Research On Lossless Data Hiding Algorithms In Encryption Domain

With the development of information technology,database outsourcing service is gaining popularity.In the outsourced database mode,an individual or organization outsources its own database business to a database service provider.The outsourcing service provider provides remote database services for data owners and database users,thus avoiding repeated human and material investments in database management for each individual or organization and providing better maintainability and scalability.However,since the database services provided by third parties are not completely trusted,the outsourced database model has to face a series of security issues such as privacy leakage,data theft,malicious tampering,and so forth.Cryptography,which converts original data into unintelligible encrypted data,is an extremely effective method for privacy protection.In the outsourced database mode,in order to avoid exposing private data content directly to third-party database servers,the data owner can encrypt the original data before uploading it.The third-party database server can only access unintelligible ciphertext,so the risk of data privacy leakage will be greatly reduced.Moreover,watermarking,which embeds meaningful data into cover media by introducing minor modifications,can provide solutions for authentication,fingerprinting and copyright protection during distribution of digital data.After encrypting the data,the data owner may embed a watermark into the encrypted data for the purposes of fingerprinting and copyright protection.For the requirements of management,fingerprinting or integrity verification,the third-party database server may also embed the corresponding watermark into the encrypted data uploaded by the data owner.In the outsourced database mode,the combination of cryptography and data hiding can significantly improve the security of the outsourced database.Note that the data owner still expects the original data to be perfectly restored,so data embedding does not affect the decryption of ciphertext.Data embedding needs to be performed with a reversible or lossless manner.In this dissertation,we propose a lossless data hiding algorithm based on orderpreserving encryption for massive and dynamic modern relational databases.The database data hiding algorithm modifies the parity of encrypted attribute values to embed secret bits and adopts the manner of grouping repetitive embedding to provide strong robustness based on the majority voting mechanism.Since the random number is introduced in the encryption algorithm,the parity modification is equivalent to modifying only the random number.The data embedding does not affect the decryption of encrypted data.Direct decryption of an encrypted database containing embedded data completely restores the original data.The experimental results and analysis show that compared with the existing algorithms,the proposed algorithm has stronger robustness,higher embedding rate,lower computational complexity,no auxiliary information and no distortion.In addition,since digital images occupy a large share of the uploaded data,many reversible data hiding algorithms for the encrypted digital image have been proposed.For static digital images,we propose a lossless data hiding algorithm based on homomorphic encryption.The data hiding algorithm uses the homomorphic and probabilistic properties of the homomorphic encryption algorithm to embed secret data.No distortion was introduced throughout the data embedding process.Directly decrypting an image containing embedded data is the same as the original image.The experimental results and analysis show that compared with the existing algorithms,the proposed algorithm has higher embedding rate,lower computational complexity,less auxiliary information and no distortion.