In Turbo Encoder/decode if we give encoder output to decoder (without adding noise) we are getting errors

Question

soujanya velugu 2015-11-13

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/254617-in-turbo-encoder-decode-if-we-give-encoder-output-to-decoder-without-adding-noise-we-are-getting-e

回答： Jakub Hoferek 2018-5-10

Ideally if we give encoder output to decoder (without adding noise) then we are supposed to get zero bit errors right? but in the following code we are getting BER as non-zero. Can any one help with this?

hTEnc  = comm.TurboEncoder('TrellisStructure', poly2trellis(4, [13 15 ], 13), 'InterleaverIndices', intrlvrIndices);
hTDec  = comm.TurboDecoder('TrellisStructure', poly2trellis(4,[13 15 ], 13), 'InterleaverIndices', intrlvrIndices,'NumIterations', 4);
          data = randi( [0 1], 1024, 1);
          encodedData = step(hTEnc, data);
          receivedBits  = step(hTDec, encodedData );

2 个评论
显示无隐藏无

Jakub Hoferek 2018-5-4

I have also this problem. Could anyone help us? .. Or give some tips.

With (4,[13 15 17], 13) it work OK. But with (4,[13 15 ], 13) is about 20% errors in output BITs.

Jakub Hoferek 2018-5-4

在 MATLAB Online 中打开

frmLen = 1024;
  rng default
  intrlvrIndices = randperm(frmLen);
  obj.hTEnc = comm.TurboEncoder('TrellisStructure',poly2trellis(4, ...
      [13 15 ],13),'InterleaverIndices',intrlvrIndices);
  obj.hTDec = comm.TurboDecoder('TrellisStructure',poly2trellis(4, ...
      [13 15 ],13),'InterleaverIndices',intrlvrIndices, ...
      'NumIterations',4);
  in = randi([0 1],frmLen,1);
  coded = step(obj.hTEnc, in(:, 1));
  out = step(obj.hTDec, coded);
  eval = sum(in-out)  % output is cca 3% bit error

Ok there is 2.7% error, but it is still to much. (if setting is 4,[13 15 ],13 - error = 0 )

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Jakub Hoferek 2018-5-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/254617-in-turbo-encoder-decode-if-we-give-encoder-output-to-decoder-without-adding-noise-we-are-getting-e#answer_319612

Thanks to Jan H., he found that problem is in format between coder and decoder. Because decoder need at the input soft bits between -1 and 1. And at the codet output are just normal bits like 0 and 1.