Hi,

I'm looking for a tool which can simulate turbo codes for general M-PSK/M-QAM constellations. I would like to test the performance of turbo coding in different channel models: Eg AWGN, additive Laplace noise, impulsive noise and I want to evaluate it for different constellation setups also. Because the channels are not necessarily AWGN now, the tool should also be flexible enough that user can specify how to calculate bit LLR (which depends on the noise distribution).

The options I found so far are:

1. Matlab: Seems to handle any modulation scheme like magic. User don't even have to specify which modulation scheme it used for decoding. Didn't understand how decoder can work without the knowledge of modulation scheme. However, the source is closed, so not able to change channel models/noise distribution (?).

2. CommPy: Opensource, but limits to only BPSK scheme. Also, only rate 1/3 turbo decoder is supported. Even though user can easily modify the core to get LLR for different noise distributions, nothing other than BPSK is supported for LLR calculation.

3 IT++: Opensource, but again only BPSK scheme. Turbo decoder seems to support a wider range of configurations than CommPy, but the decode can take only received signals in real number (BPSK signalling). I think changing channel model will not be much difficult, though not tried yet.

Can anybody suggest a Turbo decoder simulation tool which

1. Can simulate general m-PSK/m-QAM constellations

2. user can can specify the noise distribution to use for LLR calculation

3. Have control over generating polynomials and puncturing

4. Is preferably opensource

Thanks,

Vishnu Raj