Secure Query Answering for Outsourced Databases in Cloud Computing

云计算无疑是处理“大数据”的一种有效方法，它通过由第三方云托管的强大且可配置的计算资源提供按需高质量的服务。云计算的一个应用是“数据库即服务”，数据所有者将复杂的数据库管理系统外包到云中。此应用程序的一个主要障碍是敏感数据的机密性，因为云服务器不在数据所有者的控制之下，并且不受信任。为了保护数据隐私和机密性，敏感数据必须在外包给云服务器之前进行加密。然而，这使得传统的查询处理方法（如索引）不适用，因为所有这些方法都对明文数据进行操作。因此，对加密数据的查询处理是一个非常重要的新研究课题。**有三个要求需要解决。首先是保密性，即云服务器通过存储数据和计算查询，不应该了解查询结果以外的任何信息。第二种是身份验证，如果服务器没有遵循规定的协议或篡改数据或查询结果，用户应该能够检测到事件的发生。第三是效率，该方法应该可扩展到云规模的数据。这需要索引和计算方法直接处理加密数据，而不需要扫描整个数据库。密码学传统上是解决第一个需求的领域，但是所开发的方法不能扩展到云规模，因为通常需要对数据库进行全面扫描。数据库和信息检索的最新发展解决了其中的一些问题，但据我们所知，还没有工作解决了所有这三个要求。此外，数据是动态增长的，必须随着时间的推移满足上述需求。最初的研究表明，对于某些类型的查询，在一些宽松的保密要求下，解决上述所有要求的“可搜索加密”是可能的。本提案的目标是研究可搜索性和机密性之间的权衡，并在提供实际可接受的机密性的同时，为更大的查询类别提出可搜索加密的解决方案。该提案的创新之处在于将数据库、密码学和信息检索技术结合起来，以解决这一具有挑战性的问题，而这些技术传统上有不同的重点。该项目的成果将提供更安全的云计算执行方法，从而使所有数据所有者、云服务器和数据用户受益。

Cloud computing is no doubt an effective approach to deal with "big data", through providing on-demand high quality services from powerful and configurable computing resources hosted by a third party called the cloud. One application of cloud computing is "database as a service", where the data owner outsources complex database management systems into the cloud. A major obstacle to this application is the confidentiality of sensitive data since the cloud server is not under the control of the data owner and is untrusted. To protect data privacy and confidentiality, sensitive data must be encrypted before being outsourced to the cloud server. This, however, renders traditional query processing methods (such as indexing) inapplicable because all such methods operate on the plaintext data. Thus, query processing over encrypted data is a new research issue and of paramount importance. **There are three requirements to be addressed. The first is confidentiality, that is, the cloud server should not learn anything other than the query result through storing the data and computing the query. The second is authentication where the user should be able to detect the occurrence of the event if the server does not follow the prescribed protocol or tampers with the data or query result. The third is efficiency where the method should be scalable to data at the cloud scale. This requires indexing and computational methods that work directly on encrypted data without scanning the entire database. Cryptography is traditionally a field addressing the first requirement, but the methods developed do not scale up to the cloud scale because a full scan of database often is required. Recent developments in databases and information retrieval address some of these issues, but to our knowledge, no work has addressed all three requirements. In addition, data grows dynamically and the above requirements must be met over time. Initial studies suggest that a "searchable encryption" that addresses all of the above requirements is possible for certain types of queries under some relaxed requirement of confidentiality. The objective of this proposal is to investigate the trade-off between searchability and confidentiality and propose solutions to searchable encryption for larger classes of queries while providing practically acceptable confidentiality. The innovation of this proposal is combining the techniques from databases, cryptography, and information retrieval, which traditionally has different focuses, to address this challenging problem. The outcome of this project will provide methodologies for more secure execution of cloud computing, therefore, benefit all of the data owners, cloud servers, and data users.