Research On Some Security Problems Of Buyer-Seller Watermarking Protocols

With the rapid development of multimedia technologies and the wide deployment ofbroadband networks, an increasing amount of multimedia data are distributed through net-works. This introduces an urgent need to protect the proper distribution and use of multime-dia content, especially in view of the ease of copying and manipulating digital multimediadata. In an open network environment, however, people come to realize that it is almostimpossible to strictly prevent pirates from committing piracy, and thus the research focuseson how to effectively trace the source of illegal copies. Digital fingerprinting is an emerg-ing technology of digital copyright protection due to this requirement. In the early stage ofdigital fingerprinting schemes, the major concern is to obtain collusion-resistant fingerprintcodes and the corresponding trace mechanisms by encoding the customer information. Allthese schemes assumed that sellers are always trustworthy and allowed them to generate thefingerprinted copies. Because both the sellers and the buyers have access to the fingerprintedcopies, it is impossible to determine who should be responsible for it when an illegal replicawith a user's fingerprint is found. For this case, different cryptographic techniques were in-troduced into digital fingerprinting to regulate the transaction behavior of the sellers and thebuyers. This makes a new stage in the research of digital fingerprinting and forms a digitalfingerprinting protocol with the ability to truly achieve the goal of copyright protection.Buyer-seller watermarking protocols are a kind of digital fingerprinting protocols basedon homomorphic encryption schemes. They have attracted considerable attention for theirsimplicity and practicality and lots of works have been done in this area. In general, ex-cept for the above so-called customer's right problem, a secure buyer-seller watermarkingprotocol should consider the piracy tracing problem, unbinding problem, double embeddingproblem, secure verification problem, conspiracy problem, anonymity problem, etc. Thisthesis focuses on these security problems and aims to design some more reasonable, secure,and efficient buyer-seller watermarking protocols in light of the shortcomings and problemsexisting in the previous ones. The main contributions of the thesis are as follows: 1. The design and analysis of buyer-seller watermarking protocols without double em-bedding. We first investigate the research situation of the double embedding problemin buyer-seller watermarking protocols. Then, based on the famous Lei et al.'s scheme,we propose a new buyer-seller watermarking protocol without double embedding. Inour new scheme, the seller only needs to embed one watermark in the encrypted do-main. If the seller finds an unauthorized copy later, the embedded watermark can beutilized as an index to locate the pirate in the encrypted domain. Then the seller cantake the same watermark as an undeniable proof against the pirate. We examine howthe design goals are achieved and discuss some issues related to practicality. Comparedto the original Lei et al.'s scheme, the new scheme not only retains several advantagesof its predecessor, but also provides more reasonable and efficient solutions to thedouble embedding problem and the unbinding problem.2. The design and analysis of buyer-seller watermarking protocols suitable for proba-bilistic privacy homomorphisms. We analyze the weaknesses of deterministic pri-vacy homomorphisms in depth and point out two common problems with two fa-mous schemes. We then improve the two schemes. The comparison between the newschemes and the original ones shows that the former not only successfully remove therestriction of deterministic privacy homomorphisms, but also solve the secure verifi-cation problem existing in the latter. Our new schemes are the first ones in the existingliterature that solve the two problems simultaneously.3. The design and analysis of based on zero-knowledge watermark detection (ZKWD)buyer-seller watermarking protocols. Based on Adelsbach-Sadeghi ZKWD protocol,we propose the first buyer-seller watermarking protocol with the ability to truly solvethe secure verification problem in zero-knowledge information leakiness.4. The design and analysis of buyer-seller watermarking protocols suitable for web con-text. We first analyze Frattolillo's watermarking protocol in detail and point out its twoserious problems. We then propose a new web-oriented and anonymous buyer-sellerwatermarking protocol based on Frattolillo's scheme. By detailed security analysisand discussion, we show the new scheme not only solves the problems existing in theoriginal Frattolillo's scheme, but also achieves the main goals of the original one. Thenew scheme is a truly efficient and web-oriented buyer-seller watermarking protocol.