COde to generate H matrix for LDPC code with R=4/5 and tailbiting case

The goal of this project is to design an efficient LDPC code for operation over the binary erasure channel (BEC). You will begin by determining the optimal degree distributions. Upon finding the degree distribution, you will design a full-rank H matrix. Finally, you will run a simulation of your code on a cluster computer that is accessible by a web application.

You code must satisfy the following constraints: • R = 4/5 • The H matrix must be full rank. • You will generate H matrices for two lengths: n = 1, 000 (short) and n = 25, 000 (long). You are allowed to use whatever type of LDPC code you like, as long as the H matrix is full rank. An acceptable choice is to implement the code as a check-regular tail-biting eIRA code, but you are free to explore other options.

Answer 1

[Ben Morse]

This sounds like a homework question.

To start you should read about gf(2) and how to convert between H and G matrices. You should also find a LDPC solver or write your own. Attached is a document describing LDPC in flash memory and explains a iterative soft decision decoder. It's a good start.

[Achala G]

Can you share matlab code?