Research Of Cloud Database Security Based-on Digital Watermark Technique

Cloud database is a network service which possesses key functions like dataindex and inquiry. Compared with traditional databases, it possesses particularfeatures, its design is simple, and there is no need for prophase investment as well asthe maintenance and management of the database administrator, it is maintained bythe third-party which provides service. Thus, this pattern reduces the burden ofmanagement relating to data modeling, index maintenance and performanceimprovement. Due to this, it is favored by increasing users. However, there are alsosome drawbacks. In cloud database, users no longer own the infrastructure resourcesand business datum are stored in the "cloud". It is convenient to access, copy ordisseminate the data, but making it hard to trace. This provides favorable conditionsfor data piracy and illegal dissemination, which threatens the owner’s legitimate rightsand interests. Therefore, copyright protection of cloud database becomes an urgentand realistic problem. Digital watermark, as a kind of common security technique forprotecting digital information, becomes a potential solution to the problem of clouddatabase copyright protection.Compared with relational database watermark, cloud database is commonly usedto store huge amounts of data, requiring highly efficient watermarking algorithms;Data stored in cloud database is more various than that of the relational database,leading to the generality decline of watermarking algorithms; Cloud databases adopt"scattered file" storage technology which induces the desynchrony of embedding anddetection. Thus, it needs better synchronization mechanism to ensure the safety ofwatermark. These put forward higher request on cloud database watermarkingresearch. Cloud database watermarking, as a brand new research field, is still in itspreliminary stage. There is still much exploration work needed to be done.This thesis introduces related foundation as well as theory of cloud database anddigital watermark, analyzes existing security issues of cloud database and presentstwo digital watermarking algorithms for cloud database security: cloud databasewatermarking algorithm based on polar angle expansion and cloud databasewatermarking algorithm based on quantum k-means clustering. The former algorithmis a reversible watermarking. First, it adopts key as a seed of pseudo-random numberto generate pseudo-random numbers. Second, it uses these numbers to select theposition of watermark embedding, then it maps these attributes to polar coordinates one by one, and embeds watermark into those extending polar angle of point sequence.The latter, it views database table as multidimensional space. On the basis of this,quantum k-means clustering is implemented. Each class represents a specific meaningof the data, and then for each type of data, it embedding specific bits repeatedly toimprove the robustness of watermark. Meanwhile, to ensure the results of dataclustering are the same in before and after the watermark is embedded, adaptivewatermark embedding is realized. Comparative experiments show that these twomethods possess high robustness for subset change, subset selection, and subsetadding attacks, and the former can truly recover the original data.