Rayleigh one tap fading channel

[anta]

Hi all. I am doing ofdm for my diploma thesis and i am facing some problems with rayleigh fading channel....Apparently there's a mistake in my code that i can't find...Any suggestion will be helpful...thank you in advance
Here's my code:


%% DEFINE PARAMETERS %%
%%%%%%%%%%%%%%%%%%%%%%%%%%


M=4;%size of signal constellation

L=1024; % length of data

cp=32;%cyclic prefix


% maximum value of SNR to be simulated (in dB)
max_SNR =25;


%maximum number of iterations
max_ITER =1000;

bits_in_error= zeros(1,max_SNR);
symbols_in_error=zeros(1,max_SNR);




% generate impulse response h of the channel
% independent Rayleigh fading
h=1.*randn(1,1);


%Rayleigh factor
%follows gaussian distribution with mean zero and
%standard deviation 1
%h2=1.*randn(1,K+1);
%energy of the rayleigh factor
%Eh2=sum(abs(h1).^2)/(L);

% h1=heaviside(0);

%h=h2.*h1;

% find frequency response
H=fft(h,L);

% clear figure from previous runs
clf;

%%%%%%%%%%%%%%%%%%%%%%%%%
%% OFDM TRANSMITTER %%
%% %%
%%%%%%%%%%%%%%%%%%%%%%%%%

for iter=1:max_ITER;

%%Modulation
%Modulate using 16-QAM.

%generates random symbols
x=randint(1,L,M);

%qam modulation
y=qammod(x,M);

%normalization according to M-QAM signal constellation
z=y*(1/2)*sqrt(6/(M-1));

%signal energy
Ez=sum(abs(z).^2)/L;

%%IFFT
Z = ifft(z);
%signal energy after the ifft
EZ=sum(abs((Z).^2))/L;

%apply the fourier transform factor
Z1 = sqrt(L)*Z;
%signal energy
EZ1=sum(abs((Z1).^2))/L;

%cyclic prefix

%apply cyclic prefix
Z2 = [Z1(L-cp+1:L) Z1];
%energy of the new signal(with cyclic prefix)
EZ2=sum(abs((Z2.^2))/(L+cp));


%transmitted signal after rayleigh factor applied
Y=Z2*h;





%%%%%%%%%%%%%%%%%%%%%%
%% GAUSSIAN CHANNEL %%
% %%
%%%%%%%%%%%%%%%%%%%%%%

%given SNR in dB
for SNR_dB=1:1:max_SNR
% convert snr
snr=10^(SNR_dB/10);
%find the noise for every SNR
n=sqrt(1/(2*snr))*(randn(1,(L+cp))+j*randn(1,(L+cp)));
%noise energy
En=sum(abs(n.^2))/L;

%find y_noisy for every SNR
ynoisy=Y+n;
%ynoisy=0;

%find the signal energy
Eynoisy=sum(abs(ynoisy.^2))/(L+cp);

%undo cyclic prefix
ynoisy_cut=ynoisy(cp+1L+cp));

%ynoisy_cut(without cyclic prefix) energy
Eynoisy_cut=sum(abs(ynoisy_cut.^2))/L;

%signal normalization
Y1=ynoisy_cut./sqrt(L);

%fast fourier transform
Y2=fft(Y1);





%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%% OFDM RECEIVER %%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%% DEMODULATION %%
%% CHECK y_demod SHOULD BE THE SAME WITH THE TRANSMITTED x when there is no noise applied %%%
%% CHECK y_demod WHEN GAUSSIAN NOISE IS APPLIED%%%%%


%%% undo the rayleigh factor
Y3=Y2./H;


%%% M-QAm normalization
y4=ynoisy_cut*2/sqrt(6/(M-1));

%%%demodulated signal when noise is applied
y_demod=qamdemod(y4,M);

%%difference between transmitted and received signal when noise is applied
y_dif=abs(y_demod-x);



%%%% convert to binary%%%%%%

x_binary=dec2bin(x);
y_demod_binary=dec2bin(y_demod);



%%the difference should be zero when the channel is zero
y_binary=sum(abs(dec2bin(x(1:L))-dec2bin(y_demod(1:L))),2);

%find the total bits in error
total_bits_in_error = sum(y_binary);

%find total symbols in error
total_symbols_in_error=sum(y_dif);

%bits in error for every SNR
bits_in_error(SNR_dB) = bits_in_error(SNR_dB) + total_bits_in_error;

%symbols in error for every SNR
symbols_in_error(SNR_dB)=symbols_in_error(SNR_dB)+total_symbols_in_error;


%%binary difference
%%the difference should be zero when the channel is zero
y_binary_total= sum(sum(dec2bin(x(1:L))-dec2bin(y_demod(1:L)),1),2)/L;

%%shows the binary difference between the symbols due to the noise%%
y_dif_binary=y_demod_binary-x_binary;


end

%end for iter=1:max_ITER
end

%%find bit error rate
BER = bits_in_error/(log2(M)*L)/max_ITER;

%%find symbol error rate
SER=symbols_in_error/(M*L)/max_ITER;


%%BER plot
semilogy(1:max_SNR,BER);

 

[anta]

any help so far? it's urgent...thank you in advance

 

[Communications_Engineer]

visit www.dsplog.com, search for OFDM and modify the code to suit your work.

 

[anta]

I looked through the site but it didn't help me..
I think that the problem is in the choice of rayleigh factor...
When the channel is flat (aka one tap) what is the rayleigh factor,h..
Since the channel is flat fading h should be the same for every ofdm symbol...

If L is the length of data then h is 1/sqrt(2)*randn(1,L)?

any help?
thanx

 

[lemontree45]

I have been working on the Rayleigh fading channel in MIMO.

If the no: of transmitters- nt
and no: of receivers - nr

Acc to Rayleigh flat fading channel,

Then channel matrix h= 1/sqrt(nt) * randn(nr,nt)+ sqrt(-1)*randn(nr,nt)

I am not sure whether you are looking for this one.

Regards

 

[anta]

@Lemontre45 thank you for you reply. I've tried simulating according to your suggestion...However the plot is rather similar to the one of an awgn channel...\

So if N is the num of subcarriers
num of taps=1 with one transmitter and one receiver i still wonder which is the right rayleigh factor...

Any suggestion will be helpful
Thanx in advance