| Resumo: | There are two dominant themes that have become increasingly more important in our technological society. First, the recurrent use of cloud-based solutions which provide infrastructures, computation platforms and storage as services. Secondly, the use of applicational large logs for analytics and operational monitoring in critical systems. Moreover, auditing activities, debugging of applications and inspection of events generated by errors or potential unexpected operations - including those generated as alerts by intrusion detection systems - are common situations where extensive logs must be analyzed, and easy access is required. More often than not, a part of the generated logs can be deemed as sensitive, requiring a privacy-enhancing and queryable solution. In this dissertation, our main goal is to propose a novel approach of storing encrypted critical data in an elastic and scalable cloud-based storage, focusing on handling JSONbased ciphered documents. To this end, we make use of Searchable and Homomorphic Encryption methods to allow operations on the ciphered documents. Additionally, our solution allows for the user to be near oblivious to our system’s internals, providing transparency while in use. The achieved end goal is a unified middleware system capable of providing improved system usability, privacy, and rich querying over the data. This previously mentioned objective is addressed while maintaining server-side auditable logs, allowing for searchable capabilities by the log owner or authorized users, with integrity and authenticity proofs. Our proposed solution, named Chameleon, provides rich querying facilities on ciphered data - including conjunctive keyword, ordering correlation and boolean queries - while supporting field searching and nested aggregations. The aforementioned operations allow our solution to provide data analytics upon ciphered JSON documents, using Elasticsearch as our storage and search engine.
|
|---|