Decentralized federated learning-based intrusion detection in IOT systems

• Ano de defesa: 2023
• Autor(a) principal: Nascimento, Francisco Assis Moreira do
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul; Escola Politécnica; Brasil; PUCRS; Programa de Pós-Graduação em Ciência da Computação
• Palavras-chave em Português: Federated Learning; lnternet of Things; Cybersecurity; Machine Learning; Distributed Ledgers; Aprendizagem Federada; Internet das Coisas; Cibersegurança; Aprendizagem de Máquina; Registro Distribuido; CIENCIA DA COMPUTACAO:TEORIA DA COMPUTACAO
• Link de acesso: https://tede2.pucrs.br/tede2/handle/tede/11398

Systems based on the Internet of Things (IoT) are vulnerable to several types of attacks, mainly due to the weakness of IoT devices, which have little computational and memory power necessary to implement more sophisticated security features. In addition, IoT systems are distributed systems (interconnected and collaborative autonomous devices) and thus inherit all problems related to the need to guarantee confidentiality, integrity, authenticity, and availability. One of the traditional strategies to deal with some of these problems involves intrusion detection and prevention techniques. It is usual to implement them in a centralized way, which in addition to not being scalable for IoT systems with an increasing number of distributed components, implies an unacceptable single point of failure in the system. Besides, sending all collected data to a centralized server in the cloud poses a significant risk to the privacy of information. With data processing at the edge, this problem is minimized in a distributed approach. This thesis presents a decentralized security architecture for supporting detecting intrusion in IoT-based systems, which is based on federated machine learning techniques for intrusion detection, combined with the use of distributed ledger technologies for the implementation of authentication and authorization in the access to resources, allowing to obtain an effective and efficient mechanism to minimize security risks in IoT systems. A prototype was implemented to evaluate the decentralized architecture, allowing several experiments, which proved its effectiveness, with results similar to those obtained with a centralized approach but with all the advantages offered by a decentralized, federated learning-based architecture.