Optimisation of reinforcement learning-based decoding strategies for binary linear codes

• Main Author: Ang, Rosamund Pei Yin
• Other Authors: Frederique Elise Oggier
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Science::Mathematics
• Online Access: https://hdl.handle.net/10356/156951
• Institution: Nanyang Technological University

Linear codes are a class of error-correcting codes, whereby any linear combination of two codewords always results in another codeword. In general, they are defined over a finite field, and have broad applications in the fields of communications and information systems. The present work surveys the construction and decoding methods for binary linear codes, and approaches the decoding of such linear codes as a reinforcement learning (RL) problem. The present work also presents a general theoretical RL-based framework for the decoding of binary linear codes over a binary symmetric channel (BSC).