Distributed Processing for Encoding and Decoding of Binary LDPC codes using MPI

TL;DR

This paper explores distributed processing techniques using MPI to accelerate encoding and decoding of LDPC codes, demonstrating performance improvements through multi-core CPU implementations.

Contribution

It introduces a distributed processing approach for LDPC codes using MPI, comparing stream and batch processing methods for practical error correction applications.

Findings

Distributed processing reduces encoding/decoding time

Performance improves with more CPUs and cores

Stream and batch processing methods are evaluated

Abstract

Low Density Parity Check (LDPC) codes are linear error correcting codes used in communication systems for Forward Error Correction (FEC). But, intensive computation is required for encoding and decoding of LDPC codes, making it difficult for practical usage in general purpose software based signal processing systems. In order to accelerate the encoding and decoding of LDPC codes, distributed processing over multiple multi-core CPUs using Message Passing Interface (MPI) is performed. Implementation is done using Stream Processing and Batch Processing mechanisms and the execution time for both implementations is compared w.r.t variation in number of CPUs and number of cores per CPU. Performance evaluation of distributed processing is shown by variation in execution time w.r.t. increase in number of processors (CPU cores).