Dear sir ,
I am trying to create a matlab model of HARQ ( all types) using LDPC (llr based) as fec . 
I wrote a  code using matlab library (fec.ldpcenc-fec.ldpcdec) , but the BER curve is somehow different than the other literature works (the drop is around 3-5 dB SNR ,not between 0-1dB or 1-1.5dB ,also for high ldpc iteration(50) numbers bottom line of the curve is only 10^(-0.9)) . 
Could you explain me what is wrong with my approach ,did i miss something (maybe sigma calculations?) ?

R=1/2;
t_er=zeros(1,21);
er=zeros(1,21);
SNRdB=-1:0.20:3;
trial=70;

      enc = fec.ldpcenc;  % Construct a default LDPC encoder object
      dec = fec.ldpcdec;  % Construct a companion LDPC decoder object
      dec.DecisionType = 'Hard decision';     % Set decision type
      dec.OutputFormat = 'Information part';  % Set output format
      dec.NumIterations = 50;                 % Set number of iterations
      dec.DoParityChecks = 'Yes';  % Stop if all parity-checks are satisfied

for z=1:21%z=1:length(SNRdB)
    
    for run=1:trial
      msg = randi([0 1],1,enc.NumInfoBits);   % Generate a random binary message
      codeword = encode(enc,msg);          % Encode the message

      % Construct a BPSK modulator object
      modObj = modem.pskmod('M',2,'InputType','Bit');

      % Modulate the signal (map bit 0 to 1 + 0i, bit 1 to -1 + 0i)
      modulatedsig = modulate(modObj, codeword);

       %dB to actual SNR 
      SNR(z)=10^(SNRdB(z)/10);
      ebno_c(z)=SNR(z)*R;                    %Eb/No for coded signal!!!!!!!!!!!!!!<----!

      % Noise parameters
      var(z) = 1 / (2 * ebno_c(z));%No/2 !!!!!!!!!!!!!!!

      %or-->sigma = sqrt((10^(-SNRdB(z)/10))/(2*R));
      
      % Transmit signal through AWGN channel
      receivedsig = modulatedsig+sqrt(var(z)).*randn(1,enc.blocklength);

      %or-->receivedsig = modulatedsig+sigma.*randn(1,enc.blocklength);
      
      
      % Construct a BPSK demodulator object to compute log-likelihood ratios
      demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',var(z));

      %or-->demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',sigma^2);
      
      
      % Compute log-likelihood ratios (AWGN channel)
      llr = demodulate(demodObj, receivedsig);

      % Decode received signal
      decodedmsg = decode(dec, llr);

      % Actual number of iterations executed
      iter =  ...
           num2str(dec.ActualNumIterations);

      % Number of parity-checks violated
      parity_checks_violated = num2str(sum(dec.FinalParityChecks));
           
      % Compare with original message
      er(z) = (nnz(decodedmsg-msg));
       t_er(z)=t_er(z)+er(z);
       
    end
    av_er(z)=t_er(z)/trial;
    ber(z)=av_er(z)/32400;       %for R=1/3-->ber(z)=av_er(z)/21600;
end
semilogy(SNRdB,ber,'--rs','LineWidth',2,'MarkerEdgeColor','r')


Could you help me  to get LDPC(DVB-S2) Eb/No-BER by using  matlab library functions??
THANK YOU IN ADVANCE

       

for R=1/3 BER graphs i used :

      R=1/3;
      H = dvbs2ldpc(R);
      enc = fec.ldpcenc(H);  
      dec = fec.ldpcdec(H);  

(enc=fec.ldpcenc; %creates 1/2 Rate DVB-S2 LDPC encoder object.)

"on on" <matidoisaduende@yahoo.com> wrote in message <ic6fo0$pqp$1@fred.mathworks.com>...
> Dear sir ,
> I am trying to create a matlab model of HARQ ( all types) using LDPC (llr based) as fec . 
> I wrote a  code using matlab library (fec.ldpcenc-fec.ldpcdec) , but the BER curve is somehow different than the other literature works (the drop is around 3-5 dB SNR ,not between 0-1dB or 1-1.5dB ,also for high ldpc iteration(50) numbers bottom line of the curve is only 10^(-0.9)) . 
> Could you explain me what is wrong with my approach ,did i miss something (maybe sigma calculations?) ?
> 
> R=1/2;
> t_er=zeros(1,21);
> er=zeros(1,21);
> SNRdB=-1:0.20:3;
> trial=70;
> 
>       enc = fec.ldpcenc;  % Construct a default LDPC encoder object
>       dec = fec.ldpcdec;  % Construct a companion LDPC decoder object
>       dec.DecisionType = 'Hard decision';     % Set decision type
>       dec.OutputFormat = 'Information part';  % Set output format
>       dec.NumIterations = 50;                 % Set number of iterations
>       dec.DoParityChecks = 'Yes';  % Stop if all parity-checks are satisfied
> 
> for z=1:21%z=1:length(SNRdB)
>     
>     for run=1:trial
>       msg = randi([0 1],1,enc.NumInfoBits);   % Generate a random binary message
>       codeword = encode(enc,msg);          % Encode the message
> 
>       % Construct a BPSK modulator object
>       modObj = modem.pskmod('M',2,'InputType','Bit');
> 
>       % Modulate the signal (map bit 0 to 1 + 0i, bit 1 to -1 + 0i)
>       modulatedsig = modulate(modObj, codeword);
> 
>        %dB to actual SNR 
>       SNR(z)=10^(SNRdB(z)/10);
>       ebno_c(z)=SNR(z)*R;                    %Eb/No for coded signal!!!!!!!!!!!!!!<----!
> 
>       % Noise parameters
>       var(z) = 1 / (2 * ebno_c(z));%No/2 !!!!!!!!!!!!!!!
> 
>       %or-->sigma = sqrt((10^(-SNRdB(z)/10))/(2*R));
>       
>       % Transmit signal through AWGN channel
>       receivedsig = modulatedsig+sqrt(var(z)).*randn(1,enc.blocklength);
> 
>       %or-->receivedsig = modulatedsig+sigma.*randn(1,enc.blocklength);
>       
>       
>       % Construct a BPSK demodulator object to compute log-likelihood ratios
>       demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',var(z));
> 
>       %or-->demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',sigma^2);
>       
>       
>       % Compute log-likelihood ratios (AWGN channel)
>       llr = demodulate(demodObj, receivedsig);
> 
>       % Decode received signal
>       decodedmsg = decode(dec, llr);
> 
>       % Actual number of iterations executed
>       iter =  ...
>            num2str(dec.ActualNumIterations);
> 
>       % Number of parity-checks violated
>       parity_checks_violated = num2str(sum(dec.FinalParityChecks));
>            
>       % Compare with original message
>       er(z) = (nnz(decodedmsg-msg));
>        t_er(z)=t_er(z)+er(z);
>        
>     end
>     av_er(z)=t_er(z)/trial;
>     ber(z)=av_er(z)/32400;       %for R=1/3-->ber(z)=av_er(z)/21600;
> end
> semilogy(SNRdB,ber,'--rs','LineWidth',2,'MarkerEdgeColor','r')
> 
> 
> Could you help me  to get LDPC(DVB-S2) Eb/No-BER by using  matlab library functions??
> THANK YOU IN ADVANCE
> 
>        

Any luck getting an answer on this problem?


"on on" <matidoisaduende@yahoo.com> wrote in message <ic6fo0$pqp$1@fred.mathworks.com>...
> Dear sir ,
> I am trying to create a matlab model of HARQ ( all types) using LDPC (llr based) as fec . 
> I wrote a  code using matlab library (fec.ldpcenc-fec.ldpcdec) , but the BER curve is somehow different than the other literature works (the drop is around 3-5 dB SNR ,not between 0-1dB or 1-1.5dB ,also for high ldpc iteration(50) numbers bottom line of the curve is only 10^(-0.9)) . 
> Could you explain me what is wrong with my approach ,did i miss something (maybe sigma calculations?) ?
> 
> R=1/2;
> t_er=zeros(1,21);
> er=zeros(1,21);
> SNRdB=-1:0.20:3;
> trial=70;
> 
>       enc = fec.ldpcenc;  % Construct a default LDPC encoder object
>       dec = fec.ldpcdec;  % Construct a companion LDPC decoder object
>       dec.DecisionType = 'Hard decision';     % Set decision type
>       dec.OutputFormat = 'Information part';  % Set output format
>       dec.NumIterations = 50;                 % Set number of iterations
>       dec.DoParityChecks = 'Yes';  % Stop if all parity-checks are satisfied
> 
> for z=1:21%z=1:length(SNRdB)
>     
>     for run=1:trial
>       msg = randi([0 1],1,enc.NumInfoBits);   % Generate a random binary message
>       codeword = encode(enc,msg);          % Encode the message
> 
>       % Construct a BPSK modulator object
>       modObj = modem.pskmod('M',2,'InputType','Bit');
> 
>       % Modulate the signal (map bit 0 to 1 + 0i, bit 1 to -1 + 0i)
>       modulatedsig = modulate(modObj, codeword);
> 
>        %dB to actual SNR 
>       SNR(z)=10^(SNRdB(z)/10);
>       ebno_c(z)=SNR(z)*R;                    %Eb/No for coded signal!!!!!!!!!!!!!!<----!
> 
>       % Noise parameters
>       var(z) = 1 / (2 * ebno_c(z));%No/2 !!!!!!!!!!!!!!!
> 
>       %or-->sigma = sqrt((10^(-SNRdB(z)/10))/(2*R));
>       
>       % Transmit signal through AWGN channel
>       receivedsig = modulatedsig+sqrt(var(z)).*randn(1,enc.blocklength);
> 
>       %or-->receivedsig = modulatedsig+sigma.*randn(1,enc.blocklength);
>       
>       
>       % Construct a BPSK demodulator object to compute log-likelihood ratios
>       demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',var(z));
> 
>       %or-->demodObj = modem.pskdemod(modObj,'DecisionType','LLR','NoiseVariance',sigma^2);
>       
>       
>       % Compute log-likelihood ratios (AWGN channel)
>       llr = demodulate(demodObj, receivedsig);
> 
>       % Decode received signal
>       decodedmsg = decode(dec, llr);
> 
>       % Actual number of iterations executed
>       iter =  ...
>            num2str(dec.ActualNumIterations);
> 
>       % Number of parity-checks violated
>       parity_checks_violated = num2str(sum(dec.FinalParityChecks));
>            
>       % Compare with original message
>       er(z) = (nnz(decodedmsg-msg));
>        t_er(z)=t_er(z)+er(z);
>        
>     end
>     av_er(z)=t_er(z)/trial;
>     ber(z)=av_er(z)/32400;       %for R=1/3-->ber(z)=av_er(z)/21600;
> end
> semilogy(SNRdB,ber,'--rs','LineWidth',2,'MarkerEdgeColor','r')
> 
> 
> Could you help me  to get LDPC(DVB-S2) Eb/No-BER by using  matlab library functions??
> THANK YOU IN ADVANCE
> 
>