Implementação e otimização do algoritmo de decodificação de LDPC no GRC baseado em redes neurais

• Ano de defesa: 2019
• Autor(a) principal: Jerji, Fadi
• Orientador(a): Akamine, Cristiano
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Presbiteriana Mackenzie
• Palavras-chave em Português: ATSC 3.0; SDR; GRC; LDPC; NN; BP.; NNL
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA
• Link de acesso: http://dspace.mackenzie.br/handle/10899/24508

This manuscript presents a software implementation of the Low-Density Parity-Check (LDPC) codes, defined by the Third Digital TV Standard developed by the Advanced Television Systems Committee (ATSC 3.0). In order to decode the messages in software and in real-time, the decoder was implemented using a structure that is based on the Neural Networks (NN) and a modified Back-Propagation (BP) training method using Software Defined Radio (SDR). This implementation was realized using the GNU Radio Companion (GRC) software and the C++ programming language. The mathematical analysis and the experimental results showed that the NN-based LDPC decoder has a performance that exceeds that of the classical decoder BFA by up to 4.66 dB for the ATSC 3.0 LDPC codes of the size 16200 bits and up to 3.49 dB for the ATSC 3.0 LDPC codes of the size 64800 bits. The complexity analysis of the proposed decoder highlighted its low complexity in comparison to the traditional methods, the thing that permits the implementation of the proposed decoder in software and in real-time.