Introduction

• Communication is defined as the process of

conveying messages between two/multi
distant points, i.e., source(s) to destination(s).

• Generally, Digital Communication theory is

applicable to the exchange of (any form of)
message signals in the digital form
irrespective of the distance between men &
m/c(s).
 • Digital communication deals with the detailed
fundamentals relating to the analysis & design of
modern age EM/electronic communication systems
for providing local/global connectivity for
information exchange, including the
Sampling process,
quantization,
PCM system,
digital modulation Techniques,
Signal detection,
various line codings for digital data transmission
along with most of the
 modern multiple access techniques.
 Text Books:-
1. Taub’s Principles of Communication Systems, by
Taub, Schilling & Saha, 4/e, MGH.

2. Modern Digital and Analog Communication Systems,

by B. P. Lathi and Z. Ding, 4/e, Oxford University Press.

Reference Books :-
1. Digital Communications by Simon Haykin,
John Wiley & Sons.
2. Communication Systems Engg by J.G. Proakis
& M. Salehi, 2/e, Pearson Education.
 Source
• After the year 1990, digital communication
systems (DCSs) are being dominated in our daily
lives.
• DC deals with the problems of transmitting the
digital/binary data/bit string (groups of “1s” &
0s”) over a communication channel.
• We assign a distinct waveform (pulse) to each of
these 2 symbols. The resulting sequence of these
pulses is transmitted over a channel.
• At the receiver these pulses are detected & are
converted back to binary data (“1s” & 0s”).
• A DCS consists of several components as shown
in Fig.-1.
• The I/P to a digital system is a sequence of binary
bits or a string of (“1s” & “0s”) which is the o/p
from a data set, a computer or a digitized audio
signal (PCM, DM, etc,) or a HDTV or telemetry
data, etc.
• The process of converting the o/p of either an
analog or digital source into a sequence of binary
digits efficiently is called source encoding or data
compression.
• The use of the channel encoder is to introduce in
 a controlled manner some redundancy in the binary
information sequence that can be used at the receiver
to overcome the effects of noise & interference
encountered in the transmission of the signal through
the channel. So the added redundancy serves to
increase the reliability of the received data &
improves the strength of the received signal.
Line Coder
• The digital o/p of a source/channel encoder (are the
binary bit string) is converted into electrical
waveforms (pulses) to modulate carrier (or directly
used) for transmitting over the channel. This process
is called line coding or transmission coding.
I/P

Received Source Baseband Demodulator Demux

O/P Decoder Demodulator

Fig.1(i): Basic building blocks of DCS

• Before digital transmission, analog signal is coded into
binary forms, i.e., 0s & 1s.
• But how these 0s & 1s are converted into electrical
 Digital Transmitter
Noise

Digital Receiver

Fig.-1(ii):-Block diagram of a typical digital communication system.

 voltages/pulses is known as line coding.
• Quality line coding depends on the following
factors: (i) Minimum power & BW for
transmission, (ii) ability to extract/recover
timing/clock information, (iii) ability to reduce
low frequency or dc component as it is unsuitable
for ac coupled circuits, (iv) Favorable PSD for
matching to the available channel frequency
response, (v) ease of signal detection & decoding,
(vi) improve error detection & correction
capability, (vii) transparency to correctly transmit
a digital signal regardless of 0‟s & 1‟s, etc.
• The line codes are classified as shown in Fig.-2 &
its digital representation is shown in Fig.-3 for
digital data 111001…, i.e., there are many
possible ways of assigning waveforms (pulses) to
the above digital data as shown in Fig.-3.
• In unipolar (on-off) code, logic „0‟ is represented
by 0v & logic „1‟ by constant +Vv, i.e., in the on-
off case, „1‟ is transmitted by a pulse p(t) & a „0‟
is transmitted by no pulse as shown in Fig.-3 (a)
& (d).
• In case of NRZ +Vv is maintained throughout its
bit period whereas in case of RZ +Vv is
maintained half of its bit period as shown in Fig-
3.
• Unipolar NRZ coding is simple but creates
Fig.-2 Classification of line codes.
 On-off (RZ)

Polar (RZ)

Bipolar/AMI (RZ)

On-off (NRZ)

Polar (NRZ)

Fig.3(i): Few line code examples.

 serious problems due to high dc as well as low
frequency components. Long string of 1‟s & 0‟s
also make clock recovery difficult & causes
synchronization problem. Unipolar RZ reduces
these problems to some extent & requires less
power.
• Another commonly used code is polar, where „1‟
is encoded by a pulse p(t) & „0‟ is transmitted by
a pulse -p(t) as shown in Fig.-3 (b) & (e). In polar
code logic 0 is represented by -Vv & logic 1 by
constant +Vv.
• The polar scheme is the most power-efficient code,
since it requires the least amount of power for a
given noise immunity (error probability).
• Another popular code in PCM is bipolar, also known
as pseudoternary or alternate mark inversion (AMI),
where „0‟ is encoded by no pulse & „1‟ is encoded by
a pulse p(t) or –p(t), i.e., consecutive „1s‟ alternate in
sign as shown in Fig.-3 ©.
• Thus in alternate mark (=1) inversion code, logic „0‟
is represented by 0v & logic „1‟ by constant +Vv or –
Vv alternately for a fraction of the bit interval. It
removes accumulation of a dc component & helps in
bit error correction.
• Thus the AMI code has the advantage that if one
single error is made while detecting the pulses, the
received pulse sequence will violate the bipolar rule
& the error can be detected (although not corrected)
immediately.
• But, if multiple errors happen in the sequence,
then error detection in AMI code is impossible
due to cancellation effects. However the
probability of multiple errors is much smaller
than that of single errors.
• Even for single errors, we cannot tell exactly
where the error is located. So the AMI code can
detect the presence of single errors, but it cannot
correct them.
• A modified version of AMI known as the High
Density Bipolar (HDB) code introduces
deliberately 1‟s when a long string of zeros
 (a)

(b)

(d)

(e)

(f)

© N

Fig.3(ii)- Electrical representation of 0s & 1s.

 appear in AMI to help clock synchronization. In
HDB-3 code, if there are more than 3 zeros in a
string the 4th one is replaced by a violation pulse.
• In Biphase (split-phase) or (Manchester) coding,
no dc component & clock recovery is easier as
shown in Fig.-3 (f).
• For example:- Bipolar RZ-AMI code has the best
overall characteristics among all these discussed
line codes.
 Multiplexer
• Generally, the capacity of a physical channel
(e.g., coaxial cable, optical fiber) for transmitting
data is much larger than the data rate/BW
requirement of individual source.
• Hence to utilize this capacity effectively, we
combine several users/sources by means of a
digital multiplexer. The digital multiplexing can
either be FDM or TDM.
• Thus, by the use of mux-demux pair, a physical
channel is normally shared by several messages
simultaneously.
 Regenerative Repeater
• These are used at regular intervals along a digital
transmission line to detect the incoming signal &
regenerate new “clean” pulses for further
transmission along the line/channel.
• This process periodically eliminates & thereby
combats, accumulation of noise & signal
distortion along the transmission path.
• The ability of such regenerative repeaters to
effectively eliminate noise & signal distortion
effects is one of the biggest advantages of digital
communication systems over analog counterparts.
• If it is a long distance comn, then sufficient no. of
repeaters (known as amplifiers over analog comn
channel) can be located in the midpoints to raise
the SNR value till reaching the destination.
• Thus, in analog comn, repeater amplifies both
signal & in-band noise while in digital comn,
Regenerative repeaters are placed at intermediate
places during long distance comn to remove noise
completely from the original binary information
before retransmitting them.
• Hence digital transmission of signals is most
suitable for very very long distance comns.
 Periodic timing information extraction
• If the bits/pulses are transmitted at a rate of Rb
pulses per sec, then we need the periodic timing
information – the clock signal at Rb Hz – to
sample the incoming pulses/bits at a repeater.
• This timing information can be extracted from the
received signal itself if the line code is chosen
properly.
• For example, when the RZ polar signal is in Fig.-
3(i) (b) is rectified, it gives a periodic signal
having clock frequency Rb Hz as shown in Fig.-4.
• Fig.-4 contains the desired periodic timing signal
which can be applied to a resonant ckt tuned to
frequency Rb Hz, thus the o/p of the tuned ckt will
be a sinusoid of Rb Hz, which can be used for
timing.
• The timing signal (the resonant ckt o/p) is
sensitive to the incoming bit pattern.
• For example, in the on-off or bipolar case, a „0‟ is
transmitted by „no pulse.‟ Hence, if there are too

Fig.4- To extract clock frequency.

 many zeros (no pulses) in a string/sequence, then
there is no i/p signal to the resonant ckt for a
longer time. Then the o/p of the resonant ckt
starts decaying, which causes error in the timing
information.
• A line code in which the bit pattern does not
affect the accuracy of the timing information is
said to be transparent line code.
• Thus the RZ Polar scheme is transparent, whereas
the on-off & bipolar line codes are
nontransparent.
 Advantages of DC over AC
• Analog signals are varying continuously with
time whereas digital signals has finite number of
discrete levels. Thus, in digital Comn, the
message is discrete in nature & the no. of
discrete levels are finite.
• DC, which can withstand channel noise &
distortion much better than analog as long as the
noise & distortion are within limits, i.e., DC is
more noise immune than AC. With analog
messages, any distortion or noise, no matter how
small, will distort the received signal.
• Analog comn is less costly & require less BW, but
digital comn is less affected by noise & hence
suitable for long distance comn via regenerative
repeaters.
• Regenerative repeaters are placed at intermediate
places during long distance comn to remove noise
completely from the original binary information & to
retransmit them. But, in analog comn, repeater
amplifies both signal & in-band noise.
• In digital comn we can employ different coding
techniques to improve BER & we can implement
compression technique to reduce the bit rate which in
turn reduces BW requirement. It is not possible in
analog comn.
• Every 2-3 yrs, digital hardware cost reduces by
half while its capacity doubles. Storage &
indexing of digital signals is easy & inexpensive.
Thus it has the ability to search & select
information from distant electronic databases.
• Also, digital hardware implementation is flexible
in nature that helps moving to newer technology
standard with relative easiness. Hence we can
coordinate with microprocessors, digital
switching, & VLSI.
• It is easier & more efficient to multiplex several
digital signals due to common formats for all
types of messages.
• Reproduction & transportation of digital
messages are easier, extremely reliable without
deterioration, & less costlier than analog
messages.
• Thus the use of digital methods offers several
important operational advantages over analog
methods, like:-
01. Increased immunity to channel noise & external
interference; error correction is easy.
02. Flexible operation of the system.
03. A common format for the transmission of all
types of message (e.g., voice, video, data) signals.
04. Improved security of comn through the use of
encryption.
• The development of sophisticated, high-speed digital
comn systems has been accelerated by concurrent
developments in inexpensive high speed ICs &
programmable DSP chips.
 Sampling Theorem
• Analog signal is converted to digital form
through the process of Sampling, Quantization,
and Encoding. This A/D conversion sets the
foundation of modern digital comn systems.
• Naturally an analog signal is converted into
digital form by (1) sampling in time, then (2)
quantizing in amplitude & (3) coding into binary
form.
• Through the use of the sampling process, an
analog signal is converted into a corresponding
sequence of samples that are usually spaced
 uniformly in time, known as discrete time signal.
• These discrete time signals when coded in binary
offer advantages like storage, error correction,
greater immunity to noise, etc.
• Clearly, for such a procedure to have practical
utility, it is necessary that we choose the sampling
rate properly (lossless), so that the sequence of
samples uniquely defines the original analog
signal.
• This is the essence of the sampling theorem to
reconstruct/build the original analog signal
without any distortion (with sufficient accuracy).
• Nyquist theory states that to represent an analog
(Low pass) signal by its sampled version, the
sampling frequency should be greater than or equal
to twice the maximum frequency present in the low
pass band limited analog signal, known as sampling
1 1
theorem, ie,. f s ( sampling rate)  2 f m  Ts   .
fs 2 fm
• The process of uniformly sampling a continuous
time signal of finite energy results in a periodic
spectrum with a repetition frequency equal to the
sampling rate.
• Thus Sampling Theorem For Low pass signal:- Let
m(t) be a message signal, which is band-limited such
that its highest spectral component is fM, then the
sampling rate, fs should be ≥ 2fM for the faithful
reconstruction of the original analog signal from its
samples without any distortion.
fs = 2 fM ← Nyquist/ideal/critical rate of sampling.
fs < 2 fM ← under sampling is avoided due to signal
aliasing or distortion.
fs > 2 fM ← Over sampling is desirable for sampling.
• The pulses in Fig.-5(b) are narrow width of dt. The
spectral density of the
band-limited message, fm
is shown in Fig.-6 (a).

Samples of the signal m(t)

Fig.5: A signal m(t) is sampled, controlled by the unity

pulse train S(t) having period = Ts.
• fs= 2fm is the Nyquist rate. An increase in sampling
rate above the
Nyquist rate
increases the Spectrum of band limited
width of the message, fm
guard band,
hence easing fs = 2 fm
the problem
of filtering.
• Bandpass
Sampling
Theorem:- For
a bandpass Fig.-6(b) Spectrum (DTFT) of sampled message signal.
 C
M(f) M(-f)

Fig.-6 :- Spectrum of sampled message

(d) with spectrum aliasing.
 Signal with highest frequency fH & bandwidth B,
can be recovered from its samples through bandpass
filtering by sampling it with frequency fs = 2fH/k
where, k is the largest integer not exceeding fH/B.
• Pb-01:- A signal m(t) = 2cos 6000πt + 4cos8000πt +
6 cos10000πt is to be confidently represented by its
samples. What is the minimum sampling rate as per
(a) low pass Sampling theorem & (b) band pass
Sampling theorem ?
• Solution:- (a) fs = 10000 Hz & (b) B = 5000 -3000 =
2000 & k = floor(5000/2000) = 2. Then sampling
frequency, fs = 2fH/k = 5000 Hz.
 Sampling in transmitting multiple band-
limited signals (Concept of TDM)
• A technique by which we may take the advantage of
the sampling principle for the purpose of TDM is
shown in Fig.-7.
• At the transmitting (left side of Fig.-7) end, a no. of
band-limited signals are connected to the contact
point of a rotary switch (Commutator).
• As the rotary arm swings around, it samples &
collects each user signal sequentially for
transmission via a single channel. Similarly, the
rotary switch (Decommutator) at the receiving end
is in syncrhronism.
• The commutator samples & combines samples,
Fig.-7:- Block diagram of a PAM-TDM system
 While the decommutator separates samples
belonging to individual signals so that these signals
may be reconstructed.
• Both the rotary switches must make at least 2fm = fs
revolutions per second for individual message signal
reconstruction.
• Multiplexing of several PAM signals is possible
because the various signals are kept distinct & are
separately recoverable as these are sampled at
different times. Hence this system is an example of a
TDM system.
• You may transmit the multiplexed signal directly (as
shown in Fig-7) over a high-speed transmission line
 Or the TDM-PAM signals can be transmitted via antenna
through analog/digital modulation processes.
• Thus when a band limited analog signal is sampled then
the samples are known as the discrete time (PAM)
signals as shown in Fig.-8(a).
• Thus putting a large no. of samples coming from
different sources one after another in a time band forms
the TDM system.
• Generally, in a PAM/sampling system, the duration of a
pulse (τ) is much less than the time period of pulses (Ts),
i.e., τ << Ts. As a result a large numbers of PAM signals
from other users can be accommodated within that pulse
duration known as TDM-PAM system as shown in Fig.-
7. Fig.-8(b) shows TDM of 2 signals.
 Message signal

Unit pulse train

Product of (a) & (b)

Fig.-8(a) The Natural samples of a message signal.
Fig.-8(b) TDM of two message signals.
• Again, TDM system is superior to FDM system
in the following ways:-
(i) FDM system requires different
subcarriers/channels hence different BPFs are
required to multiplex & separate whereas, in
TDM system all the sub-carriers/channels require
identical circuits comprised of simple
synchronous switches, gates, & a LPF. Hence the
circuitry needed in the TDM system is much
simple than the one needed in the FDM system.
(ii) The nonlinearities in various amplifiers of an
FDM system produce harmonic distortion &
hence introduce interference within channels.
Whereas in a TDM system, the signals from
different channels are allotted at different time
slots & they are not present simultaneously.
Hence, the TDM system is relatively immune to
interference (inter channel cross-talk) in
comparison to the FDM system.
 Signal Reconstruction
• In the A/D converter, the sampling rate must be large
enough to permit the analog signal to be
reconstructed from the samples with sufficient
accuracy.
• The Sampling Theorem, which is the basis for
determining the proper (lossless) sampling rate for a
given signal.
• Nyquist theory states that to represent an analog
(Low pass) signal by its sampled version, the
sampling frequency should be greater than or equal
to twice the maximum frequency present in the low
pass band limited analog signal, known as sampling
 1 1
theorem, i.e., f s ( sampling rate)  2 f m  Ts   .
fs 2 fm
• Thus, Sampling Theorem For Low pass signal:- Let
m(t) be a message signal, which is band-limited such
that its highest spectral component is fM, then the
sampling rate, fs should be ≥ 2fM for the faithful
reconstruction of the original analog signal from its
samples without any distortion.
fs = 2 fM ← Nyquist/ideal/critical rate of sampling.
fs < 2 fM ← under sampling is avoided due to signal
aliasing or distortion.
fs > 2 fM ← Over sampling is desirable for sampling.
 C

Fig.-6 :- Spectrum of sampled message

(d) with spectrum aliasing.
• The process of reconstructing a continuous time
signal g(t) from its samples (collected
uniformly), ḡ(t) is known as Interpolation.
• If the analog signal g(t) is band limited to B Hz,
then it can be reconstructed (interpolated) exactly
from its samples, ḡ(t) as:

1
g (t )  g (t ) Ts (t )   g (nTs ) (t  nTs )   g (t ) e jn 2f s t
...(a)
Ts n  

• This means not only the uniform sampling at

above the Nyquist rate preserves all the signal
information, but also that simply passing the
sampled signal, ḡ(t) through an ideal LPF of BW
 B Hz & gain Ts will reconstruct the original
message.
• Such an ideal LPF has transfer function:
    f 
H ( f )  Ts    Ts   ...(b)
 4B   2B 
• Ideal LPF is noncausal & unrealizable as its
impulse response is a sinc pulse extending upto
∞ duration.
• For practical signal reconstruction purpose a
causal & a practically realizable LPF of BW B
Hz can be used.
 Quantization of signals
• By the process of sampling analog signal is
discretized in the time axis known as PAM signal
where amplitudes are still continuous.
• Similarly by the quantization process amplitude
axis of the analog signal is discretized, i.e., the
sample (continuous amplitude) belongs to the mid
point value falling to a particular quantization
level out of finite number of levels which then
helps for encoding into binary digits for complete
conversion of analog signal to digital form,
known as PCM.
• When a message signal, m(t) is quantized a new signal,
mq(t) is created which is an approximation to m(t) & free
from the additive noise while transmitted for a longer
distance.
• The quantization process is explained in Fig.-9. Here a
message/analog signal m(t) is confined to a range of VH
(7.5 V) toVL (-0.5 V). Suppose it is quantized to M (say,
8) no.s of levels. Then the step size, S = ( VH -VL)/M.
• The characteristic of a quantizer is described by a
staircase function. Quantizer can be of a uniform or non-
uniform type.
• Due to quantization process quantization noise, qe is
introduced = true signal value of m(t)- quantized signal,
value mq(t) .
 7.5 V

-0.5 V
Fig.-9 The quantization process.
• Due to quantization process the amount of noise
introduced in each sample is called the
quantization error, S
qe  ....(i )
2
• Hence, the quantization error can be minimized
by reducing the step size (S).

• Then the mean-square quantization error or

quantization noise power is :-
2
S
q 2
e  ....(ii )
12
Fig.-10 A quantized signal wave & the corresponding error curve.
• The conversion of an analog (continuous)
sample/amplitude of the message signal into a
digital (discrete ) form is called the quantizing
process.
• The difference between 2 adjacent discrete values
is called a quantum or step size, S.
• The quantization error is the difference between
the i/p & o/p signals of the quantizer.
• Quantizing noise signal or quantizing error, qe
varies randomly within the interval –S/2 ≤ qe ≤
S/2.
• Average power of quantizing noise, Pq is :-
2
1 S /2 S
Pq   S / 2 q dqe  12
2
e ....(ii )
S
(assuming samples are uniformly distributed over all the
levels)
• Thus, the av. Power of quantizing noise grows as the
square of the step size, S where the S is under the
designer‟s control.
• Hence, the message signal distortion due to quantizing
noise can be controlled by choosing the appropriate
smallest step size.
• In a practical system, 256 quantization levels give a quality
commercial voice/colour TV, while 64 quantization levels
give only fairly good performance.