EC4019 : DIGITAL COMMUNICATION LAB

REPORT
Experiment 3

Performance Evaluation of BPSK

Modulation
with Error Control Coding
Submitted By
Pal Joyanusha (B110595EC)
Ramavathu Kavitha (B110681EC)
Rameswaree Ganesh (B110083EC)
Rassal Shaji (B110744EC)

Objectives

Simulate a digital communication system that uses coherent BPSK with antipodal signaling and BCH code in the presence of Rayleigh fading channel and
white noise and plot BER curves.
Implementation of BPSK modulation with Error Control Coding
Synthesizing a frequency flat Rayleigh fading channel and plot its PDF.
Understanding the BER performance of uncoded digital modulation scheme
in the presence of Rayleigh fading channel and AWGN channel
Implementation of (15,7,5) BCH encoding.
Analysis of BER performance of the modulation scheme with BCH code
in the presence of Rayleigh Fading Channel
Plotting of Ideal and Practical BER curves
1

 Find the Eb/N 0 required for achieving a BER performance of 103 in all
comparisons

Theory

2.1

Rayleigh Fading Channel:

The delays associated with different signal paths in a multipath fading

channel change in an unpredictable manner and can only be characterized statistically. When the number of paths is very large, Central Limit Theorem can
be invoked in order to model the time - varying impulse response of the channel
as a circularly symmetric complex valued Gaussian random process. The channel is said to be Rayleigh fading channel if the impulse response is modeled
as a zero - mean complex Gaussian process.

Figure 1: pdf of Rayleigh Fading Channel

Channel Model
A circularly symmetric complex Gaussian random variable is of the form,
Z = X + jY

(1)

where X and Y are zero - mean independent and identically distributed

(iid) Gaussian random variable. For a circularly symmetric complex random variable Z, we have:
E[Z] = E[ej Z] = ej E[Z]
2

(2)

The statistics of a circularly symmetric complex Gaussian random variable

is completely specified by its variance.
2 = E[Z 2 ]

(3)

The probability density of the magnitude |Z| is given by:

p(z) =

z z22
e 2 , z 0
2

(4)

This is the Rayleigh random variable. This model is reasonable for an

environment where there are a large number of reflectors.
BER for BPSK modulation in a Rayleigh Fading Channel:
The probability of error in BPSK modulation scheme in the presence of
AWGN channel is given by:
r
Eb
1
1

) = erf c( )
(5)
Pb = erf c(
2
N0
2
In the presence of the Rayleigh Fading Channel, the effective bit energy to
2
b
. Hence, to find the bit error probability is determined
noise ratio is |hN|E
0
by evaluating the conditional probability density function Pb|h over the
probability density function of :
Z
1

Pb =
erf c( )p()d
(6)
2
0
This can be reduced to yield the following result:
v
u Eb
u
1
Pb = (1 t EbN0 )
2
N0 + 1

2.2

(7)

Error Control Coding

Error detection and correction or error control are techniques that enable
reliable delivery of digital data over unreliable communication channels.
A code is a mapping that takes a sequence of information symbols and produces a (larger) sequence of code symbols so as to be able to detect/correct
errors in the transmission of the symbols.
The simplest class of codes is the class of binary linear block codes. Here
each vector of k information bits xi = [xi,1 ... xi,k ] is mapped to vector of
n code bits ci = [ci,1 ... xi,n ], with n >k.

2.3

BCH Code

The BCH codes form a class of cyclic error-correcting codes that are constructed using finite fields.
There is a precise control over the number of symbol errors correctable by
the code. In particular, it is possible to design binary BCH codes that can
correct multiple bit errors.
They are easy to decode, using the algebraic Syndrome method.For any
positive integers m >= 3 and t < 2m1 , there exists a binary code with
block length n = 2m 1, number of parity-check digits n k <= mt with
a minimum distance dmin >= 2t + 1, called the t-error correcting BCH
codes. The generator polynomial of this code is specified in terms of its
roots from the Galois field GF (2m ).

Implementation

3.1

Rayleigh Fading Channel - System Model

The received signal in the Rayleigh fading channel is of the form:

y = hx + n
where,
y is the
h is the
x is the
n is the

(8)

received symbol
complex scaling factor corresponding to Rayleigh fading channel
transmitted symbol
Additive White Gaussian Noise (AWGN)

In this experiment, we have assumed that the Rayleigh fading channel has
only two multipath components, X(t) and Y(t).
Rayleigh fading can be obtained from zero - mean complex Gaussian processes (X(t) and Y(t)).
In order to get the Rayleigh fading channel envelope, we add the two Gaussian random variables and take its absolute value (equivalent to squaring
and adding the two random variables and then taking its square root).
This gives a Rayleigh distributed process. The phase follows uniform distribution.

3.2

BCH Code

We are required to implement a (15, 7, 5) BCH code,which is 2- error

correcting.

 We first construct the generator polynomial for this code. Let be a

primitive element of GF(2m ). The generator polynomial g(x) of the terror-correcting code of length 2m 1 is the lowest degree polynomial over
GF(2) which has , 2 ,3 , ...2t and their conjugates as its roots.
If i (x) is the minimal polynomial of i , g(x) = LCM (1 (x), 3 (x), ...., 2t1 (x))
We have the H matrix as

The entries of H are elements from GF (2m ). Each element in GF (2m ) can
be represented by a m-tuple over GF (2). Then we get the binary parity
check matrix for the code. Therefore H for our above problem is :

BCH decoding
For a t error correcting BCH code, the syndrome is a 2t-tuple, S =
(S1 , S2 , ..., S2t ) = rH T , where r is the received code vector.
The ith component of the syndrome is Si = r(i ).
The decoding procedure for 2-error correcting code is as follows:
Compute S1 and S3 .
If S1 = 0 and S3 = 0 , then assume no error.
If S1 6= 0 and S3 = S13 , then assume 1 error. Error location is i1
satisfying X1 = i1 .

 If S1 6= 0 and S3 6= S13 , then assume 2 errors. Error location numbers

S +S 3
are the zeros in GF (2m ) of R (x), where R (x) = x2 + S1 x + 3S1 1
Error locations are i satisfying X = i .
For all other conditions, more than two errors have occurred and
these are uncorrectable.

MATLAB Code

clear all;ctr=0;
n=15; k=7; t=2;
N=k*1000;
ebyn0db = 0 : 2 : 20;
ebyn0 = 10.^(ebyn0db/10);
ip=rand(1,N)>0.5;
si=2*ip-1;
g=[1 0 0 0 1 0 1 1 1 0 0 0 0 0 0];
gen=[];
for i=0:k-1
gg= circshift(g,i);
gen=[gen gg];
end
gen=gen;
a=0:15;
b=gf(a,4);
alpha=b(3);
A=[];
for i= 0: 15
aa= alpha^i;
A=[A aa];
end
s=reshape(ip,N/k,k);
coded = mod(s*gen,2);
code = reshape(coded,1,(N/k)*n);
BER = zeros(1,length(ebyn0db));
nn = randn(1,N) + 1i*randn(1,N);
nn1 = randn(N/k,n) + 1i*randn(N/k,n);
for ii=1: length(ebyn0db)
rxd1 = si + 10^(-ebyn0db(ii)/20)*nn;
rxd = real(rxd1)>0;
coded_n = (2*coded-1) + 10^(-ebyn0db(ii)/20)*nn1;
coded_n = real(coded_n) >0;
rcv=[];
for i=1:N/k
pos=find(coded_n(i,:));

for m=1:length(pos);
if m==1
s1= A(pos(m));s3= A(pos(m))^3;
else
s1=s1+A(pos(m));s3=s3+(A(pos(m)))^3;
end
end
s1d=double(s1.x);
s11=de2bi(s1d,2*t);
s3d=double(s3.x);
if (s1==b(1)) && (s3==b(1))
rcv=[rcv coded_n(i,:)];
elseif (s1 ~= b(1)) && (s3 == s1^3)
e=s11;
im = find(A(1:15) == s1);
rr=xor(coded_n(i,:),[zeros(1,im-1) 1 zeros(1,n-im)]);
rcv=[rcv rr];
elseif (s1 ~= b(1)) && (s3 ~= s1^3)
pr=(s3+s1^3)/s1;
er=[];
for f=1:length(b)
if (b(f))^2 + s1*b(f) + pr == b(1)
er = [er b(f)];
end
end
if isempty(er)
rcv=[rcv coded_n(i,:)]; ctr=ctr+1;
else
e1 = find(A(1:15) == er(1));e2 = find(A(1:15) == er(2));
if(e1>e2)
xx= [zeros(1,e2-1) 1 zeros(1,e1-1-e2) 1 zeros(1,n-e1)];
elseif(e1<e2)
xx= [zeros(1,e1-1) 1 zeros(1,e2-1-e1) 1 zeros(1,n-e2)];
else
xx= [zeros(1,e1-1) 1 zeros(1,n-e1-1)];
end
rr= xor(coded_n(i,:),xx);
rcv=[rcv rr];
end
else
rcv=[rcv coded_n(i,:)];
end
end
rcv=reshape(rcv, n, N/k);
y= length(find(xor(rcv,coded)));
BER(ii)= y/(n*(N/k));
7

y1= length(find(xor(rxd,ip)));
thber(ii) = y1/N;
end
%thber = 0.5*erfc(sqrt(10.^(ebyn0db/10)));
%Rayleigh channel
variance=0.2;
X = randn(1, N)+1i*randn(1,N);
Y = randn(1, N)+1i*randn(1,N);
R = sqrt(variance*(real(X).^2 + real(Y).^2));
range = 0:0.1:3;
%Get histogram values and approximate it to get the pdf curve
h = hist(R, range);
approxPDF = h/(0.1*sum(h)); %Simulated PDF from the x and y samples
%Theoritical PDF from the Rayleigh Fading equation
theoretical = (range/variance).*exp(-range.^2/(2*variance));
figure;
plot(range, approxPDF,b*, range, theoretical,r);
title(Simulated and Theoretical Rayleigh PDF for variance = 0.5)
legend(Simulated PDF,Theoretical PDF)
xlabel(r);
ylabel(P(r));
grid;
BERr=[]; BERu=[];
nn = randn(N/k,n) + 1i*randn(N/k,n);
hn = randn(N/k,n) + 1i*randn(N/k,n);
for ii=1:length(ebyn0db)
rxd1 = Y.*si + 10^(-ebyn0db(ii)/20)*X;
rxd11 = rxd1./Y;
rxd = real(rxd11)>0;
coded_nr1 = hn.*(2*coded-1) + 10^(-ebyn0db(ii)/20)*nn;
coded_nr11 = coded_nr1./hn;
coded_nr = real(coded_nr11) >0;
rcv=[];
for i=1:N/k
pos=find(coded_nr(i,:));
for m=1:length(pos);
if m==1
s1= A(pos(m));s3= A(pos(m))^3;
else
s1=s1+A(pos(m));s3=s3+(A(pos(m)))^3;
end
end
8

s1d=double(s1.x);
s11=de2bi(s1d,2*t);
s3d=double(s3.x);
if (s1==b(1)) && (s3==b(1))
rcv=[rcv coded_n(i,:)];
elseif (s1 ~= b(1)) && (s3 == s1^3)
e=s11;
im = find(A(1:15) == s1);
rr=xor(coded_nr(i,:),[zeros(1,im-1) 1 zeros(1,n-im)]);
rcv=[rcv rr];
elseif (s1 ~= b(1)) && (s3 ~= s1^3)
pr=(s3+s1^3)/s1;
er=[];
for f=1:length(b)
if (b(f))^2 + s1*b(f) + pr == b(1)
er = [er b(f)];
end
end
if isempty(er)
rcv=[rcv coded_nr(i,:)];ctr=ctr+1;
else
e1 = find(A(1:15) == er(1));e2 = find(A(1:15) == er(2));
if(e1>e2)
xx= [zeros(1,e2-1) 1 zeros(1,e1-1-e2) 1 zeros(1,n-e1)];
elseif(e1<e2)
xx= [zeros(1,e1-1) 1 zeros(1,e2-1-e1) 1 zeros(1,n-e2)];
else
xx= [zeros(1,e1-1) 1 zeros(1,n-e1-1)];
end
rr= xor(coded_nr(i,:),xx);
rcv=[rcv rr];
end
else
rcv=[rcv coded_nr(i,:)];
end
end
rcv=reshape(rcv, n, N/k);
y= length(find(xor(rcv,coded)));
BERr(ii)= y/(n*(N/k));
y1= length(find(xor(rxd,ip)));
BERu(ii) = y1/N;
end
thray = 0.5.*(1-sqrt(ebyn0./(ebyn0+1)));
figure;
semilogy(ebyn0db,BER,md-,Linewidth,3);hold on;
semilogy(ebyn0db,thber,b.-,Linewidth,3);hold on;
9

semilogy(ebyn0db,BERr,rs-,Linewidth,3);hold on;
semilogy(ebyn0db,BERu,go-,Linewidth,3);hold on;
semilogy(ebyn0db,thray,b.-,Linewidth,3);grid on;
axis([0 20 10^-5 0.5]);
title(BER of BPSK with BCH coding in Rayleigh channel); legend( BCH CODED-AWGN,THEORETI
ylabel(BER);

MATLAB Simulation

The PDF of AWGN channel was plotted as in the figure below:

The BER plots of BPSK modulation scheme- theoretical and practical scenarios
with and without BCH error coding and Rayleigh fading channel were obtained

10

as follows:

Results
BPSK modulation scheme was implemented in the presence of AWGN
noise and Rayleigh fading channel
BCH code was implemented for error detection and correction of the BPSK
modulated signal.
BER plots for ideal and practical scenarios were obtained for BPSK modulation scheme, BPSK modulation in the presence of Rayleigh fading channel, and also when implementing error detection and correction using BCH
code.
The performance of the various schemes were analyzed.
The required SNR for BER of 103 was obtained as follows:
11

 For the theoretical Rayleigh channel as well as transmission without

error control coding for the Rayleigh channel, the required SNR is
over 20dB.
The theoretical AWGN channel requires an SNR of about 10dB. In
this simulation, the BCH coded Rayleigh channel signal also requires
a similar SNR.
There is a significant reduction in the SNR required for the same
BER rate for the BCH coded AWGN channel, an SNR of only around
6dB. This results in a tremendous improvement in performance of the
system.

References
[1] Simon Haykin, Communication Systems, John Wiley & Sons, Inc., 4th Edition
[2] Ken Gentile, The care and feeding of digital, pulse-shaping filters
[3] Wikipedia, the Free Encyclopedia
http://en.wikipedia.org/wiki/
[4] Gaussian Waves
http://www.gaussianwaves.com/

12