Research On Asymmetric Digital Fingerprinting Technology

With the rapid development of digital and network technologies, all kinds of digital contents are distributed and shared over the Internet widely. At the same time, piracy becomes increasingly rampant as the dishonest customer can easily duplicate, modify and redistribute the digital content to a large audience illegally. How to protect the related digital copyright, ensuring the proper distribution and usage of digital product has become increasingly critical under the Internet environment. Encryption can only provide digital data with the desired security during transmission, once the digital data is decrypted, the dishonest customer can redistribute it arbitrarily. The traditional digital watermarking is a relatively passive digital copyright protection technology, which lacks an efficient mechanism for tracing traitors. When a pirated copy is found, digital watermarking cannot find the original buyer who makes a piracy and hence, it cannot prevent the digital product from being duplicated and redistributed by the dishonest customer.As a new efficient technology for protecting digital copyright developed in recent years, digital fingerprinting achieves piracy tracing by embedding a unique serial number into each distributed digital copy secretly. The embedded unique serial number is called digital fingerprint and associated with the corresponding customer. Once an unauthorized content copy is discovered somewhere, the embedded fingerprint can uniquely identify the source of the pirated copy. Digital fingerprinting has become an important tool of digital copyright protection and received much attention. In order to guarantee the fairness between merchant and customer, enabling the dishonest merchant cannot frame the innocent customer while the cheating customer cannot deny his responsibility for a copyright violation that caused by him, the most effective way is to make the fingerprinting scheme is to be designed asymmetric, that is called asymmetric digital fingerprinting.Designing secure and efficient asymmetric digital fingerprinting schemes is the basis for the applications of digital fingerprinting technology. The high computational complexity, high burden of the merchant and high communication overhead of the existing asymmetric fingerprinting schemes deters their practical application. This dissertation combines cryptography, digital watermarking, image processing, signal detection and coding theory to systematically study the asymmetric fingerprinting technology, and proposes three efficient asymmetric fingerprinting schemes. The main contributions of this dissertation are as follows:(1) A1-out-of-n oblivious transfer based asymmetric fingerprinting is proposed. In this scheme, the merchant transmits a same encrypted copy to multiple customers and, each customer gets a corresponding decryption key from the merchant by a key distribution protocol that implements the1-out-of-n oblivious transfer. The generation process of the fingerprinted copy is completed at the client end through decryption, and different decryption keys will produce different fingerprinted copies. Since the merchant only needs to generate a same encrypted copy for multiple customers, which not only reduces the merchant’s burden, but also reduces the latency time. In addition, since the merchant only needs to transmit a same encrypted copy to multiple customers, multicast that is an efficient transport mechanism for one-to-many communication can be exploited, which can reduce the bandwidth requirement significantly.(2) In these existing homomorphic public key encryption algorithm based asymmetric fingerprinting schemes, the homomorphic public key encryption algorithm is directly used to encrypt the multimedia content, and then it not only causes a high computational complexity, but also causes a high communication cost. Based on the first asymmetric fingerprinting scheme proposed in this dissertation, a new homomorphic public key encryption algorithm based asymmetric fingerprinting scheme is proposed, where the issue of asymmetric fingerprinting is transformed into the issue of the asymmetric distribution of the decryption key. Differing from the existing homomorphic public key encryption algorithm based asymmetric fingerprinting schemes, the homomorphic public key encryption algorithm is only used to construct the key distribution protocol and digital signature, not perform on the multimedia content, and then reducing the computational complexity and communication cost from previous schemes.(3) A l-out-of-2oblivious transfer based asymmetric fingerprinting is proposed. In order to reduce the computational complexity and communication overhead, the merchant uses the selective encryption algorithm to encrypt the media content and transmits a single encrypted copy to multiple customers by multicast. Each customer is assigned an incomplete decryption key, and some components will be not decrypted correctly after decryption, which can be considered as the fingerprint information and used to identify the corresponding customer. To achieve the asymmetric property, each decryption key is composed of two parts, one is generated by the merchant and the other one is generated through a key distribution protocol that implements the l-out-of-2oblivious transfer.The proposed asymmetric fingerprinting schemes in this dissertation can be used to protect digital copyright in different environments, such as digital library, online press, and online audio/video on demand.