Analysis of CNN Based Schemes for LDPC Code Classification Using LUT Based Algorithms

Abstract

This paper analyzes the performance of an LDPC code classification system that determines membership of code-words among 3 randomly generated binary LDPC codes. These codes all have the same codeword size and coderate. High classification accuracies are obtained with relatively small neural networks. The analysis presented here determines the accuracies of various look up tables and compares them to the performance of the neural networks.