Framework para autocura cognitiva de redes de banda larga

• Ano de defesa: 2024
• Autor(a) principal: Vieira, Enock Cabral Almeida
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Uberlândia; Brasil; Programa de Pós-graduação em Ciência da Computação
• Palavras-chave em Português: Aprendizado de Máquina; Autocura; Banda Larga; Redes; Framework; Redes Neurais Artificiais; Machine Learning; Self Healing; Broadband; Network; Artificial Neural Networks; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::SISTEMAS DE COMPUTACAO::ARQUITETURA DE SISTEMAS DE COMPUTACAO; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA::TELECOMUNICACOES::SISTEMAS DE TELECOMUNICACOES; ODS::ODS 9. Indústria, Inovação e infraestrutura - Construir infraestrutura resiliente, promover a industrialização inclusiva e sustentável, e fomentar a inovação.
• Link de acesso: https://repositorio.ufu.br/handle/123456789/44359; http://doi.org/10.14393/ufu.di.2024.568

With the growing volume of data traffic demanded by corporate, business, and retail consumers, telecommunications operators are becoming an increasingly important player in the world economy. However, the operators must prepare themselves with solutions that allow dealing with incidents more quickly or even avoid them, always focusing on maintaining an acceptable customer service level. In this context, SH solutions, supported by ML mechanisms, emerge as possibilities to address this challenge. This work presents a cognitive self-healing framework for telecommunications operators. This framework encompasses self-diagnosis, analysis, and automatic actuation for failure mitigation in fiber broadband telecommunications based on GPON. In addition, we did an experimental evaluation using an anonymized dataset from the operators’ users, extracted from its NMS and CRM, bringing more reliability to our results. This work shows that using ML in telecommunication broadband networks is viable and can change how telecom operators manage and improve customer experience. We show that an intelligent model could do machine learning in telecom networks and make decisions without human intervention. Three automatic cognitive models were tested as experimental proof of concept with an average accuracy above 96%.