| Abstract | In the last few years, businesses offering digital contents (such as music and video) via internet have come to be established with the development of broadband networks. At the same time, to adapt to the diversification of people's preferences, television broadcasts tend to use multi-channel broadcasting, which leads to fewer viewers on one channel. Hence, it is conjectured that free broadcasting based on advertisement revenue will become difficult. Under such circumstances, live pay-television broadcasts via internet, i.e., internet-based pay broadcasting is becoming a promising business.
To get internet-based pay broadcasting implemented, it is necessary to protect distributed contents from illegal copying and redistributing after they are accessed. Hence, fingerprinting system is a useful tool for it.
Fingerprinting schemes embed unique user information (e.g., ID or fingerprints) into each user's copy and if an illegal copy appears, user information can be extracted to help trace or identify illegal users. So, the goal of digital fingerprinting is to deter or discourage people from illegally redistributing the digital data that they have legally purchased. The major challenge to fingerprinting is collusion attacks from illegal users. A collusion attack is a cost-effective attack, where colluders (illegal users) combine several copies with the same content but different fingerprints to try to remove the original fingerprints or frame innocent users.
It is necessary to study what kind of fingerprinting code is appropriate for our internet-based pay broadcasting system and what kind of fingerprinting code is efficient and effective to protect distributed contents from illegal copying and redistributing. For superior scalability, the most widely used model is a multicast distribution model. That is, we send same digital content to all subscribers, and then at the user end, build a uniquely fingerprinted content. So the fingerprinting code has to manage a large amount of users, and at the same time it must be robust against collusion. These are important from a business and economic point of view.
In this paper, we first show that the anti-collusion code (ACC) has advantages over other existing fingerprinting codes on live pay-television broadcasting. Based on several survey results on existing fingerprinting codes, ACC is considered to be superior in terms of efficiency and effectivity. The major reasons consist in that ACC adapts to multimedia data, identifies all colluders and is shorter in code length and larger in code size than traceability code in some sense. Here, code efficiency is referred to as the number of users that can be supported by code length. Code effectivity is defined as a ratio between the maximum tolerated collusion size and the total number of users, which describes the resistance against collusion attacks.
Next, we devote to presenting how to achieve and construct efficient and effective ACC codes since there is no practical construction of ACC by now.
1. We examine the known balanced incomplete block design (BIBD) examples and demonstrate that efficient and effective ACC codes can be derived from unital and affine plane. Then we investigate their properties, i.e., unital has higher efficiency but lower effectivity, while affine plane has higher effectivity but lower efficiency. In other words, unital supports more users but exhibits weaker resistance, while affine plane exhibits stronger resistance but supports fewer users.
2.Meanwhile, performances of the ACC codes generated by unital and affine plane are evaluated from three aspects: the number of users that can be supported, collusion size and facility of implementing.
For example, assume that both unital and affine plane support 10000 users, the same amount of users, then the collusion size of affine plane is about 100. It can be considered secure enough due to the difficulty to bring together 100 illegal users. Similarly, assume the collusion sizes of affine plane and unital as about 100, then unital can support about 100,000,000 users. This is amazing ability to our live multicasting.
Last, we give out the practical explicit constructions for ACC codes derived from unital and affine plane.
In brief, we contribute in two aspects:
1.We have shown that anti-collusion code has advantages over other existing fingerprinting codes on live pay-television broadcasting, based on the analyses of survey results.
2.We first introduced affine plane and unital into fingerprinting code design. By investigating their properties and evaluating their performances, we have present how to generate efficient and effective anti-collusion codes. Finally, we have given out their practical explicit constructions, which are desirable from a users・point of view. |