隨著網際網路快速傳播的特性,目前有許多雲端業者提供雲端平臺給使用者使用,使用者只要上傳所擁有的數位內容,即可輕易地使用各式各樣的雲端服務,為了保護數位內容,我們會將多媒體加密後以密文的方式傳播,但是,在數位驗證、資料探勘分析等應用下,訊號處理的第三方必須得事先知道數位內容的明文,才能進行數位驗證、資料分析等訊號處理;然而,我們不一定信任第三方,我們不想讓訊號處理的第三方知道具有隱私的數位內容,例如:身體健康資訊、商業機密等,希望第三方可以在不知明文下進行數位驗證、資料分析等應用,因此,假設數位內容能夠先被加密後再進行訊號處理與分析,這樣一來,就不必擔心洩漏具有隱私的數位內容問題。有鑑於此,為了保護具有隱私的數位內容,使得第三方能夠在不知原始內容下進行數位驗證、資料檢索等應用,我們將針對加密多媒體,提出數位驗證之分離式可逆浮水印技術與具隱私保護的資料檢索技術之研究主題。 本計畫基於公開金鑰加密系統,第一年提出加密域之數位驗證研究,我們提出了分離式可逆浮水印技術,利用機率安全性特性,浮水印取出與內容解密可分開獨立進行。第二年提出加密域之資料分析技術,利用同態加密性特性,在不知原始內容下,進行具隱私保護之資料檢索。與其他使用串流加密的方法相比,此計畫可在不降低安全層級下應用於雲端服務,提供具隱私保護之訊號處理。
With the Internet distribution for digital media, many cloud service providers provide cloud-based storage services that allow users to upload data into cloud storage. However, the owner may not trust the cloud service provider, which disposed the owner to encrypt the media before sending it to the cloud. Traditional authentication and data analytics are usually processed before encryption or after decryption. Hence, the ability to directly process the encrypted signal while keeping the plain text unrevealed is desired. As a result, in this proposal, for encrypted multimedia we propose reversible watermarking applied for authentication and the privacy-preserving data analytics. Based on public key cryptography, in the first year, we propose a separable reversible watermarking for encrypted data authentication. At the receiver side, the data extraction and data decryption are separable based on probabilistic property. In the second year, we focus on privacy-preserving data analytics. We propose the privacy-preserving retrieval by using homomorphic cryptosystem. Compared with other schemes using a cipher stream to encrypt the data, the proposal is more appropriate for cloud services without degrading the security level.