DIGITAL COMMUNICATION

Overview
Communication Channel Classification
• Wired Channel
– Telephone wire
– TV cable or ethernet cable
• Wireless Channel
– Underwater ocean channel carrying acoustic wave
– Free space carrying electromagnetic wave
• Co-located Transmitter and Receiver
– Magnetic tape, optical disk
• Parameters – data rate, attenuation
 Performance measure
• Spectral efficiency
– Data rate / Bandwidth
• BER
– Increasing transmitted signal power
– Improving frequency filtering techniques
– Modulation and demodulation
– Coding and decoding technique
 Geometric representation of signals
• Signal space
– complete set of all signals
• Basis functions
– Collection of minimum no. of functions to represent
a given signal
– Independent
– Orthogonal
T


0
j (t )  k (t ) dt K j jk , 0 t Tj , k 1,2,....N
---(1)
– Kj is a non zero constant 1 j k
 jk 
0 otherwise
• Orthogonality of the basis functions demands
that the inner product of two different basis
functions should always be zero
• Dimension of the signal space
– Minimum no. of basis functions
• Basis set
– Collection of basis functions
• Orthonormal basis set
– When the energy for all the basis functions are
normalised so that each Kj = 1, the basis set is called
an orthonormal basis set.
• Any arbitrary finite set of waveforms sk(t)
(k=1,2,3,…….M), where each member of the set is
physically realisable and of duration T, can be
expressed as a linear combination of N
orthogonal waveforms  1 (t ), 2 (t )..... N (t ), where N M
such that
s1 (t ) a11 1 (t )  a12 2 (t )  .....  a1N N (t )
s2 (t ) a21 1 (t )  a22 2 (t )  .....  a2 N N (t )

 ----(2)


sM (t ) aM 1 1 (t )  aM 2 2 (t )  .....  aMN N (t )
• These relationship can be expressed in more
compact notation as
N
sk (t )  akj j (t ), k 1,2..., M; j 1,2....N ---(3)
j 1

where,
T
1
akj  sk (t ) j (t )dt , k 1,...M; j 1,...N; 0 t T
Kj 0 ---(4)
• The basis coefficient akj is the value of the  j (t )
component of signal sk(t). The set of signal
waveforms, {sk(t)}, can be viewed as a set of
vectors {sk} = {ak1,ak2,……,akN} ----(5)
• Once a set of N basis functions has been adopted,
each of the transmitted signal waveforms, sk(t) is
completely specified by the vector of its
coefficients.
Sk = (ak1,ak2,….akN) k = 1,2,…N ----(6)

• Using Eq. (1 & 3), the Normalized energy Ek

associated with the waveform sk(t) over a symbol
interval T can be written as 2
T
 T

E 
k s (t ) dt   a  (t )  dt
2
k kj j
0 0  j 
T
 a kj j (t ) a ki i (t )dt
j i
0
 T
  a kja ki  j (t) i (t)dt
j i 0
N
  a kja ki K j jk  a 2kjK j , k 1,2...M
j i j1 ---(7)
• Equation ( 7 ) is a special case of Parseval’s
theorem relating the integral of the square wave
of the waveform sk(t) to the sum of the square of
the basis coefficients. If orthonormal functions
are used (i.e. Kj = 1), the normalized energy over
a symbol duration T is given by N
E k  a 2
kj
---(8)

j1
 Summary
• To Check Orthogonality:
T


0
j (t )  k (t ) dt K j jk , 0 t Tj , k 1,2,....N
1 j k
 jk 
0 otherwise

• To expressN signal in terms of orthogonal functions

sk (t )  akj j (t ), k 1,2..., M; j 1,2....N
j 1

• where T
1
akj  sk (t ) j (t )dt , k 1,...M; j 1,...N; 0 t T
Kj 0
 Problem - 1
• A set three waveforms s1(t), s2(t) and s3(t) are
shown in the figure 2.1.
– (a) Demonstrate that these waveforms do not
form an orthogonal set.
– (b) Show that  1 (t ) and  2 (t ) form a basis set
– (c) Express the signal set si(t) in terms of basis set
 3 (t )  4 (t )
– (d) Verify that and also form a basis set
– (e)Express the signal set si(t) in terms of basis set
 Gram-Schmidt Orthogonalization
Procedure (GSOP)
• To construct a set of orthonormal basis functions
from a set of finite energy signal waveforms
{sk(t), k=1,2,….M}.
• The Energy associated with s1(t) is given by
T
E1 s12 ( t ) dt
0
s1 ( t )
1 ( t ) 
• First basis function is constructed as E1
• Second basis function:

c12  s 2 ( t )1 ( t )dt


 '2 ( t ) s 2 ( t )  c12 1 ( t )
'
 (t)
 2 (t)  2

E2
• Generally the kth basis function can obtained
as  (t) 
 'k ( t )
k
Ek
k 1
 'k ( t ) s k ( t )  c
i 1
ik i (t)

c ik  s k ( t ) i ( t )dt , i 1,2,3....k - 1

• If the dimensionality of the signal space
corresponding to the given M signal waveforms
is also M , this procedure would lead to exactly
M number of basis functions.
• If the dimensionality of the signal space is N
(N≤M), the GSOP generates only N basis
functions.
• GSOP automatically detects the dimensionality
of the signal set and generates the same
number of basis functions.
• Once the set of orthonormal waveforms  n ( t )
the M signals {sk(t)} as linear combinations of the
basis functions  n (t ) . Therefore, it can be
written as
N
s k ( t )  s kn  n ( t ), k 1,2...M
n 1

s kn  s k ( t ) n ( t )dt , k 1,2,3...M; n 1,2..N;

 2
N
and E k  s k ( t ) dt  s 2kn  s k
2

 n 1

• Each signal may be represented by the vector Sk

= [sk1 sk2 sk3……..skN] or equivalently, as a point in
the N-dimensional signal space with coordinates
{s , i=1,2…N}.
 Signal space
• Representation of M message points on N
dimensional space
• Procedure to draw signal space
– Determine the dimension of the signal space (N)
– Find the number of message points (M)
– Find the coordinates for each message point sk
– sk = [sk1 sk2 sk3……..skN], k=1,2…..M
Signal space
s1 ( t )  2T 1 ( t )
s 2 ( t )  2T  2 ( t )
T T
s 3 ( t ) 3 1 ( t )   2 (t)
2 2

s1 ( 2T ,0)
s 2 (0, 2T )
 T T
s 3  3 , 
 2 2 
Bandwidth
 Bandwidth
• Half-power Bandwidth
– Interval between frequencies at which Gx(f) has
dropped to half-power or 3 dB below the peak
value
• Noise equivalent Bandwidth Px
Wn 
G x (f c )
– Px – total signal power over all frequencies
– Gx(fc) – value of the maximum spectral component
• Null-to-null Bandwidth
– Width of the main spectral lobe
• Fractional Power Containment Bandwidth
– Power contained within the band is 99 percent;
above and below the band is 0.5 percent of the total
signal power residues
• Bounded power spectral density
– Everywhere outside the specified band Gx(f) must
have fallen at least to a certain level below that
found that at the band center
– Typical attenuation level 35 or 50dB.
• Absloute bandwidth
– Interval between the frequencies outside which the
spectrum is zero
Bandwidth of Digital signal or Data
• Let s(t) be a some practical waveform (signal,
noise or signal-noise combination) to be
represented over the interval a<t<b.
• As we know, it the signal s(t)) can be represented
by an equivalent orthogonal series as
• Where an is the
s( t )basis
coefficient
a n  n ( t ) and is a
basis set. Complex exponential,
n
Harmonic
 n (t)
sinusoidal, Bessel Function, Legendre Polynomial
and Sinc function are examples of orthogonal sets
• Complex Fourier series also represent a basis
set where  n ( t ) e jn t 0

2 T0
• n takes integral values,0  , T0 b  a , and k n 
T0 2
• Alternatively, any physical waveform (i.e.
finite energy) can be represented over the
interval a<t<a+T0 as

s( t )   C n e jn0 t

n 
a T0
1

- jn0 t
where, C n  s ( t ) e dt
T0 a
• If the waveform is periodic T0, this Fourier series
representation becomes valid over the entire
time axis, -∞ < t < ∞ by virtue of the periodicity
of the basis functions  n ( t ) with a common
period of To. But it is not so in digital
communication signal.
• In order to represent a non-periodic signal s(t)in
a small time window in terms of their basis 1
*
functions, let us take n n and - B nf 0 B , f 0  T
C  C
0
• Now, it can be written as
BT0
s( t )   n
C
n  BT0
e jn0 t
• C0 is a DC component does not convey any
information regarding the signal. Hence, 4BT0
real coefficients to characterise s(t).
C n C*n
• Due to , 2BT0 coeffecients only
required to describe s(t)
• Coeffecients Cn actually depend on the shape
of s(t).
• Dimensionality Theorem
– A real waveform can be completely specified by N
independent pieces of information where N is
given by N = 2BT0.
• Following the theorem, the signal s(t) can be
represented by N finite no. of basis functions as
N
s( t )  s k  k ( t )
• Where {sk} and { } are k 1 the digital data set and

 k of the signal s(t).

basis set respectively
• For storing and reconstructing a band limited
signal over an interval of To, at least N points
should be stored.
 Bandwidth
• A digital signal is represented by N points which
are transmitted over an interval of T0 secs.
Hence, the symbol rate R s  N  2BT0 2B
T0 T0
Rs
• Bandwidth of the signal is B
2
– where Rs is the symbol rate.
– Lower limit is achieved for sinc pulse basis set
– For rectangular pulse basis, the bandwidth equals
the symbol rate (baud rate D)D
– Bandwidth of the data B  2
Mathematical Models of
Communication Channel