IT++ Logo
Simulation of PCCCs in an AWGN channel

This program simulates Parallel Concatenated Convolutional Codes (PCCCs) of coding rate 1/3 using a turbo decoder with two SISO RSC modules.

Reference: S. Benedetto, D. Divsalar, G. Motorsi and F. Pollara, "A Soft-Input Soft-Output Maximum A posteriori (MAP) Module to Decode Parallel and Serial Concatenated Codes", TDA Progress Report, nov. 1996

#include "itpp/itcomm.h"
using namespace itpp;
using std::cout;
using std::endl;
using std::string;
int main(void)
{
//general parameters
double threshold_value = 10;
string map_metric="maxlogMAP";
ivec gen = "07 05";//octal form, feedback first
int constraint_length = 3;
int nb_errors_lim = 3000;
int nb_bits_lim = int(1e6);
int perm_len = (1<<14);//total number of bits in a block (with tail)
int nb_iter = 10;//number of iterations in the turbo decoder
vec EbN0_dB = "0:0.1:5";
double R = 1.0/3.0;//coding rate (non punctured PCCC)
double Ec = 1.0;//coded bit energy
//other parameters
int nb_bits = perm_len-(constraint_length-1);//number of bits in a block (without tail)
vec sigma2 = (0.5*Ec/R)*pow(inv_dB(EbN0_dB), -1.0);//N0/2
double Lc;//scaling factor
int nb_blocks;//number of blocks
int nb_errors;
ivec perm(perm_len);
ivec inv_perm(perm_len);
bvec bits(nb_bits);
int cod_bits_len = perm_len*gen.length();
bmat cod1_bits;//tail is added
bvec tail;
bvec cod2_input;
bmat cod2_bits;
int rec_len = int(1.0/R)*perm_len;
bvec coded_bits(rec_len);
vec rec(rec_len);
vec dec1_intrinsic_coded(cod_bits_len);
vec dec2_intrinsic_coded(cod_bits_len);
vec apriori_data(perm_len);//a priori LLR for information bits
vec extrinsic_coded(perm_len);
vec extrinsic_data(perm_len);
bvec rec_bits(perm_len);
int snr_len = EbN0_dB.length();
mat ber(nb_iter,snr_len);
ber.zeros();
register int en,n;
//Recursive Systematic Convolutional Code
cc.set_generator_polynomials(gen, constraint_length);//initial state should be the zero state
//BPSK modulator
BPSK bpsk;
//AWGN channel
AWGN_Channel channel;
//SISO modules
SISO siso;
siso.set_generators(gen, constraint_length);
siso.set_map_metric(map_metric);
//BER
BERC berc;
//Randomize generators
//main loop
for (en=0;en<snr_len;en++)
{
cout << "EbN0_dB = " << EbN0_dB(en) << endl;
channel.set_noise(sigma2(en));
Lc = -2/sigma2(en);//normalisation factor for intrinsic information (take into account the BPSK mapping)
nb_errors = 0;
nb_blocks = 0;
while ((nb_errors<nb_errors_lim) && (nb_blocks*nb_bits<nb_bits_lim))
{
//permutation
perm = sort_index(randu(perm_len));
//inverse permutation
inv_perm = sort_index(perm);
//bits generation
bits = randb(nb_bits);
//parallel concatenated convolutional code
cc.encode_tail(bits, tail, cod1_bits);//tail is added here to information bits to close the trellis
cod2_input = concat(bits, tail);
cc.encode(cod2_input(perm), cod2_bits);
for (n=0;n<perm_len;n++)//output with no puncturing
{
coded_bits(3*n) = cod2_input(n);//systematic output
coded_bits(3*n+1) = cod1_bits(n,0);//first parity output
coded_bits(3*n+2) = cod2_bits(n,0);//second parity output
}
//BPSK modulation (1->-1,0->+1) + AWGN channel
rec = channel(bpsk.modulate_bits(coded_bits));
//form input for SISO blocks
for (n=0;n<perm_len;n++)
{
dec1_intrinsic_coded(2*n) = Lc*rec(3*n);
dec1_intrinsic_coded(2*n+1) = Lc*rec(3*n+1);
dec2_intrinsic_coded(2*n) = 0.0;//systematic output of the CC is already used in decoder1
dec2_intrinsic_coded(2*n+1) = Lc*rec(3*n+2);
}
//turbo decoder
apriori_data.zeros();//a priori LLR for information bits
for (n=0;n<nb_iter;n++)
{
//first decoder (terminated trellis)
siso.rsc(extrinsic_coded, extrinsic_data, dec1_intrinsic_coded, apriori_data, true);
//interleave
apriori_data = extrinsic_data(perm);
//threshold
apriori_data = SISO::threshold(apriori_data, threshold_value);
//second decoder (unterminated trellis)
siso.rsc(extrinsic_coded, extrinsic_data, dec2_intrinsic_coded, apriori_data);
//decision
apriori_data += extrinsic_data;//a posteriori information
rec_bits = bpsk.demodulate_bits(-apriori_data(inv_perm));//take into account the BPSK mapping
//count errors
berc.clear();
berc.count(bits, rec_bits.left(nb_bits));
ber(n,en) += berc.get_errorrate();
//deinterleave for the next iteration
apriori_data = extrinsic_data(inv_perm);
}//end iterations
nb_errors += int(berc.get_errors());//get number of errors at the last iteration
nb_blocks++;
}//end blocks (while loop)
//compute BER over all tx blocks
ber.set_col(en, ber.get_col(en)/nb_blocks);
}
it_file ff("pccc_bersim_awgn.it");
ff << Name("EbN0_dB") << EbN0_dB;
ff << Name("BER") << ber;
ff.close();
return 0;
}

When you run this program, the results (BER and EbN0_dB) are saved into pccc_bersim_awgn.it file. Using the following MATLAB script:

clear all
itload('pccc_bersim_awgn.it');
figure
semilogy(EbN0_dB, BER, 'o-')
grid on
xlabel('E_b/N_0 [dB]')
ylabel('BER')

the results can be displayed.

Similarly, the results can be displayed using the following Python script (pyitpp, numpy and matplotlib modules are required):

#!/usr/bin/env python
from pyitpp import itload
from matplotlib.pyplot import *
out = itload('pccc_bersim_awgn.it')
semilogy(out['EbN0_dB'], out['BER'].T, 'o-')
grid()
xlabel('$E_b/N_0 [dB]$')
ylabel('$BER$')
show()
SourceForge Logo

Generated on Sat Jul 6 2013 10:54:28 for IT++ by Doxygen 1.8.2