This source code is intended to reproduce the results published in [LB18,LBT+18] and [SLB18, Sta18]. In these works, it is explained how the information bottleneck method can be used to design decoders for regular and irregular LDPC codes.

1. Download and install Python 3.6 (we recommend Anaconda)
2. Download and install the ib_base package
3. Clone the git repository
4. Installation requires the following packages:

• numpy
• IB_Base
• PyOpenCl.2018
• mako
• progressbar

To construct a decoder, make use of the scripts decoder_config_generation.py in the different folders. Here, you can enter:

• the maximum number of decoding iterations $i_{max}$
• the cardinality of the exchanged messages $|\mathcal{T}|$
• the the degree distributions $\mathbf{d}_c$ and $\mathbf{d}_v$
• the design-$E_b/N_0$ for which you want to construct the decoder

You can run either benchmark simulations for the belief-propagation decoding, belief-propagation decoding with an channel output quantizer or min-sum decoding. Please make sure that you have OpenCL set up correctly.

To run the benchmark simulations just use BER_simulation_OpenCL_min_sum.py or BER_simulation_OpenCL_quant_BP.py in the respective folders.

To run BER simulations just use BER_simulation_OpenCL_enc.py in the respective folders. The ending "enc" indicates, that the transmission chain uses the appropriate LDPC encoder, not the all-zeros codeword is transmitted.

Note: Make sure that you have an IB decoder generated before running a simulation.

A detailed documentation of all provided functions and a more complete test suite will be available soon.

The code is distributed under the MIT license. When using the code for your research please cite our work.

[LBT+18] J. Lewandowsky, G. Bauch, M. Tschauner, and P. Oppermann, “Design and Evaluation of Information Bottleneck LDPC Decoders for Software Defined Radios,” in Proc. 12th International Conference on Signal Processing and Communication Systems (ICSPCS), Australia, 2018.

[SLB18] M. Stark, J. Lewandowsky, G. Bauch. "Information-Bottleneck Decoding of High-Rate Irregular LDPC Codes for Optical Communication using Message Alignment“. Applied Sciences. 2018; 8(10):1884.

[SLB18a] M. Stark, J. Lewandowsky, and G. Bauch, “Information-Optimum LDPC Decoders with Message Alignment for Irregular Codes,” in 2018 IEEE Global Communications Conference (Globecom2018), Abu Dhabi, United Arab Emirates, 2018.

[LB18] J. Lewandowsky and G. Bauch, “Information-Optimum LDPC Decoders Based on the Information Bottleneck Method,” IEEE Access, vol. 6, pp. 4054–4071, 2018. https://ieeexplore.ieee.org/document/8268118