Requesting help with R=2/3 turbo encoding for lte waveform.

Hi all,

I am trying to recreate an LTE compliant waveform ('A2-1') but I am running into some trouble. the current command: lteTurboEncode.m only supports R=1/3 turbo coding which i believe is leading me to biterrors when i go to undo all of the rate match, demodulation, etc.

Bottom line: Is there a way to implement 2/3 encoding? I've tried used the comms toolboox too, but with no luck.

Code is posted below. I highlighted in bold where I believe the issue is. You'll notice some is underlined too. that was my attempt at using the comms tool box to fix the problem, which is separate from the lte toolbox attempt (not underlined, but still in bold).

Thank you!

frc=lteRMCUL('A2-1');

frc.TotSubframes=1;

txdata=randi([0 1], frc.PUSCH.TrBlkSizes(1)*frc.TotSubframes, 1);

TrBLKcrc = lteCRCEncode(txdata,'24A');

%% PROBLEM HERE. %2/3 Rate Turbo encoding the bits.

trellis = poly2trellis([5 4],[23 35 0; 0 5 13]);

convEncoder = comm.ConvolutionalEncoder('TrellisStructure',trellis);

TRBOcodeBLK = convEncoder(TrBLKcrc);

TRBOcodeBLK = lteTurboEncode(TrBLKcrc);

%% END

TxTrBLKcrc = lteRateMatchTurbo(TRBOcodeBLK,frc.PUSCH.CodedTrBlkSizes(1),0,frc.PUSCH);

QAMsyms = lteSymbolModulate(TxTrBLKcrc,'16QAM');

[antseq,info,layerseq] = ltePUSCHDRS(frc,frc.PUSCH);

DRSind = ltePUSCHDRSIndices(frc,frc.PUSCH);

PUSCHind = ltePUSCHIndices(frc,frc.PUSCH);

TxSYMS(DRSind)=antseq;

TxSYMS(PUSCHind)=QAMsyms;

dims = lteULResourceGridSize(frc);

TxGrid = reshape(TxSYMS,[72 14]);

txWaveform = lteSCFDMAModulate(frc,TxGrid);

RxGrid = lteSCFDMADemodulate(frc,txWaveform);

RxSYMS = reshape(RxGrid,[prod(dims) 1]);

Rxbits = lteSymbolDemodulate(RxSYMS(PUSCHind),'16QAM','Hard');

undoRatematch = lteRateRecoverTurbo(Rxbits,frc.PUSCH.TrBlkSizes(1),0,frc.PUSCH);

%% This is where the problem lies!!!%%

traceBack = 16;

vitDecoder = comm.ViterbiDecoder('TrellisStructure',trellis,'InputFormat','Hard','TracebackDepth',traceBack);

RxTrBLKcrc = vitDecoder(cell2mat(undoRatematch));

RxTrBLKcrc = lteTurboDecode(undoRatematch);

[blk,err] = lteCRCDecode(RxTrBLKcrc,'24A'); %err should 0 if no problems.

Answer 1

[Graham Freeland]

Hi,

You need to use soft demodulation of the 16QAM symbols, which is the default for that function i.e.

Rxbits = lteSymbolDemodulate(RxSYMS(PUSCHind),'16QAM');

or

Rxbits = lteSymbolDemodulate(RxSYMS(PUSCHind),'16QAM','Soft');

There is no 2/3 rate turbo coding in LTE. It's a fixed 1/3 coder with the overall code rate variation coming from the rate matching stage.

Graham

[James Long]

Graham,

Thank you for your reply. worked like a charm.

now, time to get smart on soft demod...

James

Answer 2

[Baghdadi Aya]

i try also to code and decoed a random signal and to calculate the BER, but the problem that the code is too slow ,and the code execution can take more than 30 minutes for just few number of bits !!!

this is the code :

clear all

close all

M = 64; % Modulation order

bps = log2(M); % Bits per symbol

EbNo = (2:0.5:10);

pktLen = 100;

%Initialize the bit error rate vector.

ber = zeros(size(EbNo));

%Create System objects for a turbo encoder and decoder pair, where the interleaver indices are supplied as input arguments<

turboEnc = comm.TurboEncoder('InterleaverIndicesSource','Input port');

turboDec = comm.TurboDecoder('InterleaverIndicesSource','Input port', ...

'NumIterations',1);

%Create an AWGN channel System object and an error rate counter System object

awgnChannel = comm.AWGNChannel('NoiseMethod','Variance','Variance',1);

errorRate = comm.ErrorRate;

rate = pktLen/(3*pktLen+4*3);%rate de codeur

for k = 1:length(EbNo)

errorStats = zeros(1,3);

EsNo = EbNo(k) + 10*log10(bps);

snrdB = EsNo + 10*log10(rate); % in dB

noiseVar = 1./(10.^(snrdB/10));

awgnChannel.Variance = noiseVar;

while errorStats(2) < 100 && errorStats(3) < 1e7

% Generate random binary data

data = randi([0 1],pktLen,1);

% Interleaver indices

intrlvrInd = randperm(pktLen);

% Turbo encode the data

encodedData = turboEnc(data,intrlvrInd);

% Modulate the encoded data

modSignal = qammod(encodedData,M,'InputType','bit','UnitAveragePower',true);

% Pass the signal through the AWGN channel

rxSignal = awgnChannel(modSignal);

% Demodulate the received signal

demodSignal = qamdemod(rxSignal,M,'UnitAveragePower',true,'OutputType','llr','NoiseVariance',noiseVar);

% Turbo decode the demodulated signal. Because the bit mapping from the

% demodulator is opposite that expected by the turbo decoder, the

% decoder input must use the inverse of demodulated signal.

rxBits = turboDec(-demodSignal,intrlvrInd);

% Calculate the error statistics

errorStats = errorRate(data,rxBits);

% Save the BER data and reset the bit error rate object

ber(k) = errorStats(1);

reset(errorRate);

end

end

%Plot the bit error rate

fprintf('Bit error rate = %5.2e\nNumber of errors = %d\nTotal bits = %d\n', ...

errorStats)

semilogy(EbNo,ber,'-*')

grid

xlabel('Eb/No (dB)')

ylabel('Bit Error Rate')

I need t know where is the problem , by the way the most function that takes the most of time is in turbodecoder