Improving cloud based encrypted databases.

• Ano de defesa: 2019
• Autor(a) principal: Cominetti, Eduardo Lopes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Algorithm; Algoritmos; Banco de dados; Computer security; CriCryptographic; Criptologia; Database; Segurança de computadores
• Link de acesso: http://www.teses.usp.br/teses/disponiveis/3/3141/tde-29052019-072659/

Databases are a cornerstone for the operation of many services, such as banking, web stores and even health care. The cost of maintaining such a large collection of data on-premise is high, and the cloud can be used to share computational resources and mitigate this problem. Unfortunately, a great amount of data may be private or confidential, thus requiring to be protected from agents. Moreover, this data needs to be manipulated to provide useful information to its owner. Hence, encrypted databases that allow the manipulation of data without compromising its privacy have surfaced in the recent years. Albeit promising, the solutions available in the literature can still be improved in terms of eciency, flexibility and even security. In this work, we propose modifications to CryptDB, a state-of-the-art encrypted cloud database, aiming to enhance its eciency, flexibility and security; this is accomplished by improving or changing its underlying cryptographic primitives. The eciency of CryptDB was improved by substituting a new homomorphic algorithm proposed by us for the Paillier cryptosystem. The flexibility of the cloud database was augmented by modifying how a text is encrypted using the Song, Wagner and Perrig algorithm, thus enabling wildcard searches. Finally, the security of the system was enhanced by substituting the Song, Wagner and Perrig algorithm for the AES in CMC mode at the deterministic layer.