Avaliação de Desempenho e Custo Computacional na Utilização da Blockchain Ethereum em Dispositivos de Internet das Coisas

• Ano de defesa: 2023
• Autor(a) principal: Silva, Maykon Valério da
• Orientador(a): Rodrigues, Luiz Antônio
• Banca de defesa: Oyamada, Marcio Seiji, Camargo, Edson Tavares de, Santos, Aldri Luiz dos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual do Oeste do Paraná; Cascavel
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
• Departamento: Centro de Ciências Exatas e Tecnológicas
• País: Brasil
• Palavras-chave em Português: IoT; Blockchain; Ethereum; Blockbench; Ethash; Clique
• Palavras-chave em Inglês: IoT; Blockchain; Ethereum; Blockbench; Ethash; Clique
• Área do conhecimento CNPq: CIÊNCIA DA COMPUTAÇÃO
• Link de acesso: https://tede.unioeste.br/handle/tede/6734

As the Internet of Things expands, billions of new devices will be networked in the coming years, collecting and transmitting data that requires secure communication mechanisms and data storage. Blockchain is currently emerging as a system capable of meeting these requirements. However, in its classic form, it comes with high computational and energy costs, which is prohibitive for embedded devices that form IoT networks due to their limited resources. Therefore, changes to the way blockchain works are constantly proposed to improve it. In order to investigate the applicability and impact of some proposed changes, we used a modified version of Blockbench to perform systematic benchmarks on private Ethereum networks by adding an embedded Raspberry Pi device to the network as a participating node and taking performance measurements of the network and devices. We compared scenarios with and without the presence of the embedded device, as well as the Ethash consensus protocol (proof-of-work) and clique (proof-of-authority). The results show that the Raspberry Pi cannot be a network miner in the Ethash protocol, since it can only participate as a light node, leading to an analysis of the consequences of the presence of light nodes in the network. The results also show the rapid network saturation in proof-of-work and the expected superiority in latency, throughput, and consumption of computational resources when a consensus proof-of-authority is used.