21-09-2017, 11:55 AM
As cloud computing becomes prevalent, sensitive information is increasingly centralized in the cloud. For the protection of data privacy, sensitive data have to be encrypted prior to outsourcing, making efficient use of data a very difficult task. Although traditional search encryption schemes allow users to securely search for encrypted data through keywords, these techniques only support Boolean searches, without capturing any relevance to the data files. This approach suffers two major drawbacks when applied directly in the context of Cloud Computing. On the one hand, users, who do not necessarily have a prior knowledge of the encrypted data of the cloud, have to register each recovered file to find the ones that most coincide with their interest. On the other hand, invariably retrieving all the files containing the queried data. The keyword also incurs an unnecessary network traffic, which is absolutely undesirable in the current pay-per-use cloud paradigm. In this article, for the first time, we define and solve the problem of finding effective and safe keywords above the encrypted data of the cloud. The classified search greatly improves the usability of the system by returning the matching files in a ranked order against certain relevance criteria (for example, keyword frequency), thus making a further step towards the practical deployment of data hosting services that preserve privacy in Cloud Computing. First of all, we offer a simple yet ideal keyword search under the definition of state-of-the-art symmetric search encryption (SSE) security and demonstrate its inefficiency. In order to achieve a more practical performance, we propose a symmetric cryptographic definition, and give an efficient design through the proper use of the existing cryptographic primitive, symmetric encryption order preservation (OPSE). An in-depth analysis shows that our proposed solution enjoys the security assurance "as strong as possible" compared to the previous SSE schemes, while also correctly accomplishing the goal of finding classified keywords.