Loading...
Publication
Encriptação Homomórfica: Técnicas e Ferramentas de Segurança e Privacidade para Processamento de Dados
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.11 MB | Adobe PDF |
Authors
Abstract(s)
Perante os desafios atuais da segurança da informação, esta dissertação explora
a Encriptação Homomórfica (Homomorphic Encryption - HE) como uma solução
inovadora para o processamento seguro de dados sensíveis, num cenário cada vez
mais digital e dependente da computação em nuvem, onde o risco de ciberataques
é cada vez maior. A HE distingue-se por permitir a realização de operações aritmé ticas e lógicas diretamente sobre dados encriptados, assegurando, em simultâneo, a
privacidade da informação e o cumprimento de normas internacionais de proteção
de dados.
A metodologia adotada integrou uma revisão bibliográfica teórica e uma compo nente experimental aplicada. Foram realizados testes sobre dados encriptados para
avaliar os tipos de operações suportadas, o desempenho (em particular, a latência),
a precisão numérica e a complexidade de integração. Estes testes decorreram em di ferentes Sistemas Operativos (SO) e Sistemas de Gestão de Bases de Dados (SGBD),
recorrendo a bibliotecas criptográficas amplamente reconhecidas tanto na investiga ção como na indústria, bem como a um conjunto de dados sintéticos representativos
do domínio da saúde.
Os resultados obtidos evidenciaram tanto as vantagens como as limitações prá ticas dos diversos esquemas de HE avaliados, nomeadamente BFV, CKKS, BGV,
TFHE, Paillier e RSA sem padding. Foram identificadas diferenças no suporte a
operações, na latência, na precisão dos resultados numéricos e na complexidade de
integração com sistemas existentes. Com base nesses resultados, foram desenvol vidos modelos comparativos que sintetizaram o comportamento de cada esquema
em diferentes contextos de aplicação, permitindo definir diretrizes práticas para a
sua adoção. Este contributo constitui uma base sólida para investigações futuras na
área.
In light of the current challenges in information security, this dissertation explores Homomorphic Encryption (HE) as an innovative solution for the secure processing of sensitive data. This study is set against an increasingly digital landscape, where reliance on cloud computing continues to grow and the risk of cyberattacks is ever increasing. HE stands out for enabling arithmetic and logical operations to be performed directly on encrypted data, while simultaneously ensuring data privacy and compliance with international data protection standards. The adopted methodology combined a theoretical literature review with an ap plied experimental component. Tests were conducted on encrypted data to evaluate the types of supported operations, performance (particularly latency), numerical precision, and integration complexity. These tests were carried out across differ ent Operating Systems (OS) and Database Management Systems (DBMS), using cryptographic libraries widely recognised in both research and industry, as well as a synthetic dataset representative of the healthcare domain. The results revealed both the advantages and practical limitations of the eval uated HE schemes, namely BFV, CKKS, BGV, TFHE, Paillier, and RSA without padding. Differences were identified in terms of operation support, latency, numeri cal accuracy, and ease of integration with existing systems. Based on these findings, comparative models were developed to synthesise the behaviour of each scheme in various application contexts, allowing for the definition of practical guidelines for their adoption. This work provides a solid foundation for future research in the field.
In light of the current challenges in information security, this dissertation explores Homomorphic Encryption (HE) as an innovative solution for the secure processing of sensitive data. This study is set against an increasingly digital landscape, where reliance on cloud computing continues to grow and the risk of cyberattacks is ever increasing. HE stands out for enabling arithmetic and logical operations to be performed directly on encrypted data, while simultaneously ensuring data privacy and compliance with international data protection standards. The adopted methodology combined a theoretical literature review with an ap plied experimental component. Tests were conducted on encrypted data to evaluate the types of supported operations, performance (particularly latency), numerical precision, and integration complexity. These tests were carried out across differ ent Operating Systems (OS) and Database Management Systems (DBMS), using cryptographic libraries widely recognised in both research and industry, as well as a synthetic dataset representative of the healthcare domain. The results revealed both the advantages and practical limitations of the eval uated HE schemes, namely BFV, CKKS, BGV, TFHE, Paillier, and RSA without padding. Differences were identified in terms of operation support, latency, numeri cal accuracy, and ease of integration with existing systems. Based on these findings, comparative models were developed to synthesise the behaviour of each scheme in various application contexts, allowing for the definition of practical guidelines for their adoption. This work provides a solid foundation for future research in the field.
Description
Keywords
Encriptação Homomórfica Esquemas Homomórficos Computação Segura Desempenho Criptográfico Homomorphic Encryption Homomorphic Schemes Secure Computation Cryptographic Peformance