Channel coding course project

Course is outlined here.

Project structure

• simulation.py is the executable file that actually test codes for various SNR

• report/ contains LaTeX file and dependencies (plots, ...) needed for project presentation

• Makefile makes straightforward to generate relevant plots in report/figures/ from simulation result file with make all

• specs/ is the place where compressed encoding matrix for the code should be put, in order to perform simulations. A suitable Python module is embedded to ease matrix decompression for various code lengths.

• LDPC/ is a Python module that contains relevant procedures to encode and decode, given specification. Most expensive operations are written in the much faster Cython, a Python-like language that compiles to C.

• results/ contains all relevant CSV output of simulation.py execution

• test-matrices.py checks if matrices are symmetric in some sense, with a standard output report

How to run project

• put relevant compressed encoding matrices in specs/, as specified specs/README.md

• install needed Python packages, for example creating a virtual environment (see Unix example, where virtualenv command is required)

virtualenv venv/
source venv/bin/activate
pip install -r requirements.txt

• run wanted simulation, after checking its parameters (in the code)

python3 simulation.py

• check result file is created, namely SNRvsPe.csv.gz in our context

Missing specs/ files

Private version (with licensed specs/ matrices) is here; I have done this because of parity check matrices licensing by IEEE.