Convolutional Coding over BEC (MATLAB)

This repo implements a convolutional (3,4) encoder and two decoding methods over the Binary Erasure Channel (BEC).

Files

• conv_encoder_34.m — (3,4) convolutional encoder using generator matrices H0, H1, H2 with memory M=2.
• SW_decoder.m — sliding window decoder that solves a linear system per window to recover erasures.
• peeling_decoder.m — iterative decoder for erasure recovery using a sparse parity-check matrix H.
• conv_34_simulate.m — runs FER/BER simulations over varying erasure probabilities (p) and plots FER.

Simulation Setup

• Code rate: 3/4 (k=3, n=4)
• Erasure decoding tested for p ∈ (0.001, 0.15)
• 500 trials per erasure probability
• Outputs: FER and BER vs. p (only FER plotted)

Notes

• Only BEC supported
• All parameters are hardcoded (no argument parsing)
• Matrix H is fixed and used in both decoders