AFF3CT - Release 2.2.1

Changes listed below:

• CI: add public CI for the master and development branches.
• CI: fix wrong display of SNR points in regression tests.
• CI: add long regression tests (30 minutes max. per SNR).
• Fix wrong version detection in the CMakeLists.txt file.
• Add badges to the README file.
• Update 2018 to 2019... and Happy New Year!

AFF3CT - Release 2.2.0

Changes listed below:

• Add LDPC Gallager B/E and Probabilistic Parallel Bit-Flipping (PPBF) decoders.
• Add documentation on AList and QC file formats.
• Improve the implementation of the PSK, QAM, PAM and USER modems with a generic abstract class.
• Add a generic implementation of the SCMA for arbitrary codebooks.
• Externalize the AFF3CT help strings in a dedicated file.
• Improve the CI: check automatically the documentation coherence between Sphinx and the simulator help.
• Automatically find the help strings file and the configuration files on a regular AFF3CT installation.
• Ease the installation of multiple AFF3CT versions on the system.

Minor bug fixes and improvements

Changes listed below:

• Fix a bug in 'mutual_unique' routine.
• Improve the documentation: add AVX-512 flags.
• Fix an under scope variable bug in the Task class.
• Refresh the AFF3CT headers file.
• Improve the CI.

The AFF3CT user documentation is out!

Many changes listed below:

• Introduce the first version of the user doc, most of the simulator parameters are detailed.
• Add a fast intra-frame SIMD implementation of the LDPC BP flooding (SPA) decoder.
• Change GSL Gaussian distribution generation function from the Box-Muller to the Ziggurat method which is faster.
• Bug fix: in the fast event generator, when the event probability is null then fill the event vector with non event values to avoid border effects.
• The channel type is given to the modem directly in the parameters instead of an argument to the build factory function.
• Add the possibility to display a subset of frames in the debug mode and with an inter frame level higher than 1.
• Fix the error tracker replay function when using BEC or BSC channels by adding USER_BEC and USER_BSC channels that use events from a given file.
• In the CLI arguments, shorten the 'debug' word to 'dbg', same for 'frame' to 'fra'.

New major version, many changes

We are proud to announce that AFF3CT v2.0.0 is out !

Many changes listed below:

• Add new channels: optical, BEC and BSC
• Add new noise generators: event generator, from user pdf generator
• Add new code families: Reed-Solomon, Turbo Product
• Add new BCH decoder implementations: fast and genius
• Add LDPC decoders: vertical layered, bit flipping, peeling
• Add LDPC SIMD inter-frames implementations for LDPC decoders: flooding/layered + SPA/LSPA/MS/NMS/OMS/AMS
• Add LDPC encoder from IRA
• Improve LDPC encoder from H
• Add OOK modulations
• Add the possibility to get the mutual information in BER/FER simulations
• Use smart pointers whenever possible
• Improve the arguments management from the command line and the help display
• Bug fixes and little improvements

Double binary Turbo codes are here !

We are proud to announce the support of the double binary Turbo codes in AFF3CT. The DVB-RCS1 and DVB-RCS2 standards are natively supported with their associate interleavers (ARP based).

Many other improvements come with this new version:

• a generic management of the command line arguments,
• support of the SCMA coded modulation/demodulation,
• a fast LDPC min-sum layered decoder has been added (based on inter frame SIMD),
• polar decoders code source generation has been detached from AFF3CT, you can get the polar_decoder_gen project if you are interested in.
• same for the PyBER plotter,
• etc.

Multi-nodes support and various bug fixes

This beta introduces the multi-nodes support (with the MPI standard): the BER/FER simulations can run over many computers to increase the simulation speed. We also add the support of StarPU, a task model for the parallelization of the BER/FER simulations.

Many bugs are fixed from the previous v1.1-alpha.0 version:

• LDPC BP layered decoders now perfectly works in iterative BER/FER simulation context,
• inter frame level per module is now working in most cases,
• etc.

Introducing the LDPC BP decoders

This pre-release contains some LDPC BP decoders (the min-sum and sum-product algorithms with flooding and layered sequencing).
The quality of the code has been constantly improved and the interface with the user is better now (a lot of things can be directly configured from files and be displayed in the PyBER plotter GUI application).

A lot of features have been added:

• Rayleigh channel,
• PSK-PAM-QAM-GSM modulations,
• turbo demodulation (BFERI simulations),
• self-corrected turbo decoder,
• SystemC/TLM standard support,
• etc.

It is time for a checkpoint, we hope that you will enjoy this new release.

The first software release

This release contains Polar, Turbo, RSC (Recursive Systematic Convolutional), Repetition and RA (Repeat and Accumulate) codes. The simulation includes a BPSK modulation and an AWGN channel. Some Polar, RSC and Turbo decoders are optimized with SIMD instructions.