DSPRelated.com
Forums

Trouble in generate SER in OFDM

Started by ericknightX February 26, 2007
hi all ,

Recently , I have generated a OFDM signal , but meanwhile I have noticed
that my Analytical & Computer Simulation of my SER are totally difference,
Moreover , the Numbers of Errors in my signal also quite large, therefore i
am wondering whether there are some Errors in my signal detection or the
progress of generate OFDM signal. Actually I should generate channel
estimation of OFDM signal , but i have stuck at this part quite long time
, therefore i am really need help from all of you with my sincere.

Thank you.

Eric

*********************************************************************************************
%generate 64 QAM signal 
aqam=[-7 -5 -3 -1 1 3 5 7];
A=repmat(aqam,8,1);
B=flipud(A');
konst_qam=A+j*B;
konst_qam=konst_qam(:);

qam=konst_qam(randint(1024,1,64)+1);  %Random Generate 1024 symbol of 64
QAM signal


F=ifft(eye(1024));                   %Convert the signal from Frequency
domain to Time domain

Fs=F*qam;

%Fs1=repmat(Fs,1,100);


C=zeros(1024,1);
R=zeros(1,1024);


for(i=1:6)
h(1,i)=rand+j*rand;
end

R(1,1:6)=h;
C(1:6,1)=h';

H=toeplitz(C,R);

x=H*Fs;

%generate Noise                                          
n=randn(size(x))+j*randn(size(x));            %randn generates Gaussian
distributed noise
SNR=20;                                                %SNR-ratio

p=std(x)/(std(n)*10^(SNR/20));                     

   
%add noise to signal
x1=x+n*p;                      

%Receiver
F1=fft(eye(1024));  %Convert the Signal back from Time Domain to Frequency
Domain.

y1=F1*x1;


D=F1*H*F;

%Equalization
s_final1=(inv(D))*y1;

%Symbol Detection
s_f1=repmat(s_final1,1,64).';
konst_block=repmat(konst_qam,1,1024);
distance1=abs(s_f1-konst_block);
[z1,ind_2]=min(distance1);
qam_det1=konst_qam(ind_2);


%Theoretical SER
d=2;
sigma=std(real(n*p));
Q=0.5*erfc(d/(sqrt(2)*2*sigma));
sep_theo1=3.5*Q-3.0625*Q^2;


%Simulated SER
error=sum(qam~=qam_det1);
sep=error/1024;
On Feb 26, 9:42 am, "ericknightX" <wil...@gmail.com> wrote:
> hi all , > > Recently , I have generated a OFDM signal , but meanwhile I have noticed > that my Analytical & Computer Simulation of my SER are totally difference, > Moreover , the Numbers of Errors in my signal also quite large, therefore i > am wondering whether there are some Errors in my signal detection or the > progress of generate OFDM signal. Actually I should generate channel > estimation of OFDM signal , but i have stuck at this part quite long time > , therefore i am really need help from all of you with my sincere. > > Thank you. > > Eric >
Eric, Your code is quite confusing to follow. I'm not entirely certain what the F=ifft(eye(.....))) is all about? When you say that you have generated an OFDM signal have you actually veriifed the performance with the theoretical AWGN, without sticking it through a channel? Have a look at this link for a freq flat fading to get some ideas and then come back with questions. http://groups.google.com/group/comp.dsp/browse_frm/thread/febe0248d368fbe8/0f61c99347e8de3a?lnk=gst&q=Maybe+a+short+example+will+help%3F+&rnum=1#0f61c99347e8de3a col
Thanks col , I have checked your code , & i think i have get some idea.
I am wondering that maybe i have the same problem with Shahin. Actually i
have skipped the step in insert cyclic prefix & remove cyclic prefix in my
code , since i assume the process is ideal. Therefore there is no IBI.
Meanwhile , there is only the AWGN might affect my SER , but i really no
idea , why the theoretical SER & simulate SER is totally difference , &
the number of error is too large.   

the  F=ifft(eye(.....))) & direct ifft the transmitted signal actually is
the same , since i am try to do in block transmission & equalization.

And I am referring this Journal
http://www.ece.iastate.edu/~zhengdao/group/publications/papers/wagi00m.pdf

Regards 

Eric