Model
• Analog channel is AWGN channel

• Receiver should observe the signal over Tb interval and estimate the
transmitted signal si(t), i = 1, 2

• Due to noise, the estimation might be in error

46
Signalling over AWGN Channels
AWGN Channel Model
• Average probability of symbol error is

• 1, 2 is the probability of transmitting 1 and 2

• ,, is the estimate of the symbol 0 and 1

• Conditional probabilities

47
Signalling over AWGN Channels
AWGN Channel Model
• How to optimize the design of receiver to minimize the probability of error

• How to set the signals s1(t), s2(t), …, sM(t) to represent the symbols m1, m2, …,
mM

• Less probability of error  high reliability

• More probability of error  low reliability

48
Geometric Representation of
Signals
• How to represent a energy signal as a linear combination of orthonormal
basis function?

• The M energy signals si(t), i = 1, 2, …, M can be represented as linear

combination of N orthonormal basis function, N  M

• The signals s1(t), s2(t), …, sM(t) over T interval

• The coefficients sij

49
Geometric Representation of
Signals
• The real valued basis functions 1(t), 2(t), …, N(t) form a orthonormal set

• ij is the kronecker delta function

• ij = 1, each basis function is normalized to unit energy

• The basis functions are orthogonal to each other

• Given the coefficients sij, j = 1, 2, …, N, a N dimensional signal vector can

50
be
Geometric Representation of
Signals
• How to generate the signal si(t) with the coefficients?  N bank of multipliers
• How to calculate the coefficients given the signal si(t)?  N bank of correlators

Analyzer
Synthesizer

51
Geometric Representation of
Signals
• Geometric representation of a signal with N = 2 and M = 3

52
Geometric Representation of
Signals
• Signal can be determined by the signal vector

• si can be represented as a N dimensional vector/ N dimensional Euclidean

space

• Signal vectors {si|i = 1, 2, …, M} defines set of M points drawn in N

dimensional Euclidean space (signal space)

53
•
Geometric Representation of
Signals
• How to define the length of a vector and angle between vectors?

• Absolute operator for a one dimensional or two dimensional signal vector

• Norm operator for a multi dimensional signal vector

• The squared length of any signal vector si  inner product or dot product of si
with itself

• j  jth element of si and T  matrix transpose

54
Geometric Representation of
Signals
• How to define the length of a vector and angle between vectors?

• Absolute operator for a one dimensional or two dimensional signal vector

• Norm operator for a multi dimensional signal vector

• The squared length of any signal vector si  inner product or dot product of si
with itself

55
Geometric Representation of
Signals
• Relation between energy of a signal and length of signal vector

• Energy

• Summation is a linear operator, hence interchanging summation and

integration

56
Geometric Representation of
Signals
• Relation between energy of a signal and length of signal vector

• 1(t) is a orthonormal set

• Energy of signal is equal to the squared length of the signal vectors

• Pair of signals represented by signal vectors

57
Geometric Representation of
Signals
• Relation between energy of a signal and length of signal vector

• Euclidean distance between signals

58
Geometric Representation of
Signals
• How to define the length of a vector and angle between vectors?

• Angle ik subtended between two vectors si and sk is

• Cosine of the angle ik is inner product of these two vectors divided by the
product of the individual norms

• When two vectors are perpendicular or orthogonal, their inner product is zero
and ik = 90o
59
Gram – Schmidt
Orthogonalization Procedure
• Given the M energy signals s1(t), s2(t), …, sM(t), how to calculate the
orthogonal basis function 1(t), 2(t), …, N(t) ?

• Gram-Schmidt orthogonalization procedure

• Step 1: Find the first orthogonal basis function 1(t) from s1(t) as

• E1 is the energy of the signal s1(t)

• s11 is the coefficient, E1 = (s11)2 60

Gram – Schmidt
Orthogonalization Procedure
• Step 2: Using s2(t), 1(t) can be obtained as
• Calculate the coefficient of the signal s2(t)

• Intermediate function is g2(t) found

• g2(t) is orthogonal to 1(t) but does not have unit energy

61
Gram – Schmidt
Orthogonalization Procedure
• Step 2: Using s2(t), 1(t) can be obtained as
• orthogonal basis function 2(t) is

62
Gram – Schmidt
Orthogonalization Procedure
• Generalized procedure

• Usually N  M

• The signals s1(t), s2(t), …, sM(t), forms linearly independent set N = M

• The signals s1(t), s2(t), …, sM(t), are not linearly independent set N < M,
63
the
Gram – Schmidt
Orthogonalization Procedure
• Determine the orthogonal basis function for the signals given below.

64
Conversion of Continuous AWGN Channel into
Vector Channel
• The input to the bank of correlators is x(t) (modulated signal plus noise) not
si(t) (Modulated signal)

• w(t)  sample of white Gaussian noise process W(t) with mean zero and
variance/ PSD No/2

• Output of jth correlator is a sample value of random variable Xj

65
Conversion of Continuous AWGN Channel into
Vector Channel
• sij  deterministic component due to si(t)

• wj  sample value of random variable Wj

• New stochastic process X’(t) with a sample function x’(t) related with x(t) as

66
Conversion of Continuous AWGN Channel into
Vector Channel
• Sample function x’(t) depends on channel noise term w(t)

67
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Let the received signal x(t) is a sample function of a random process X(t)

• Xj  random variable with sample value represented by xj

• X(t)  Gaussian random process, Xj  Gaussian random variable

• Wj  random variable generated by sample value wj is generated by the jth

correlator in the presence of noise w(t)
68
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Mean of Xj

• Variance of Xj

69
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Variance of Xj

70
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Variance of Xj

71
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Covariance of Xj Xk

72
Conversion of Continuous AWGN Channel into
Vector Channel
Statistical Characterization of Correlator Outputs

• Outcome of N correlators can be represented as N vectors of mean sij and

variance No/2

• The elements are independent Gaussian distributed

73
Principles of Spread Spectrum
Signals
• The bandwidth of spread spectrum signals is greater than the information
rate R in bits per sec

• Bandwidth expansion factor is greater than unity Be = W/ R

• Redundancy is introduced  helps to overcome interference

• Coding also introduces redundancy and increases the bandwidth expansion

factor over unity

• Pseudorandomness  makes signal appear like a random noise

74
Principles of Spread Spectrum
Signals
Uses of Spread spectrum signals

• Reduces the detrimental effects of interference due to jamming 

interference from other users or self interference

• Hide the signal by using low power for signal transmission  listener will
identify as a noise

• Provides security or privacy with other listeners

• Avoids intentional interference/ jamming

75
Principles of Spread Spectrum
Signals
Uses of Spread spectrum signals

• Pseudorandomness in the coded transmitted signal  prevents jamming

• Pseudorandom pattern (Code) is known to the intended receiver and

unknown to the jammer

• Interference  multiple users share the same channel

• Knowing the pseudorandom pattern the intended receiver can retrieve the
information avoid interference
76
Principles of Spread Spectrum
Signals
Uses of Spread spectrum signals

• Multipath component creates self interference due to time dispersive nature

can be resolved

• Message can be hidden by suppressing the power and spreading the

bandwidth

• Lower power communication  covert  low probability of interception

• Message privacy  only intended user can decode the message

77
Principles of Spread Spectrum
Signals
Model of Spread Spectrum Communication Systems

78
Principles of Spread Spectrum
Signals
Model of Spread Spectrum Communication Systems

• Pseudorandom pattern generator act as interface between Tx and Rx

• Pseudorandom pattern generator generates a pseudorandom/ pseudonoise

binary sequence (PN sequence)

• Synchronization of the PN sequence and the received signal is required

• Synchronization is done by transmitting the pseudonoise sequence which

can be received with higher probability of correct decision
79
Principles of Spread Spectrum
Signals
Model of Spread Spectrum Communication Systems
• Interference  narrow band or broad band based on the information bearing
signals bandwidth

• Modulation used are PSK or FSK

• PSK  phase coherence can be achieved, FSK  phase coherence cannot be

attained

• PSK modulation + PN sequence direct sequence or PN spread spectrum signal

• FSK modulation + PN sequence  Frequency hopping spread spectrum signal80

PN Sequence Generation
Generation of PN Sequence
• PN Sequence  periodic binary sequence  feedback shift register

81
PN Sequence Generation
Generation of PN Sequence
• Feedback shift register  shift register made of m flipflops and logic circuit

• All flipflops are regulated by a single clock signal

• For every clock signal the state of flipflop is changed by the previous flipflop,
logic circuit computes Boolean function

• The output of logic circuit is fed as input to the first flipflop  prevents shift
register getting emptied

• The PN sequence length is determined by the number of flipflops 82

PN Sequence Generation
Generation of PN Sequence
• The PN sequence is determined by the number of flipflops, initial state of
the shift register and feedback logic

• Assume initial state as 1 0 0 83

PN Sequence Generation
Generation of PN Sequence
• The state of the shift register after kth clock pulse is

• sj(k)  state of the jth shift register after kth clock pulse

• The function of the shift register

84
PN Sequence Generation
Generation of PN Sequence
• The Boolean function of the logic network determines the states of the shift
register

• The total possible states of m flipflop is 2m

• The pn sequence generated will be periodic after 2m

• Feedback shift register is linear if it has logic network made of modulo-2

adder, zero state (initial state of all FF are not zero)

85
PN Sequence Generation
Generation of PN Sequence
• Period of pn sequence created by feedback shift register with m FFs will not
exceed 2m-1

• PN sequence  maximal length sequence/ m-sequence  period is 2m -1

• The pn sequence generated will be periodic after 2m

• Feedback shift register is linear if it has logic network made of modulo-2

adder, zero state (initial state of all FF are not zero)

86
PN Sequence Generation
Generation of PN Sequence
• The PN sequence generator with 3 flipflops, initial state of the shift register
is 100 and feedback logic with modulo-2 adder

87
PN Sequence Generation
Properties of m-sequence
• Balance property: Number of ones is always one more than the number of
zeros

• Run property: 1 half the runs of length one, quarter of the runs of length
two, one eight of the runs of length three.

• Run  subsequence of identical symbols, length of the subsequence  length

of a run

• Total runs  (N+1)/2, N = 2m -1

88
PN Sequence Generation
Properties of m-sequence
• Correlation property: autocorrelation function is periodic and binary valued

• The period of m-sequence with m flipflops is N = 2m -1

• 0 and 1 of the m-sequence is represented by -1 and +1 (antipodal

modulation)

• Period of the waveform of c(t) is Tb = NTc

• Tc  chip interval or duration of a symbol 0 or 1 of m-sequence 89

PN Sequence Generation
Properties of m-sequence
• Correlation property: The autocorrelation of c(t) is

90
PN Sequence Generation
Properties of m-sequence
• Correlation property: The power spectral density of c(t) is

91
PN Sequence Generation
Properties of m-sequence
• Correlation property: The autocorrelation of m-sequence looks like a
autocorrelation of random binary wave

• The power spectral density is made of delta function spaced at 1/NTc

interval

Shift registers and

feedback logic

92
PN Sequence Generation
Properties of m-sequence
Auto Correlation and Cross Correlation of PN Sequence

93
PN Sequence Generation

94
Spread Spectrum
• Protection against jamming signal

• Jamming signal  high power broadband noise or multitone waveform

directed towards receiver

• Increasing the bandwidth of the transmission signal  noise like signal

• Anyone listening to the signal believes it to be a noise

• Spread spectrum  camouflaging the information bearing signal

95
Spread Spectrum
• Method of widening the bandwidth of the signal  modulation

• bk  binary message sequence ck  pseudonoise sequence

• b(t)  line coded version (polar NRZ) of message sequence

• c(t)  line coded version (polar NRZ) of pseudonoise sequence

• Multiplying b(t) and c(t) using product modulator (spreading) gives m(t)

96
Spread Spectrum
• b(t)  narrow band signal

• c(t)  wide band signal

• m(t)  wide band signal similar to c(t)

97
Spread Spectrum
• The received signal r(t)

• i(t)  interference signal

• Signal reception

98
Spread Spectrum
• The multiplier output signal z(t)

• b(t)  narrow band signal

• c(t)i(t)  wide band signal

• Passing through LPF the noise or interfering component can be reduced

99
DSSS with BPSK
DS/BPSK Transmitter

100
DSSS with BPSK
DS/BPSK Transmitter

101
DSSS with BPSK
DS/BPSK Receiver

102
DSSS with BPSK
DS/BPSK Receiver

103
DSSS with BPSK
DS/BPSK Receiver
• Synchronization of the PN sequences used at the transmitter and receiver is
required

• Synchronization  acquisition (coarse) and tracking (fine)

• Acquisition (coarse synchronization)  two PN sequences are aligned within a

fraction of chip

• Acquisition  Multiplying the received signal with PN sequence and calculating

the correlation and decision rule, search strategy

104
• Tracking  phase locked techniques
Signal Space Dimensionality and Processing Gain
• The set of orthogonal basis functions

• Tc  Chip duration, N  Number of chips

105
Signal Space Dimensionality and Processing Gain
• The transmitted signal

• Eb  Energy per bit, +  1, -  0

• c0, c1, …, cN-1  ck ± 1 code sequence

106
• Transmitted signal represented as N orthogonal function
Signal Space Dimensionality and Processing Gain
• The jammer signal

• Jammer tries to occupy only the transmitted signal bandwidth but does not
know the phase
107
Signal Space Dimensionality and Processing Gain
• The average power of the jammer signal

• Jammer will apply equal power on sine and cosine term of the signal

108
Signal Space Dimensionality and Processing Gain
• The output of coherent detector

• vs  despread binary PSK signal s(t), vj  spread interference c(t) j(t)

109
Signal Space Dimensionality and Processing Gain
• The first component vs

• The second component

110
Signal Space Dimensionality and Processing Gain
• Approximating PN sequence as independent and identical binary sequence,
the second term

• Vcj and Ck are random variables with sample values vcj and ck

• The probability of ck = 1 and -1 are equiprobable

• Mean of Vcj is zero

111
Signal Space Dimensionality and Processing Gain
• The variance of Vcj for a given pseudo sequence

• Using spread factor N = Tb/Tc, average jamming power

• The Mean and variance of Vcj is zero and JTc/2

112
Signal Space Dimensionality and Processing Gain
• The output signal to noise ratio is

• The input signal to noise ratio is

• Equating Eb/J of output and input SNR gives

113
Signal Space Dimensionality and Processing Gain
• The SNR in terms of decibel

• PG  processing gain  gain in SNR obtained by spread spectrum

• 3-dB  gain in SNR due to coherent detection

• The processing gain and spread factor are equal  spread spectrum

• Large PN sequence  large PG  large SNR 114

Probability of Error
• The coherent detector output of DS/BPSK v is the sample value of RV V

• vcj is the sample value of the RV Vcj

• The decision rule of coherent detection is 1 (v > 0) and 0 (v< 0)

• The transmitted bits 0 and 1 are equiprobable, the probability of error is

115
Probability of Error
• The probability of error depends on the RV Vcj

• Vcj is the sum of N identically distributed RV, by central limit theorem, for
large N, Vcj follows Gaussian distribution

• Vcj can be approximated as Gaussian distribution with mean zero and

variance JTc/2

• The average probability of error for large N

116
Anti-jam Characteristics
• The average probability of error in coherent BPSK is

• In DS/BPSK, the interference can be treated as wide band noise with PSD
N0/2

• Eb is the energy per bit = P Tb, P is the average signal power, Tb is the bit
duration

117
Anti-jam Characteristics
• Using the relation of processing gain PG = Tb/ Tc

• Ratio J/P is called the jamming margin, in decibels is

• (Eb / N0)min is the minimum value needed to support a prescribed average

probability of error

• Jamming margin defines a level of interference a system accepts to provide a

reliable performance 118
Frequency-Hop Spread Spectrum

• DSSS use PN sequence to spread the signal to attain wide bandwidth

• PG = N determines the ability to withstand jamming

• Large PG means small chip duration, more chips per bit and wide BW

• Practical physical devices have some limits on the processing gain (i.e PG
may not be too large)

• Alternate method is to randomly hop over the frequency spectrum

119
Frequency-Hop Spread Spectrum
• The random hopping of data modulated carrier from one frequency to the
other is called frequency-hop spread spectrum (FHSS)

• Common modulation format for FHSS is M-ary frequency shift keying (MFSK)

• FHSS with MFSK  FH/MFSK

• FHSS does not spread across the spectrum instantaneously but sequentially

• Rate at which frequency hop occurs  fast frequency hopping SS and slow
frequency hopping SS
120
Frequency-Hop Spread Spectrum

• Fast frequency hopping SS  symbol rate Rs of the MFSK signal is smaller

than the hop rate Rh

• Several symbols are transmitted on each frequency hops

• Slow frequency hopping SS  symbol rate Rs of the MFSK signal is greater

than the hop rate Rh

• Carrier frequency hops several times during the transmission of one symbol

121
Slow Frequency-Hop Spread Spectrum
• Slow frequency hopping SS  symbol rate Rs of the MFSK signal is smaller
than the hop rate Rh

• Carrier frequency hops several times during the transmission of one symbol

122
Slow Frequency-Hop Spread Spectrum

123
Slow FHSS

124
Fast FHSS

125
Band Pass to Equivalent Low Pass
• Low pass signal  low frequency signal/ baseband signal

• Band pass signal  real, narrow band, high frequency signal

• Easier to work/ handle with low frequency equivalent signal instead of high
frequency signal

• Operating/ working with low frequency signal is easy due to low sampling
rates

• Signal processing algorithms can be implemented at low frequency

126
Band Pass to Equivalent Low Pass
• Spectrum of real valued baseband signal

• S(f) ≈ 0, | f | > W
• S (f) > 0 for some f

• Signal energy is concentrated around DC

• Imaginary part of frequency domain satisfies

conjugate symmetry of Fourier transform

127
Band Pass to Equivalent Low Pass

• Conjugate symmetry property of Fourier transform

• A signal s(t) is real valued s(t) = s*(t) only if it satisfies the following condition

• S(f) = S*(-f) Complex conjugate in Fourier transform

• Re(S(f)) = Re(S(-f)) & Im(S(f)) = -Im(S(-f))

128
Band Pass to Equivalent Low Pass
• Spectrum of real valued passband signal

• S(f) ≈ 0, | f ± fc | > W, fc > W > 0

• Signal energy is concentrated at high freq.

• Bandwidth (BW) is defined only for positive

freq. for real valued signals

• One sided BW and two sided BW

129
Complex Baseband Representation
• Passband channels  transmit and receive passband signals

• Real valued passband signal can be represented as complex valued-baseband

signal

• Complex valued baseband signal  complex envelope/ complex baseband

representation

• Complex envelope can be represented in discrete time with small sampling rate
than the passband signal

• Analog passband signal processing can be kept minimum compared to 130

digital
complex baseband signal processing
Complex Baseband Representation
• Transmitter performs encoding, modulation and filtering using DSP on the
information

• Baseband signal is upconverted before transmission and passband signals are

down converted after reception followed by synchronization, demodulation
and decoding

• Advanced algorithms can be applied for complex baseband signal irrespective

of the signal frequency

131
Time Domain Representation of Passband Signal

• The passband signal sp(t)can be expressed as

• sc(t), ss(t) are real valued signals

• sc(t)  in-phase component of passband signal
• ss(t)  quadrature component of passband signal

• The complex envelope or complex baseband representation of passband

signal sp(t) is

132
Complex Envelope

• Time domain relationship between passband and baseband signals is

• The envelope and phase of the signal is

• Substituting s(t) = e(t)ej(t) gives in sp(t) gives

133
Digital Communication through Bandlimited Channels

• No bandwidth constraint is imposed on the signal design

• If the channel is bandlimited to a bandwidth W Hz

• Channel can be modelled as linear filter with low pass frequency response
C(f)  0, f < W, C(f) = 0, f > W

• The objective is to design a signal pulse g(t) with bandwidth limited to W

• The pulse in a linearly modulated signal is

134
Digital Communication through Bandlimited Channels

• How to use the ideal channel which is bandlimited | f | ≤ W?

• The signal pulse should be designed such that the symbol rates should be
confined within the channel bandwidth

• What if the channel is not ideal?

• The signal transmission with symbol rates ≥ W causes inter symbol

interference (ISI)

• ISI occurs between adjacent symbols

135
Digital Communication through Bandlimited Channels

• How to design a receiver that can mitigate the effect of ISI and AWGN?

• Receiver should compensate/ reduce the effect of ISI in the received signal

• The compensator for ISI is called equalizer

136
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Common channels used in digital communication  Telephone channels

• Telephone channels are characterized by bandlimited linear filters when FDM

is used

• Modern telephone channel uses TDM for multiplexing and PCM for sampling
and encoding analog signal

• Filtering is used before sampling

137
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Telephone channels can be characterized as linear filters with equivalent low

pass frequency response C(f)

• C(f) has equivalent low pass impulse response c(t)

• On passing a signal s(t) through the channel

• s(t)  pass band signal, v(t)  baseband signal

138
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Equivalent low pass received signal is

• z(t)  additive noise

• The signal in frequency domain is V(f)C(f)

• If C(f) is bandlimited, C(f) = 0 |f| > W, the frequency components of V(f) > W
will not be transmitted

139
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• The frequency response of the channel within the bandwidth can be

expressed as

• |C(f)|  magnitude or amplitude response and ϴ(f)  phase response

• The envelope delay characteristic is

140
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• The channel is ideal or non-distorting when |C(f)| is constant for all | f | < W
and ϴ(f) is linear function of frequency, τ(f) is constant for all | f | < W

• If |C(f)| is not constant for all | f | < W, channel distorts the amplitude of the
transmitted signal

• If τ(f) is not constant for all | f | < W, channel introduces signal delay

• Non-ideal channel  amplitude and delay distortion

141
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Non-ideal channel causes successive symbols or pulses to look smeared/

spread  transmitted symbols/ pulses cannot be distinguished

• Pulses overlap  ISI Equalizer output

Received Pulse

Transmitted Pulse 142

Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• PAM with equally spaced pulses  peak of the pulse at zeros of other pulses

• After passing through channel with amplitude and delay distortion  zero
crossings are not equally spaced

• Successive pulses will spread with other and peak of the pulses are not
distinguishable  ISI

• To study the channel and its ability to provide ISI  frequency response
characteristics of the channel is studied
143
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Frequency response of the telephone channel (180 – 725 miles)

• The usable band is 300 Hz to 3000 Hz

144
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Impulse response of the telephone channel is

• The duration is 10 ms

• Symbol transmission rate is 2500 per sec

• ISI extends to 25 – 30 symbols

• Other channel impairments  phase jitter, frequency offset, impulsive noise,

thermal noise, non-linear distortion
145
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Non-linear distortion  non linearities in amplifiers, compandors
• Small and difficult to correct

• Frequency offset  less than 5 Hz  caused by carrier equipment

• Tolerated in low data rate system
• Intolerable in high data system with synchronous detector and adjusted by
carrier recovery system

• Phase jitter  low frequency modulation with power line signal (50 – 60 Hz)
• Serious one and can be compensated to an extent
146
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Impulsive noise  switching of the equipment
• Thermal noise  AWGN  at 30 dB or above signal levels

• Effect of channel impairments on the received signal depends on modulation

technique and data rate

• For 1800 bps (R/W < 1), FSK can be used and insensitive to all channel
impairments

• For 1800 – 2400 bps (R/W ≈ 1), bandwidth efficient QPSK can be used,
equalizer, carrier recovery are used 147
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• For > 2400 bps (R/W > 1), bandwidth efficient coded modulation, trellis
coded QAM, PSK, PAM can be used

• Linear distortion  adaptive equalizer, frequency offset  carrier recovery,

phase jitter  signal design and phase compensation

• For 9600 bps, all channel impairments are taken care

• Channel consisting of all channel impairments are difficult to analyze

• Channel  linear filter with amplitude and delay distortion with AWGN 148
Digital Communication through Bandlimited Channels

Characterization of Bandlimited Channels

• Other physical channels gives ISI through time dispersion

• Short wave ionospheric, tropospheric scatter, mobile radio channels  time dispersive
channels

• Multiple path propagation with different path delays  ISI

• Radio channel  number of path, Relative time delay of the paths vary with time  time
variant radio channel

• Time variant radio channel  variety of frequency response

149
• Telephone channels  Time invariant channel
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• Low pass transmitted signal after modulation

• In  discrete information bearing symbols

• g(t)  bandlimited pulse G(f) = 0, | f | > W

• Transmitted signal is transmitted over bandlimited channel C(f) = 0, |f| > W

• Received signal is

150
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• h(t)  convolution of the pulse with channel

• z(t)  AWGN

• Received signal passed through a filter and sampled at 1/T samples/sec gives

• x(t)  response of the filter to input pulse h(t)

• v(t)  response of the filter to noise z(t)
151
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• The sampled y(t) at t = KT + τo is

• τo  transmission delay through the channel

152
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• If xo is a arbitrary scaling factor set to one

• Ik  transmitted symbol information at kth instant

• The ISI component is

• vk  AWGN component in kth sample 153

Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• ISI can be viewed in DSO as an eye pattern

2 – level PAM 4 – level PAM

• ISI in eye pattern causes the eye to close

154
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• Effect of ISI on eye pattern

• ISI causes the eye to close  error occurs

• ISI  position of aero crossing distorts

• ISI  reduces eye opening

• ISI  synchronization error

155
Digital Communication through Bandlimited Channels
Signal Design for Bandlimited Channels
• 2D scatter diagram for ISI signals (eye pattern) of PSK and QAM

• ISI causes deviation of the received signals from signal points

• Large ISI and noise  deviation will be large
156
Design of Bandlimited Signals for No-ISI
• Conditions to avoid ISI

• Ideal Nyquist criterion for zero ISI

• Nyquist pulse shaping

157
Design of Bandlimited Signals for No-ISI

158
Design of Bandlimited Signals for No-ISI

159
Design of Bandlimited Signals for No-ISI

Drawbacks of Sinc Pulse to achieve zero ISI

• Sinc pulse is non-causal and non realizable

• Rate of convergence of sinc pulse to zero is too slow

• Sampling at improper time (timing error) will lead to infinite ISI

• Raised cosine pulse can be used

160
Design of Bandlimited Signals for No-ISI
• Raised cosine pulse

161
Design of Bandlimited Signals for No-ISI
• Raised cosine pulse

162
Design of Bandlimited Signals for No-ISI
• Raised cosine pulse

•   Roll-off factor or excess bandwidth

•  = 0, raised cosine pulse to sinc pulse

•  = 0.5, raised cosine pulse with 50% excess bandwidth

•  = 1, raised cosine pulse with 100% excess bandwidth

163
Maximum Likelihood Receiver
• Information bits / message mi are mapped to the signal parameters like
amplitude, phase, frequency of the carrier signal  modulation

• Over a symbol duration (0  T), si(t) is transmitted  i = 1, 2, …, M

• si(t)  s1(t), s2(t), …, sM(t)  all signals are equiprobable

• The signal si(t) are obtained through N orthogonal basis function j(t), j = 1, 2,
…, N

• At the receiver, the signal received is passed through bank of N correlators with
164
N orthonormal basis function
Maximum Likelihood Receiver
Correlation receiver

• Outcome of the bank of correlators for the signal si(t) without noise
si = [si1, si2, …, siN]

• The vector si corresponding to the signal si(t) mapped in the Euclidean space
of dimension N  M  transmitted signal point

• Set of message points for all signals si(t), i = 1, 2, …, M  message

constellation
165
Maximum Likelihood Receiver
Correlation receiver

• In the presence of noise

x(t) = si(t) + w(t)

• The outcome of passing x(t) in to the bank of correlators gives

x = [x1, x2, …, xN]

• x  observation vector, transmitted signal vector  si

• Mapping x vector in the Euclidean space differs the signal point si 166
Maximum Likelihood Receiver
Correlation receiver

167
Maximum Likelihood Receiver
• Scatterplot of a transmitted & received signal vector in the presence of noise

168
Maximum Likelihood Receiver
• Observation vector is different from transmitted signal vector due to the
presence of noise

• From observation vector x, the signal x(t) can be mapped to a point in the
Euclidean space

• Received signal point will be anywhere around the transmitted signal point
due to noise

• Maximum likelihood detector  maps the received signal vector x to the

estimate of the transmitted symbol m̂ i with least average symbol error
probability 169
Maximum Likelihood Receiver
Detection problem

• Given the received signal vector x how to find the estimate of the transmitted
symbol m̂ i so that the probability of error is minimized

• Probability of error is

• Decision rule is maximum a posteriori probability rule

170
Maximum Likelihood Receiver
• Using Bayes rule

• fx(x|mk)  likelihood function on transmitting mk

• pk  a priori probability of sending message mk

• If all the symbols are equiprobable pk can be omitted

• The denominator term is unconditional PDF of x and independent of the

transmitted signal 171
Maximum Likelihood Receiver
• fx(x|mk)  likelihood function is non-negative has one – to – one mapping
with log – likelihood function

• On taking natural logarithmic for the likelihood function gives

• Maximum likelihood rule  system that uses the above logic for detection 
maximum likelihood decoder

• Maximum likelihood decoder simplified version of maximum a posteriori

172
decoder with equiprobable messages
Maximum Likelihood Receiver
Graphical representation of the maximum likelihood principle

• Z  observation space  N – dimensional space for all possible observation

vector x

• Z is divided into M decision regions as Z1, Z2, …, Zm  mapped to messages m1,

m2, …, mM

• Decision boundaries will be N-1 dimensional

173
Maximum Likelihood Receiver
Graphical representation of the maximum likelihood principle

• Z  observation space  2 – dimensional space  4 decision regions

• N = 2, M = 4

• Decision boundaries are 1- dimensional

174
Maximum Likelihood Receiver
• In the presence of AWGN, The conditional PDF fx(x|mk) is

• Taking natural logarithmic gives

175
Maximum Likelihood Receiver

• The first term is constant for all possible k

• ln (fx(x|mk)) is maximum when second term is minimum

176
Maximum Likelihood Receiver
• On avoiding the variance which is constant in the second term can be
expressed as

• Decision rule can be expressed as

• Maximum decision rule calculates the distance between the transmitted

signal and observation vectors to find the estimated the transmitted signal

• A symbol is decided based on the minimum distance between transmitted

and received signal vectors 177
Maximum Likelihood Receiver
• Expanding the decision metric gives

• The first term is independent of the transmitted signal vector sk and avoided

• The second term energy of the transmitted signal vector sk

• The third term is the inner product of observation vector x and transmitted
signal vector sk
178
Maximum Likelihood Receiver
• Energy of the transmitted signal is

• The decision rule is

179
Receiver
Detector

• Detector multiplies the received signal with

the N orthogonal basis functions to form the
received signal or observation vector x

• N correlators supplied with N orthogonal

basis functions  1(t), 2(t), …, N(t)
operates on x(t)  x

180
Receiver
Decoder
• Operates on observation vector x and estimates m̂ of the transmitted symbol
mi, i = 1, 2, …, M

• Observation vector x and transmitted symbol vectors si, i = 1, 2, …, M are N

dimensional

• Observation vector x is multiplied with all M transmitted symbol vector si, i =

1, 2, …, M

• Inner product is taken by successive summation (accumulation)

181
Receiver
Decoder
• Transmitted signal vector sk which gives the largest value is mapped to the
estimated symbol

182
Matched Filter
• Alternate method of optimum receivers with linear time invariant filter with
impulse response hj(t)  matched filter

• Why matched filter?

• Orthonormal basis function are time limited over 0  T, the product

modulator at the receiver can be avoided

• Product modulators are hard to build

• Input to the filter  x(t), output of the filter  y(t)

183
Matched Filter
• Output of the filter is convolution of received signal and impulse response of
the filter

• If

• The filtered signal can be written as

184
Matched Filter
• Sampling the filtered signal at T gives

• Rearranging the limits based on the orthogonal basis function

• With bank of correlators, the output of jth correlator is

185
Matched Filter
• The output of the correlator is equivalent to the matched filter by passing the
received signal through the bank of filters and sampled at time T

• Matched filter  Filter with an impulse response which is time reversed and
delayed by T version of the signal j(t)

• The impulse response of the matched filter should be causal

186
Matched Filter

187
Vector Receiver
• The output of the correlator is equivalent to the matched filter by passing the
signal through the filter and sampled at time T

188
Duobinary Signaling
• Correlative – level coding  correlation between successive symbols are
brought

• Duo  doubling the transmission capacity

189
References

• https://www.everythingrf.com/community/what-is-oqpsk-modulation

• Simon Haykin, “Digital Communication Systems”, John Wiley & sons, First
Edition, 2014, ISBN 978-0-471-64735-5.

• Simon Haykin, “Communication Systems”, John Wiley & sons, Fourth Edition.

• John G Proakis and MasoudSalehi, “Digital Communications”, Fifth Edition,

Mc-Graw Hill.

• Upamanyu Madhow, “Fundamentals of Digital Communication”, CUP, 2008.

190