Research On Key Technologies Of Function-oriented Secure Search Over Encrypted Cloud Data

With the widespread use of cloud computing and cloud storage technologies,more and more data owners tend to outsource their data to remote cloud servers to reduce computing and storage overhead.But untrusted cloud service providers can put data at serious security threats.A common protection method is to encrypt the data before it is outsourced.However,this will make it difficult to perform search operation over the encrypted data,and thus the searchable encryption technology emerges.However,the existing searchable encryption scheme only supports basic function,the efficiency and accuracy can be improved,and the security needs to be further enhanced.In view of the above problems,this paper focuses on the key technology of function-oriented search over encrypted cloud data,starting from exact search and fuzzy search scheme,and studies on the completeness of functions,optimization of performance and improvement of security.The main contributions of this thesis are included as follows:(1)Aiming at the search requirements that can simultaneously support high performance and multiple functions,a multi-keyword ranked search scheme with high accuracy that supports results verification and dynamic update is proposed.In view of the existing search scheme,the TF-IDF algorithm only considers the number TF of a certain keyword in the document and the total number of documents IDF of the keyword in the file set,while ignoring other factors which will effect the relevance of the keyword and document,such as the location of the keyword and the length of the document.Therefore,the paper realize a multi-keyword ranked search scheme with high accuracy and high efficiency by combining keyword location and text length with traditional TF-IDF algorithm,and basing on the construction of index tree that can optimize search efficiency.Finally,based on the basic structure of the index tree,a search scheme with results verifiable and dynamic update is designed.The files in the experimental test dataset are composed of paper data containing fixed titles,abstracts,texts,conclusions,and references.The keywords have different weights in each part,thus reflecting the impact of the keyword position and text length on text relevance.Finally,the performance and safety of the scheme are proved by experiments and theoretical analysis;(2)Aiming at the requirements of search function and results security for the fuzzy search,a multi-keyword fuzzy search scheme that supports condition query and can resists ciphertext-only attack is proposed.According to the security flaw of the fuzzy search scheme based on bloom filter under only ciphertext attack proposed by Lin et al.[1],and the scheme can not support condition query,the paper designs a basic fuzzy ranked search scheme by random number redundancy confusion based on the scheme proposed by Wang et al.[2].This scheme realizes the confusion of original keyword features and makes the search pattern protected,and thus resisting the ciphertext-only attack.The paper further expands the basic scheme,and realizes widely-used condition query which can support “AND”,“OR”,“NOT” operations under fuzzy search.Through these operations,the user can customize the query for files containing certain keywords,or files that do not include certain keywords,so as to obtain the query results more in line with users' requirements.Finally,the paper theoretically analyzes the security of the scheme and verifies the function and performance of the scheme by experimenting on the actual text dataset.