A Dynamic and Scalable Encryption Framework with Similarity Awareness for Cloud Data Storage
This paper proposes a complete file management system that includes a desktop application that utilizes chaos-based cryptography to provide security for the files. This system utilizes a method of using a Chaotic Encryption Algorithm instead of using traditional ciphers so that files are encrypted using mathematical transformations that are unpredictable. To determine similarities between files, the system uses MinHash combined with Locality Sensitive Hashing (LSH) to quickly and accurately identify near-duplicate files. A Java-based server is used to provide the infrastructure necessary to support back-end processes; the secure nature of the system is maintained while also providing reasonable performance when under a heavy workload. The system employs natural language processing techniques, including using Stanford NLP tools to part of speech tag the text before encryption occurs. The user IDs of all users, document metadata, and history records of all documents are stored in a MySQL database. The Java Swing application JAVA Swing GUI is used by the end-user to allow them to upload, encrypt, decrypt, and set access control on the documents/folder(s). The results of the experiment support the fact that the Chaotic Encrypting System has high encryption accuracy, reliable decryption performance, 94.7% similarity detection accuracy for both near-duplicate and completely different documents, and 93.5% detection efficiency for duplicate and non-duplicate documents.