Ravi Shankara.M.N.

ECE Dept, SCE Bangalore

Associate Professor ravishankarmn@[Link]

Module 2 - Signalling Over AWGN Channels

.
Introduction
The conversion of analog waveforms into coded pulses represents the transition from analog communications to digital
communications. This transition has empowered by several factors like:
1. Advancement and improvement of digital silicon chips, digital signal processing, and computers,
2. Improved reliability in digital communications
3. Broadened range of multiplexing of users, which is enabled by the use of digital modulation techniques.
4. Different communication networks, uses digital communications as the preferred choice

Conceptual Model of Digital Communication System

Figure shows the conceptual model of digital communication system. The message source emits the
message sequence consisting of discrete message symbols one symbol every T sec with the symbols
belonging to an alphabet of M symbols denoted as m1, m2,…….mM. If M = 2  m1 = 0 and m2 = 1.
If M=4  m1 = 00, m2 = 01, m3=10 and m4 =11. It is assumed that all M symbols of the alphabet are
equally likely,
1
Hence pi = P emitted = for all i
M

The output of the message source is presented to a vector transmitter which produces a vector of real
numbers as output and is of the form for the message symbols mi as
si1
si2
si = . i = 1, 2 3, … … . M where N  M
.
siN .
The Modulator constructs a real valued distinct finite energy signal si(t) for every symbol emitted by
T
the message source of duration T seconds and the energy of the signal is given by Ei = 0 i
s (t)2 dt
The Channel is assumed to have two characteristics:
Channel is linear, with a bandwidth that is large enough to accommodate the transmission of the
modulator output si(t) without distortion.
The transmitted signal si(t) is perturbed by an additive, zero-mean,
stationary, white, Gaussian noise process denoted by w(t) and such a
channel is referred as AWGN ( additive white Gaussian noise ) channel.
The received signal x(t) = si(t) + w(t)
The Receiver observes the received signal x(t) for a duration T and makes a
best estimate of the transmitted signal.
Owing to the presence of channel noise, the receiver will inevitably make occasional errors.
The requirement, therefore, is to design the receiver so as to minimize the average probability of
symbol error and is defined as
Pe = P1 mi P m = 0 1 sent + P2 mi P m = 1 0 sent
This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 1
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

where P1(mi) and P2(mi) are the prior probabilities of transmitting symbols 1 and 0, respectively, and m
is the estimate of the symbol 1 or 0 sent by the source, which is computed by the receiver. The
P m = 0 1 sent and P m = 1 0 sent are conditional probabilities.

Geometrical Representation of Signals

Any set of M energy signals, {si(t)}, can be expressed as linear combinations of N Orthonormal basis
functions, where N ≤ M. This scheme provides
mathematical basis for
If s1(t), s2(t), ….., sM(t) are real valued energy signals, each of duration ‘T’ sec,
representation of energy
N signals and hence
0≤t≤T
si t = sij ∅j t − − − − − (A) provides a satisfactory
i = 1,2,3 … M way for the noise
j=1
analysis in digital
Where the coefficients of the expansion are defined by communication systems.

0≤t≤T
T
Sij = 0 i
s t ∅j t dt i = 1,2,3, … . . M -----(B)
j = 1,2,3 … … N
The real valued basis functions 1(t), 2(t), ….. … N(t) forms an orthonormal set and are orthogonal
to each other and are defined as
T 1 if i = j
∅i t ∅j t dt = δij = −−−−−− 1
0 0 if i ≠ j
where ij is the Kronecker delta
In equation (1); the first condition states that each basis function is normalized to have unit energy
(condition for orthonormal function  Inner product = 1).
T T
0
∅i t ∅i t dt = 0
[∅i t ]2 dt = 1

The second condition states that the basis functions 1(t), 2(t),…… N(t) are orthogonal with respect
to each other over the interval 0  t  T ( Inner product or dot product is zero)
T
∅i t ∅j t dt = 0
0

Scheme for Generating the signals from signal vector

Given the N elements of the vector si operating as input, the scheme to generate the signal si(t), as
given in equation (A) is shown below. This figure consists of a bank of N multipliers with each
multiplier having its own basis function followed by a summer. This scheme may be viewed as a
synthesizer.
Scheme for Generating Signal Vector si
Conversely, given the signals si(t), i = 1, 2, .., M, operating as input, the scheme to calculate the
coefficients si1, si2, .., siN as given in equation (B). This scheme consists of a bank of N product
integrators or correlators with a common input, and with each one of them supplied with its own basis
function. This scheme may be viewed as an analyzer.

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 2
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Signal Vector -The signal si(t) can be completely described by the vector si with its coefficients as
The vector si is called Signal vector. The set {si}i=1,2,..M
represents the co-ordinates of M message points in
N- dimensional Euclidean space, with N mutually
perpendicular axis labeled as 1, 2 … N
The N dimensional Euclidean space with M message points is called Signal space diagram or Signal
Constellation diagram.

Ex: If M = 3 and N = 2 ; results in three signals s1(t),

s2(t) and s3(t) and two orthonormal basis functions 1(t)
and 2(t)
−s11 s21 s31
Let s1 = s12 s2 = s s3 = s
22 32

Length of Signal Vector and Inner product

The length of the signal vector is defined by ||si||. The squared length of any signal vector si is defined
to be the inner product or dot product of si with itself. ||si||2 = [si . siT] = (si, si)
si1 si1
si2 si2
si = .  si = . . si1 si2 … . siN = s11 2 + s12 2 + s13 2 + … . +sMN 2
2

. .
siN siN

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 3
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

| si |2 = (sij )2 where sij are elements of si

j=1
s i ,s j si T . sj
Cosine of angle between si and sj Cos θij = =
s i || .||s j s i || .||s j

*** If the inner product is zero, the signals are said to be orthogonal.
Relation between Energy content of a signal and its vector length
t
The energy of the signal si(t) is given by Ei = 0 i
[s (t)]2 dt
N N
t
Ei = sij ∅j t sik ∅k t dt
0 j=1 k=1

Interchanging the order of summation and integration

N N
t
Ei = sij sik ∅j t ∅k t dt − − − − − − − −( A)
j=1 k=1 0

Consider the condition i = j

t
The second term 0
∅j t ∅k t dt forms an orthonormal set and its integral value is unity for j=k
Hence equation ( A) reduces to
N N

Ei = sij sij = [sij ]2 = | si |2

j=1 j=1
Hence the energy of the signal si(t) is equal to the squared length of the signal vector representing it

Questions from previous question papers

1. Explain the geometric representation of signals. Show that energy of the signal is equal to the squared length of the vector
representing it.
2. Explain the geometric representation of set of M energy signals as linear combinations of N orthonormal basis functions.
Illustrate for the case N = 2 and M = 3 with necessary diagrams and expressions.
3. Explain the geometric representation of signals. Illustrate the geometric interpretation of signals for the case of two
dimensional signal space with 3 signals s1(3,1), s2(1,2) and s3 (2,3).

Gram-Schmidt Orthogonalization (GSO)

To show that it is possible to construct a set of ‘N’ orthonormal basis functions ϕ1(t), ϕ2(t), ….., ϕN(t)
from the linearly independent signals {si(t)}, i = 1, 2, ….., N
If s1(t), s2(t), ….., sM(t) are real valued energy signals, each of duration ‘T’ sec,
N
0≤t≤T
si t = sij ∅j t − − − − − (A)
i = 1,2,3 … M
j=1

s 1 (t)
Let the first basis function ∅1 t =  s1 t = E1 ∅1 t = s11 ∅1 t ----------(1)
E1

T
where E1 is the energy of the signal s1(t) and E1 = 0 1
[s (t)]2 dt and s11 = E1

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 4
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Let the intermediate function g 2 t = s2 t − s21 ∅1 (t) such that g2(t) is orthogonal to 1(t)
T T T
g 2 t ∅1 t dt = s2 t ∅1 t dt − s21 ∅1 t ∅1 t dt = s21 − s21 1 = 0
0 0 0

Hence g2(t) is orthogonal to Φ1(t) in the interval 0 to T

T T Q. Explain the
2 2 Gram-Schmidt
Energy of g 2 t = [g 2 t ] dt = s2 t − s21 ∅1 t dt Orthogonalization
0 0
procedure.
T
| g2 t | = s2 2 t − 2s21 s2 t ∅1 t + s21 2 ∅1 2 t dt
0
T T T
| g2 t | = 0 2
s 2 t dt − 2s21 0
s2 t ∅1 t dt + s21 2 0
∅1 2 t dt

g2 t = E2 − 2 s21 . s21 + s21 2 . 1 = E2 − s21 2

g 2 (t) s2 t − s21 ∅1 (t)
The second basis function Φ2 t = =
energy of g 2 (t) E2 − s21 2
Where E2 is the energy of signal s2(t)
Continuing in the same procedure,
i−1
g i (t)
g i t = si t − sij ∅j (t) and ∅i t = i = 1,2,3 … … . . N
j=1
energy of g i (t)

The above equation gives the orthonormal set from which one can generate orthogonal signals as a
linear combination of set of basis functions.
Problems
1. s1(t), s2(t) and s3(t) are represented by 2 basis functions 1(t) and 2 (t). Draw the signal constellation
diagram and express s1(t), s2(t) and s3(t) interms of 1(t) and 2 (t).
4 −3 −4
The signal vectors are s1 = s2 = s3 =
0 0 4
Solution:
Given (S11, S12) = (4, 0); (S21, S22) = (–3, 0) and(S31, S32) = (–4, 0)
2
0≤t≤T
si t = sij ∅j t
i = 1,2,3
j=1

s1(t) = S111(t) + S122(t) = 4 1(t) s2(t) = S211(t) + S222(t) = –3 1(t)

and s3(t) = S311(t) + S322(t) = –4 1(t) + 4 2(t)
[Link] the constellation diagram and express s1(t), s2(t) and s3(t) as a linear combination of the basis
functions 1(t) and 2(t); given s1 = (s11, s12) = (3,0); s2 = (s21, s22) = (-2,3); & s3 = (s31,s32) = (-3,-3)

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 5
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Solution:
Given s1 = (s11, s12) = (3,0); s2 = (s21, s22) = (-2,3); & s3 = (s31,s32) = (-3,-3)
2
0≤t≤T
si t = sij ∅j t
i = 1,2,3
j=1

s1(t) = S111(t) + S122(t) = 3 1(t) s2(t) = S211(t) + S222(t) = –21(t) +32(t)

and s3(t) = S311(t) + S322(t) = –3 1(t) –3 2(t)
3. Check whether the signals 1(t) and 2(t) are orthogonal? Obtain the corresponding orthonormal
functions. Express the signals s1(t), s2(t) and s3(t) interms of 1(t) and 2(t).

Solution:
2 1 2
Consider the integral ∅1 t ∅2 (t)dt = 1 (1)dt + 1 −1 dt = 0
0 0 1
 Hence 1(t) and 2(t) are orthogonal to each other.
2 1 2
E∅1 = [∅1 t ]2 dt = 2 J and E∅2 = [∅2 t ]2 dt + [∅2 t ]2 dt = 1 2
+ (−1) 2 = 2 J
0 0 1
Corresponding orthonormal functions are
1
; 0≤t≤1
∅1 t 1 ∅2 t 2
z1 t = = and z2 t = =
E∅1 2 E∅2 −1
; 1≤t≤2
2
From the figure, the given signals can be expressed in terms of 1(t) and 2(t) as follows
s1 t = ∅1 t + ∅2 (t); s2 t = −∅1 t + ∅2 (t) and s3 t = ∅1 t − ∅2 (t)
[Link] the orthogonal signal pair shown, construct the signal constellation for s1(t) and s2(t)

T
T T
Solution: Energy of signal s1 t is given as E1 = o 1
[s (t]2 dt = o
2 1 2 dt + T −1 2 dt = T
2
s1 t 1
The first basis function is given by ∅1 t = =
E1 T
Let g2(t) be the intermediate function such that so that g2(t) and Φ1(t) are orthogonal to each other
g2(t) = s2(t) – S21Φ1(t)

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 6
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

T T/2 T
1 −1
s21 = s2 t ∅1 t dt = 1. dt + 1. dt = 0
0 0 T T/2 T
Hence g2(t) = s2(t)
T T T
Energy of signal g 2 t is given as E2 = o
[g 2 (t)]2 dt = o 2
[s (t)]2 dt = o
1 2 dt = T
g2 t 1
The second basis function is given by ∅2 t = =
Eg2(t) T
The coefficients are S11 = E1 = T S12 = 0 S21 = 0 and S22 = E2 = T
S S 0
The signal vectors are s1 = 11 = T and s2 = 21 =
S12 0 S 22 T
5. Using Gram-Schmidt Orthogonalization procedure, express s1(t) and s2(t) interms of orthogonal basis
functions.

Solution: Given number of independent signals M = 2 and the basis functions required to represent these
signals N=2
T T
2
Energy of signal s1 t is given as E1 = s1 t dt = A2 dt = A2 T
0 0

s1 t A
The first basis function is given by ∅1 t = =
E1 A2 T
1
∅1 t = ; 0t T
T
Let g2(t) be the intermediate function such that so that g2(t) and Φ1(t) are orthogonal to each other
g2(t) = s2(t) – S21Φ1(t)
T
T T
2 1 1 2A T
s21 = s2 t ∅1 t dt = 2A dt + 0 dt = +0 =A T
0 0 T T
2
T T 2
1 T
2A − A T =A 0<𝑡<
T 2
g 2 t = s2 t − s21 ∅1 t =
1 T
0−A T = −A <𝑡<𝑇
T 2
T
T T
2
2 2
Energy of g 2 t is given as E2 and E2 = A dt + −A dt = A2 T
T
0
2
1 T
; for 0 ≤ t ≤
g2 t T 2
The second basis function is given by ∅2 t = =
Energy of g 2 t 1 T
− ; for ≤t≤ T
T 2

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 7
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

The given signals can be expressed in terms of 1(t) and 2(t) as follows
s1(t) = S11 1(t) = S11 E1 = A T 1(t) ; s1(t) is defined for the period 0  t  T

s2(t) = S21 1(t) + S22 2(t) = S211(t) + E2 2(t) = A T 1(t) + A T 2(t) ;

s2(t) is defined for the period 0  t  (T/2)

6).Consider the signal s1(t), s2(t), s3(t) and s4(t) as shown in figure. Use Gram-Schmidt Orthogonalization
procedure to obtain the basis functions. Construct the corresponding signal space diagram.

Solution:
By inspection; s4(t) = s1(t) + s3(t)  s4(t) is not linearly independent.
Hence the number of linearly independent signals N= 3 and Number of distinct symbols M = 4
Since N = 3, Only three basis functions are required to represent the given signals.
Energy of the signal s1(t) is The first basis function 1(t) is
T
t
2
3 T s1 t 1 3 T
E1 = s1 t dt = 12 dt = ∅1 t = = = −−− 0≤t≤
0 0 3 E1 T T 3
3
Intermediate function
g 2 t = s2 t − s21 ∅1 (t)
T 2T
T
3 3 3 T
s21 = s2 t ∅1 t dt = 1 dt + 1 0 dt =
0 0 T 0 3

𝐓 T 3
For the period 𝟎 ≤ 𝐭 ≤ 𝟑; g 2 t = s2 t − s21 ∅1 t = 1 − . =0
3 T

𝐓 𝟐𝐓 T
For the period ≤𝐭≤ ; g 2 t = s2 t − s21 ∅1 t = 1 − . (0) = 1
𝟑 𝟑 3

𝟐𝐓 T
For the period ≤ 𝐭 ≤ 𝐓; g 2 t = s2 t − s21 ∅1 t = 0 − .0 = 0
𝟑 3
𝐓 𝟐𝐓
Energy of g2(t) is computed for the period ≤𝐭≤ only as g2(t) is zero in other intervals
𝟑 𝟑
𝟐𝐓
T
2
𝟑 T
Energy of g 2 t is given as E2 = [g 2 t ] dt = [1]2 dt =
0
𝐓 3
𝟑
g2 t 1 3 T 2T
∅2 t = = = −−−− ≤t≤
Eg T T 3 3
2 t 3

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 8
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

𝐒𝐞𝐜𝐨𝐧𝐝 𝐈𝐧𝐭𝐞𝐫𝐦𝐞𝐝𝐢𝐚𝐭𝐞 𝐟𝐮𝐧𝐜𝐭𝐢𝐨𝐧 g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t

T 2T
T T 3 3 T
s31 = 0 3
s t ∅1 t dt = 0
3 1 0 dt = 0 and s32 = 0 3
s t ∅2 t dt = T 1 dt =
T 3
3

𝐓 T T 3
𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 𝟎 ≤ 𝐭 ≤ ; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 1 − 0 − . =0
𝟑 3 3 T

𝐓 𝟐𝐓 T 3
𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 ≤𝐭≤ ; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 1 − 0 0 − . =0
𝟑 𝟑 3 T

𝟐𝐓 T
𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 ≤ 𝐭 ≤ 𝐓; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 1 − 0 0 − . (0) = 1
𝟑 3

2T
Since g3(t) has a finite value for the period ≤ t ≤ T, the energy is computed for this period only.
3
T
T
Energy of g 3 t = E3 = (1)2 dt =
2T 3
3

g 3 (t) 3 2T
The third basis function is given by ∅3 t = = ; for ≤t≤T
Energy of g 3 (t) T 3

The three basis functions 1(t), 2(t) and 3(t) forms an orthonormal set. The signals s1(t), s2(t), s3(t) and s4(t)
do not form a linearly independent set since s4(t) = s1(t) + s3(t).

Any of these four signals are expressed as a linear combination of the three basis functions as given below.
s1(t) = s11 1(t) s2(t) = s21 1(t) + s22 2(t) s3(t) = s31 1(t) + s322(t ) + s33 3(t)
s4(t) = s1(t) + s3(t) = s11 1(t) + s31 1(t) + s322(t ) + s33 3(t)
Note: s11 = E1 s22 = E2 and s33 = E3
7. Using Gram- Schmidt orthogonalization procedure, find the set of orthogonal basis functions to
represent three signals as shown in figure. Also express each of the signals in terms of basis functions.

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 9
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Solution:
From the given figures, it can conclude that s1(t), s2(t) and s3(t) are linearly independent. Hence M = 3 and
N = 3, three basis functions are required.
Energy of the signal s1(t) is The first basis function 1(t) is defined as
𝑡 1
2
𝑠1 𝑡 2
𝐸1 = 𝑠1 𝑡 𝑑𝑡 = 22 𝑑𝑡 = 4 ∅1 𝑡 = = =1 −−− 0≤𝑡≤1
0 0 𝐸1 4
𝐅𝐢𝐫𝐬𝐭 𝐈𝐧𝐭𝐞𝐫𝐦𝐞𝐝𝐢𝐚𝐭𝐞 𝐟𝐮𝐧𝐜𝐭𝐢𝐨𝐧 g 2 t = s2 t − s21 ∅1 (t)
T 1
s21 = s2 t ∅1 t dt = −4 1 dt = − 4
0 0
For the period 𝟎 ≤ 𝐭 ≤ 𝟏; g 2 t = s2 t − s21 ∅1 t = −4 − (−4) .1 = 0
For the period 𝟏 ≤ 𝐭 ≤ 𝟐; g 2 t = s2 t − s21 ∅1 t = −4 − −4 . 0 = −4
For the period 𝟐 ≤ 𝐭 ≤ 𝟑; g 2 t = s2 t − s21 ∅1 t = 0 − (−4) .0 = 0
Energy of g2(t) is computed for the period 𝟏 ≤ 𝐭 ≤ 𝟐 only as g3(t) is zero in other regions
2
Eg 2 t = E3 = −4 2 dt = 16 Joules
1
g2 t −4
The second basis function is given by ∅2 t = = = −1 −−−− 1≤ t≤ 2
Eg 16
2 t

𝐒𝐞𝐜𝐨𝐧𝐝 𝐈𝐧𝐭𝐞𝐫𝐦𝐞𝐝𝐢𝐚𝐭𝐞 𝐟𝐮𝐧𝐜𝐭𝐢𝐨𝐧 g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t

T 1 T 2
s31 = s3 t ∅1 t dt = 3 1 dt = 3 and s32 = s3 t ∅2 t dt = 3 −1 dt = −3
0 0 0 1

𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 𝟎 ≤ 𝐭 ≤ 𝟏 ; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 3 − 3 1 – −3 . (0) = 0

𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 𝟏 ≤ 𝐭 ≤ 𝟐 ; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 3 − 3 0 – −3 . (−1) = 0
𝐅𝐨𝐫 𝐭𝐡𝐞 𝐩𝐞𝐫𝐢𝐨𝐝 𝟐 ≤ 𝐭 ≤ 𝟑; g 3 t = s3 t − s31 ∅1 t − s32 ∅2 t = 3 − 3 0 – (−3) . (0) = 3
Since g3(t) has a finite value for the period 2 ≤ t ≤ 3, the energy is computed for this period only.
3
Energy of g 3 t = Eg 3(t) = 32 dt = 9 Joules
2

g 3 (t) 3
The third basis function is given by ∅3 t = = =1 for 2 ≤ t ≤ 3
Energy of g 3 (t) 9

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 10
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

(b) The three basis functions 1(t), 2(t) and 3(t) forms an orthonormal set. The signals s1(t), s2(t), s3(t) and

s4(t) form a linearly independent set. Note: 𝑠11 = 𝐸1 𝑠22 = 𝐸𝑔 2(𝑡) 𝑎𝑛𝑑 𝑠33 = 𝐸𝑔 3(𝑡)

s1(t) = 2 1(t) ; s1(t) is defined for the period 0  t  1

s2(t) = – 41(t) + 42(t) ; s2(t) is defined for the period 1  t  2 and is split into 0  t  1 & 1  t  2
s3(t) = 3 1(t) – 3 2(t) + 3 3(t) ; s3(t) is defined for the period 0  t  3

Conversion of the Continuous AWGN Channel into a Vector Channel

(Response of Bank of Correlators to Noisy Input)
The received signal x(t) is random in nature and can be expressed
as random process given by
𝑥 𝑡 = 𝑠𝑖 𝑡 + 𝑤(𝑡) 0  t  T, i = 1, 2, 3, ….. M
-------------------------(A)
Where w(t) is the sample function of White Gaussian noise
process with zero mean and PSD is (No / 2)
The output of each Correlator is also random in nature and is
defined as
𝑇 𝑇
𝑥𝑗 = 0
𝑥 𝑡 ∅𝑗 𝑡 𝑑𝑡 = 0
𝑠𝑖 𝑡 + 𝑤 𝑡 ∅𝑗 𝑡 𝑑𝑡
𝑇 𝑇
𝑥𝑗 = 0 𝑖
[𝑠 𝑡 ∅𝑗 𝑡 𝑑𝑡 + 0
𝑤 𝑡 ∅𝑗 𝑡 𝑑𝑡 = 𝑠𝑖𝑗 + 𝑤𝑗 ------(B) where j = 1, 2, 3, ……. N
The first term sij is a deterministic quantity contributed by the transmitted signal si(t) and given by
𝑇
𝑠𝑖𝑗 = 0 𝑖
𝑠 𝑡 ∅𝑗 𝑡 𝑑𝑡
And the second term wj is a random variable due to the presence of noise and is defined by
𝑇
𝑤𝑗 = 0
𝑤 𝑡 ∅𝑗 𝑡 𝑑𝑡 -------------- ( C)
Consider a new random process 𝑋 ′ 𝑡 whose sample function 𝑥 ′ 𝑡 related to the received signal x(t) as follows
𝑁
′
𝑥 𝑡 =𝑥 𝑡 − 𝑥𝑗 ∅𝑗 𝑡
𝑗 =1
𝑁
′
𝑥 𝑡 = 𝑠𝑖 𝑡 + 𝑤 𝑡 − 𝑠𝑖𝑗 + 𝑤𝑗 ∅𝑗 (𝑡)
𝑗 =1
𝑁 𝑁
′
𝑥 𝑡 = 𝑠𝑖 𝑡 + 𝑤 𝑡 − 𝑠𝑖𝑗 ∅𝑗 𝑡 − 𝑤𝑗 ∅𝑗 (𝑡)
𝑗 =1 𝑗 =1
𝑁 𝑁
′
𝑥 𝑡 = 𝑠𝑖 𝑡 + 𝑤 𝑡 − 𝑠𝑖 𝑡 − 𝑤𝑗 ∅𝑗 (𝑡) = 𝑤 𝑡 − 𝑤𝑗 ∅𝑗 𝑡 = 𝑤 ′ (𝑡)
𝑗 =1 𝑗 =1

Hence 𝑥 ′ 𝑡 depends only on noise w(t) at the front end of the receiver and not at all on the transmitted signal
si(t). Thus the received random process is expressed as
This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 11
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

𝑥 𝑡 = 𝑥𝑗 ∅𝑗 𝑡 + 𝑤 ′ 𝑡 − − − − − − − − − − − −(𝑀 )
𝑗 =1

From the above equation, it can be concluded that the Correlator output {xj}, j = 1 to N with each xj is a
Gaussian random variable and is characterized completely by its mean and Variance.
Statistical Characteristics of Correlator Output
Mean
th
Let WJ denote the random variable represented by the sample value wj produced by the j correlator in
response the white Gaussian noise component w(t). Since the noise process w(t) has zero mean, the random
variable wj extracted from w(t) also has zero mean. Hence the mean value of the jth Correlator output depends
only on sij and given by
𝜇𝑋 𝑗 = 𝐸[𝑥𝑗 ] = 𝐸 𝑠𝑖𝑗 + 𝑤𝑗 = 𝑠𝑖𝑗 + 𝐸 𝑤𝑗 = 𝑠𝑖𝑗 + 0 = 𝑠𝑖𝑗
Show that mean of Correlator output xj
Variance of xj depends only on Sij and all Correlator outputs
have a Variance equal to PSD of noise (NO/2)
Variance of xj is given by

𝜎 2 𝑋 𝑗 = 𝑉𝑎𝑟 𝑥𝑗 = 𝐸 (𝑥𝑗 − 𝜇𝑋 𝑗 )2

𝜎 2 𝑋 𝑗 = 𝐸 (𝑥𝑗 − 𝑠𝑖 𝑗 )2 = 𝐸 (𝑠𝑖𝑗 + 𝑤𝑗 − 𝑠𝑖 𝑗 )2 = 𝐸 𝑤𝑗 )2
𝑇
Using 𝑤𝑗 = 0
𝑤 𝑡 ∅𝑗 𝑡 𝑑𝑡
𝑇 𝑇
2
𝜎 𝑋𝑗 =𝐸 𝑤 𝑡 ∅𝑗 𝑡 𝑑𝑡 𝑤 𝑢 ∅𝑗 𝑢 𝑑𝑢 𝑤𝑕𝑒𝑟𝑒 𝑡 𝑎𝑛𝑑 𝑢 𝑎𝑟𝑒 𝑎𝑛𝑦 𝑑𝑢𝑚𝑚𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠
0 0

Interchanging the order of integration and expectation

𝑇 𝑇
𝜎 2 𝑋𝑗 = ∅𝑗 𝑡 ∅𝑗 𝑢 𝑑𝑡𝑑𝑢 𝐸 𝑤 𝑡 𝑤 𝑢
0 0
𝑇 𝑇
2
𝜎 𝑋𝑗 = ∅𝑗 𝑡 ∅𝑗 𝑢 𝑅𝑊 𝑡, 𝑢 𝑑𝑡 𝑑𝑢 − − − (1)
0 0

Where Rw(t,u) = E[w(t)w(u)] is the autocorrelation function of the noise process W(t) and since the noise is
stationary with a PSD value (NO/2) and Rw(t,u) depends only on the time difference (t – u ) and expressed as
𝑁𝑜
𝑅𝑊 𝑡, 𝑢 = 𝛿 𝑡 − 𝑢 − − − − − − − (𝐴 ) Shifting property of  function
2 𝑇
𝑇 𝑇 𝑇 𝑓 𝑢 𝛿 𝑡 − 𝑢 𝑑𝑢 = 𝑓(𝑡)
2
𝑁𝑜 𝑁𝑜 2 0
𝜎 𝑋𝑗 = ∅𝑗 𝑡 ∅𝑗 𝑢 𝛿 𝑡 − 𝑢 𝑑𝑡𝑑𝑢 = [∅𝑗 𝑡 ] 𝑑𝑡
2 0 0 2 0
𝑁𝑜
𝜎 2 𝑋𝑗 = 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑗 {since j(t) has unit energy}
2
Hence the Variance of the Correlator output xj is equal to the PSD (No/2) of additive noise process W(t).
Covariance of Xj
Covariance is a measure of how changes in one variable are associated with changes in the second variable.
Covariance measures the degree to which two variables are linearly associated.

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 12
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Since Xj are mutually uncorrelated, the Covariance of Xj are given by

𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = 𝐸 𝑋𝑗 − 𝜇𝑋 𝑗 𝑋𝑘 − 𝜇𝑋 𝑘 = 𝐸 𝑋𝑗 − 𝑆𝑖𝑗 𝑋𝑘 − 𝑆𝑖𝑘

𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = 𝐸 (𝑆𝑖𝑗 + 𝑊𝑗 − 𝑆𝑖𝑗 (𝑆𝑖𝑘 + 𝑊𝑘 − 𝑆𝑖𝑘 = 𝐸 𝑊𝑗 𝑊 𝑘

Show that the elements of Vector X
𝑇 𝑇 (Correltor outputs Xj) are statistically
independent
𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = 𝐸 𝑊 𝑡 ∅𝑗 𝑡 𝑑𝑡 𝑊 𝑢 ∅𝑘 𝑢 𝑑𝑢
0 0
𝑇 𝑇 𝑇 𝑇
𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = ∅𝑗 𝑡 ∅𝑘 𝑢 𝐸 𝑊 𝑡 𝑊 𝑢 𝑑𝑡𝑑𝑢 = ∅𝑗 𝑡 ∅𝑘 𝑢 𝑅𝑊 𝑡, 𝑢 𝑑𝑡 𝑑𝑢
0 0 0 0
𝑁𝑜 𝑇 𝑇
𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = 0 0 𝑘
∅ 𝑢 ∅𝑗 𝑡 𝛿 𝑡 − 𝑢 𝑑𝑡 𝑑𝑢
2
𝑇
𝑁𝑜
𝐶𝑜𝑣 𝑋𝑗 𝑋𝑘 = ∅𝑗 𝑢 ∅𝑘 𝑢 𝑑𝑡 = 0 𝑓𝑜𝑟 𝑗 ≠ 𝑘 (Since j k are orthogonal to each other)
2 0

This shows that the Correlator outputs are highly uncorrelated and each Correlator o/p are different and it can
be concluded that they are statistically independent.

Questions from previous question papers

Derive the expression for mean and variance of the correlator output. Also show that the correlator outputs are statistically
independent.
Explain how to convert continuous AWGN channel into vector channel
Show that for a noisy input, the mean value of a jth correlator output xj depends only on sij and all the correlator outputs xj ;
j =1,2,… N have variance equal to PSD (N0/2) of the additive noise process w(t).

Conditional Probability Density Function

Let X be a vector of N independent Gaussian random variables with mean value s ij and variance (NO/2). The
PDF of X is defined as
1 1
𝑓𝑋 𝑥 = exp⁡
[− 𝑥 − 𝜇𝑥 )2 − − − − − −(𝐴)
2𝜋𝜎𝑥 2 𝑁𝑜

Since the elements are statistically independent, the conditional PDF of X given that the signal si(t) or
corresponding symbol mi was sent as the product of conditional PDF of individual elements
𝑁

𝑓𝑋 𝑥|𝑚𝑖 = 𝑓𝑋𝑗 𝑥𝑗 |𝑚𝑖 𝑖 = 1,2,3, … 𝑀 − − − − − (𝐵)

𝑗 =1

x and xj are sample values of random variable X and Xj. The vector x is called as observation vector and xj is
called an element of observation vector.
Using the values of mean and variance, equation (A) can be written as
1 1 1 1
fX x = exp⁡[− xj − sij )2 = exp⁡[− x − sij )2 − − − − − −(C)
2π(No /2) No πNo No j
Using equation (C) in equation (B) results

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 13
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

𝑁 N N
1 1 2 −𝑁/2
1
𝑓𝑋 𝑥|𝑚𝑖 = exp − (xj − sij ) = πNo exp − (xj − sij )2
πNo No No
j=1 j=1
Likelihood Function
The conditional PDF fX(x|mi) of each transmitted message mi is known as likelihood function denoted by l(mi).
l(mi)= fX(x|mi) i= 1, 2, …M
For convenience, the Likelihood function is expressed in natural log and given by
N N
𝑁 1 𝑁 1
𝐿 𝑚𝑖 = ln πNo − 2 exp − (xj − sij ) 2
= − ln 𝜋𝑁𝑜 . − (xj − sij )2
No 2 No
j=1 j=1

Since the first term is constant and the expression reduces to

N
1
𝐿 𝑚𝑖 ≅ − (xj − sij )2
No
j=1

Optimum Receiver using Coherent detection – Maximum Likelihood Decoding

Figure shows the constellation of message point due to
transmitted signal si and its corresponding received signal point
constellation The received signal point and message will not be
same due to the presence of additive white noise in the channel
w(t).The correlator output at the receiver defines the observation
vector x and is given as x = si + w
Due to the presence of noise, the received signal point wanders
about the message point in a random fashion and may lie
anywhere inside in a Gaussian distributed cloud centered about the message point.
The signal detection problem is stated as “ Given an observation vector x, perform a mapping from x to an
estimate 𝑚 of the transmitted symbol mi in a way that would the minimize probability of error in the decision
making process “.
Maximum Likelihood detector (decoder)
(a) Given the observation vector x, for the chosen decision 𝑚 = 𝑚𝑖 , the probability of error is given by
𝑃𝑒 𝑚𝑖 𝑥 = 1 − 𝑃( 𝑚𝑖 𝑠𝑒𝑛𝑡 𝑥
On the basis of the above, the optimum decision rule is stated as
Choose 𝑚 = 𝑚𝑖 if
𝑝 𝑚𝑖 𝑠𝑒𝑛𝑡 𝑥 ≥ 𝑝 𝑚𝑘 𝑠𝑒𝑛𝑡 𝑥 𝑤𝑕𝑒𝑟𝑒 𝑘 = 1,2,3 … 𝑀 𝑎𝑛𝑑 𝑘 ≠ 𝑖 − − − − − −Rule(1)
Rule 1 is referred to as Maximum A posteriori Probability (MAP) rule and the system used to implement this
rule is called a MAP decoder.
(b) The MAP rule can be represented considering Likelihood probabilities using Baye’s Theorem as

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 14
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

𝜋𝑘 𝑓𝑋 (𝑥|𝑚𝑘 )
𝑠𝑒𝑡 𝑚 = 𝑚𝑖 𝑖𝑓 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑓𝑜𝑟 𝑘 = 𝑖 − − − − − Rule(2)
𝑓𝑋 (𝑥)
Where k is the prior probability of transmitting symbol mk, It can be noted from rule-2 that (i) the
fX(x|mk) is the conditional PDF of the random observation vector X denominator term fX(x) is independent

when the symbol mk is transmitted fX(x) is the unconditional PDF of X of the transmitted symbol (ii) the prior
probability k = i when all the source
Rule 2 is referred to as Maximum Likelihood rule and the system
symbols are transmitted with equal
used to implement this rule is called a Maximum Likelihood Decoder.
probability and (iii) the conditional
(c) Graphical representation of Maximum Likelihood rule. PDF fX(x|mk) bears one-to-one
Let Z denotes the N-dimensional space of all possible observation relationship to the log-likelihood
vector x and this space is known as Observation space and this space is function L(mk).

correspondingly partitioned into M decision regions denoted by Z1, Z2….ZM. The decision rule can be restated
as
𝑆𝑒𝑡 𝑚 = 𝑚𝑖 ; 𝑂𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑉𝑒𝑐𝑡𝑜𝑟 𝑥 𝑙𝑖𝑒𝑠 𝑖𝑛 𝑡𝑕𝑒 𝑟𝑒𝑔𝑖𝑜𝑛 𝑍𝑖 𝑖𝑓 𝐿 𝑚𝑘 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑓𝑜𝑟 𝑘 = 𝑖 − −Rule(3)
N
2
d Also (xj − skj )2 = x − sk
j=1
Where ||x – sk|| is the Euclidean distance between the observation vector x at the receiver input and the
transmitted vector sk. Hence the decision rule can be restated as
𝑂𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑉𝑒𝑐𝑡𝑜𝑟 𝑥 𝑙𝑖𝑒𝑠 𝑖𝑛 𝑡𝑕𝑒 𝑟𝑒𝑔𝑖𝑜𝑛 𝑍𝑖 𝑖𝑓 the Euclidean distance x − sk 𝑖𝑠 𝑚𝑖𝑛𝑖𝑚𝑢𝑚
𝑓𝑜𝑟 𝑘 = 𝑖 − −Rule(4)
The Rule (4) implies that the Maximum Likelihood rule is simply to choose the message point closest to the
received signal point.
(e) Expanding the summation term of Likelihood function
𝑁 𝑁 𝑁 𝑁 𝑁
2 2 2 2 1 2
𝑥𝑗 − 𝑠𝑗𝑘 = 𝑥𝑗 − 2𝑥𝑗 𝑠𝑗𝑘 + 𝑠𝑘𝑗 = 𝑥𝑗 −2 𝑥𝑗 𝑠𝑗𝑘 − 𝑠𝑘𝑗 − − − −(𝐵)
2
𝑗 =1 𝑗 =1 𝑗 =1 𝑗 =1 𝑗 =1
The first term in equation (B) gives the energy of the signal and is independent of index k, hence it is ignored.
The second term is the inner product of observation vector x and transmitted signal vector Sk and the third term
is the energy of the signal sk(t). Minimization of LHS is equivalent to
maximization of RHS.
N 𝑁 𝑁 𝑁
2
1 2 1
(xj − skj ) = 𝑥𝑗 𝑠𝑗𝑘 − 𝑠𝑘𝑗 = 𝑥𝑗 𝑠𝑗𝑘 − 𝐸𝐾
2 2
j=1 𝑗 =1 𝑗 =1 𝑗 =1

The Maximum Likelihood (ML) decision is reformulated based on the

above equation as
𝑂𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑉𝑒𝑐𝑡𝑜𝑟 𝑥 𝑙𝑖𝑒𝑠 𝑖𝑛 𝑡𝑕𝑒 𝑟𝑒𝑔𝑖𝑜𝑛 𝑍𝑖
𝑁
1
𝑖𝑓 𝑥𝑗 𝑠𝑗𝑘 − 𝐸𝐾 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 − − − −𝑅𝑢𝑙𝑒(5)
2
𝑗 =1
𝑘=𝑖

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 15
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

The figure illustrates the above statement for M = 4 signals and N = 2 dimensions, assuming that the signals are
transmitted with equal energy E and equal probability.
Correlation Receiver
For an AWGN channel, when the transmitted signals s1(t), s2(t)……. sM(t) are equally likely, the optimum
receiver which minimizes the average probability of error is known as a Correlation receiver. The output of the
correlation receiver is an estimate of the input based on (i) the observation vector, (ii) knowledge of the
transmitted signal and (iii) probability of occurrence of input symbols.
The correlation receiver works on Maximum Likelihood(ML) decision and consists of two subsystems namely
the detector and vector receiver.
The detector part of the receiver consists of N bank correlators supplied with orthonormal basis functions
generated locally and these correlators operate on the received signal x(t) for the period 0  t  T to produce
the observation vector x.
The Vector Receiver
The vector receiver is implemented in the form of a maximum likelihood detector that operates on observation
vector x generated by the detector to produce an estimate 𝑚 of the transmitted symbol mi. The N elements of
the observation vector x are multiplied by corresponding N elements of each of the M signal vectors s 1, s2 ….
sM and the resultant products
are summed in the accumulators to form the inner products {(x,sK)} and these are corrected by subtracting
half the signal energy (Ei/2) and finally maximum likely hood decision is employed to get the estimate 𝑚.

𝑁
1
𝑇𝑕𝑒 𝑣𝑒𝑐𝑡𝑜𝑟 𝑥 𝑙𝑖𝑒𝑠 𝑖𝑛𝑠𝑖𝑑𝑒 𝑡𝑕𝑒 𝑑𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑟𝑒𝑔𝑖𝑜𝑛 𝑍𝑖 𝑡𝑕𝑒𝑛 𝑡𝑕𝑒 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒 𝑚 = 𝑥𝑗 𝑠𝑘𝑗 − 𝐸
2 𝑘
𝑗 =1

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 16
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Matched Filter Receiver

A filter which is matched to a known signal (t) of duration T sec, characterized by an impulse response which
is time reversed and delayed version of the input (t) is known as Matched filter. The matched filter is causal
in nature and the noise in the channel is assumed to be White Gaussian noise.
Consider a linear time-invariant filter with impulse response ℎ𝑗(𝑡), the received signal 𝑥(𝑡) operating as input,
the resulting filter output yj(t) is defined by the convolution integral,
∞
yj t = x τ hj t − τ dτ − − − − − −(𝐴)
−∞

Evaluate this integral over the duration of a transmitted symbol, 0 ≤ 𝑡 ≤ 𝑇, then replace the variable τ with t,
T
yj T = 0
x t hj T − t dt − − − − − −(𝐵) 𝑕 𝑡 −→ 𝑇𝑖𝑚𝑒 𝑟𝑒𝑣𝑒𝑟𝑠𝑎𝑙−→ 𝑕 −𝑡
𝑕 −𝑡 − −𝑇𝑖𝑚𝑒𝑠𝑕𝑖𝑓𝑡 𝑏𝑦 𝑇−→ 𝑕 −𝑡 + 𝑇
Consider the detector based on a bank of correlators.
The output of 𝑗𝑡𝑕 correlator is given by,
T
xj = x t Φj t dt − − − − − −(𝐶)
0

By comparing equations (B) & (C), for 𝑦𝑗(𝑇) to be equal to 𝑥𝑗, it must satisfies the following,
hj(T – t) = Φj(t) for 0  t  T and j = 1,2,…M
Equivalently hj(t) = Φj(T –t ) and j = 1,2,…M
From the above described condition it can be concludes as follows
Given a pulse Φ(t) occupying the interval 0  t  T, a LTI filter is said to be matched to Φ(t) if its Impulse
response h(t) satisfies the condition
h(t) = Φ(T – t ) for 0  t  T
The time-invariant filter system defined in this way is known as a Matched Filter. Correspondingly, an optimum
receiver using matched filter in place of Correlators is called Matched Filter receiver.

Questions from previous question papers

 Obtain the maximum likelihood decision rule for the signal detection problem.
Obtain the decision rule for maximum likelihood decoding and explain the correlation receiver.
Explain the correlation receiver using product modulator or integrator and matched filter.
With neat diagrams and necessary equations, explain matched filter receiver.
Explain the matched filter receiver with relevant mathematical theory
This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 17
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

8) A discrete source has 8 symbols m1, m2, ….m8 and the corresponding transmitted signals are s1(t),
s2(t)… ..s8(t). These transmitted signals are equiprobable and the channel is AWGN. The signal vectors
are points in a 3-D Euclidean space with co-ordinates as given below. s1 = (0,0,0); s2 = (0,0,1); s3 = (0,1,0);
𝟏/𝟒
s4 = (0,1,1) s5 = (1,0,0); s6 = (1,0,1); s7 = (1,1,0); s8 = (1,1,1). The received signal vector is 𝐱 = 𝟓/𝟒
𝟑/𝟒
Determine the most likely transmitted signal based on Maximum likelihood decision.
Solution: The dimension of the signal space is N = 3 and M = 2N = 8
The decision rule is
𝑁
1 1
𝑐𝑕𝑜𝑜𝑠𝑒 𝑚 = 𝑚𝑖 𝑖𝑓 𝑦𝑘 = 𝑥𝑗 𝑠𝑗𝑘 − 𝐸𝐾 = 𝑥 𝑇 𝑠𝑘 − 𝐸𝐾 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑓𝑜𝑟 𝑘 = 𝑖
2 2
𝑗 =1

𝐸1 =; 0 J ; 𝐸2 = 1 𝐽 ; 𝐸3 = 1𝐽 ; 𝐸4 = 12 + 12 = 2𝐽; 𝐸5 = 1𝐽; 𝐸6 = 2𝐽; 𝐸7 = 2𝐽; 𝐸8 = 3𝐽

1 1 5 3 0 1
𝑦1 = 𝑥 𝑇 𝑠1 − 𝐸1 = . 0 − 0 =0
2 4 4 4 2
0
1 1 5 3 0 1 1
𝑦2 = 𝑥 𝑇 𝑠2 − 𝐸2 = . 0 − 1 =
2 4 4 4 2 4
1
1 1 5 3 0 1 3
𝑦3 = 𝑥 𝑇 𝑠3 − 𝐸3 = . 1 − 1 =
2 4 4 4 2 4
0
1 1 5 3 0 1
𝑦4 = 𝑥 𝑇 𝑠4 − 𝐸4 = . 1 − 2 =1
2 4 4 4 2
1
1 1 5 3 1 1 1
𝑦5 = 𝑥 𝑇 𝑠5 − 𝐸5 = . 0 − 1 =−
2 4 4 4 2 4
0
1 1 5 3 1 1
𝑦6 = 𝑥 𝑇 𝑠6 − 𝐸6 = . 0 − 2 =0
2 4 4 4 2
1
1 1 5 3 1 1 1
𝑦7 = 𝑥 𝑇 𝑠7 − 𝐸7 = . 1 − 2 =
2 4 4 4 2 2
0
1 1 5 3 1 1 3
𝑦8 = 𝑥 𝑇 𝑠8 − 𝐸8 = . 1 − 3 =
2 4 4 4 2 4
1
Since y4 is the largest, the most likely transmitted signal is m4. If two values are equal
and maximum, the receiver selects any one arbitrarily.
9) For the signal shown below, determine the impulse response of a filter matched
to s(t). Plot the matched filter output as a function of time and determine its peak
value.

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 18
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

Solution:
The impulse response of the matched filter is time-reversed and delayed version
of the input signal, h(t) = s(T-t) and the output of the filter, y(t) = x(t) * h(t).

The peak value of the output is (a2T/4) (refer class notes for detailed solution to the problem)
10) Consider the signal s(t) shown in figure. (a) Determine the impulse response of a filter matched to this
signal and sketch it as a function of time. (b) Plot the matched filter output as a function of time. (c)
What is Peak value of the output?
Solution:
The impulse response of the matched filter is time-reversed and delayed version
of the input signal, h(t) = s(T-t) and
the output of the filter, y(t) = x(t) * h(t).
Given s(t) = +1 for 0 < t < 0.5
-1 for 0.5 < t < 1.

(a) With T = 1, the impulse response h(t) is

h(t) = -1 for 0 < t < 0.5
+1 for 0.5 < t < 1.

b) The output of the filter y(t) is obtained by convolving the input s(t) and the
impulse response h(t).
The corresponding output is shown in the fig.
(c) The peak value of the output is 1.0 unit.

11) A filter is matched to a signal s(t) is given by s(t) = 1 for 0  t  1 and 3  t  4 and s(t) = 0 otherwise.
Find the peak value of the matched filter output and time at which it occurs.

Solution: For a matched filter, the output is equal to the energy of the signal.

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 19
Ravi Shankara.M.N. ECE Dept, SCE Bangalore
Associate Professor ravishankarmn@[Link]

4 1 4
∅𝑜 𝑡 = [𝑠 𝑡 ]2 𝑑𝑡 = 12 𝑑𝑡 + 12 𝑑𝑡 = 2
0 0 3
Peak value of the matched filter output = 2 and it occurs at t = 4 sec.

12)Figure shows a finite energy signal (t). Sketch the impulse response of the optimum
filter matched to (t). Determine value of the output of matched filter for t = T assuming
noise is zero and input is (t).

hopt(t) = (t –T)

Since there is no noise, the filter input x(t) = (t) and its output y(t) = 0(t). Also the correlator output and
matched filter outputs are identical at t = T
𝑇 𝑇 𝑇
2
𝑦 𝑡 𝑡=𝑇 =𝑦 𝑇 = 𝑥 𝜏 ∅ 𝜏 𝑑𝜏 = [∅ 𝜏 ] 𝑑𝜏 = 𝐴2 𝑑𝜏 = 𝐴2 𝑇
0 0 0
0 𝑡<0
𝐴2 𝑇 0≤𝑡≤𝑇
Aliter: 𝑦 𝑡 = 𝑕𝑜𝑝𝑡 𝑡 ∗∅ 𝑡 = at t = T  y(T) = A2T
𝐴2 𝑇 𝑇 ≤ 𝑡 ≤ 2𝑇
0 𝑡 > 2𝑇

This is a supporting material for the SCE students. Any use for commercial purpose has to be permitted by the author ©Ravi Shankara.M.N. Page 20