Study of the Sparse Superposition Codes and the Generalized Approximate Message Passing Decoder for the Communication over Binary Symmetric and Z Channels

TL;DR

This paper investigates the performance of the Generalized Approximate Message Passing (GAMP) decoder for sparse superposition codes over Binary Symmetric and Z channels, focusing on its effectiveness compared to other channels.

Contribution

It provides an analysis of GAMP decoder performance specifically for BSC and Z channels, extending previous evaluations from AWGN channels.

Findings

GAMP performs effectively for sparse superposition codes over BSC and Z channels.

Performance varies with channel parameters and code settings.

Comparison shows differences in decoding success across channels.

Abstract

In this project, the behavior of Generalized Approximate Message-Passing Decoder for BSC and Z Channel is studied using i.i.d matrices for constructing the codewords. The performance of GAMP in AWGN Channel is already evaluated in the previous scientific work of Jean Barbier, therefore, this project mainly focuses on the performance of GAMP decoder for BSC and Z Channel. We evaluate the performance of the GAMP decoder for sparse superposition codes at various settings and compare the performance of decoder for different channels and parameters.