Collusion attack is a very effective attack for digital fingerprinting system. In order to remove or attenuate the fingerprint information hidden in fingerprinted content, a number of users produce a new colluded copy through their own fingerprinted copies. In this paper, we address a novel collusion-resisting desynchronization fingerprinting approach using social network analysis. The novelty of this paper is that collusion attack occurred in a multimedia social network community with high

more »

... ability. Different from all existing works, with desynchronization model constructed upon social network, the original image is desynchronized to get many similar copies which are different from each other, and then they are assigned and distributed to subscribers according to social network analysis. Theory analysis shows that the presented desynchronization distribution method has significantly better performance than those existing distribution schemes. The experimental results also show that the average colluded images even with only two desynchronized copies have poor visual quality. And the visual quality of colluded images does not improve apparently with the increase of the number of copies. The copy detector then extract feature of the suspected one, and then compares the image feature with the original image feature stored in the index database [10] . Different copy detections to the illegal distribution have been proposed in the past few years [9] [10] [11] [12] [13] [14] [15] . Information hiding is another technology to protect copyright. Digital multimedia is increasingly popular in multimedia social network. With watermark embedding technology, mark information, which can help to prevent unauthorized distribution directly [16] . The embedded marks are known as digital watermarking academically. Digital watermarking, which is alternative approach to copyright enforcement, must be inserted into the original content as copyright marking before copies are made. On the other hand, due to the wide popularity of sharing multimedia in social networks, security and privacy protection has become very important. Digital watermarking can be used to protect ownership, copyright authentication, and digital fingerprinting [17] . For copyright authentication, various digital watermarking techniques [16] [17] [18] [19] [20] [21] [22] have been adopted to deter misuse of multimedia data in multimedia social network. Now the usage of digital watermarking should robust to resist common geometric transformations and some signal processing operations [18] , which should not damage the watermark information. Even if the most robust digital watermark can not trace the pirate who shares the multimedia content illegally in social network, in order to provide solution, a special watermarking scheme, fingerprinting, could be applied. Digital fingerprinting is a technology which can be used to trace the illegal redistribution in fingerprinting system, where digital fingerprints are embedded into the multimedia content as the ID. However, collusion attack is the main challenge that digital fingerprinting technology has to face, because a set of users can produce a new copy with their own fingerprinted copy, the fingerprint information hidden in the colluded copy can be removed, in this case, the detector will not detect the users who redistributed the copy in social network. As we know, these colluders must know each other or come from a same social network, so they could format a special crime unit. Inspired by this scenario, our multimedia fingerprinting will protect the multimedia content from redistribution based on social network analysis. In multimedia social network, those who take part in collusion attack must be having some social relationship. The community structure can be identified with social network analysis [27] . Therefore, colluders must come from the same community, and they will decide to choose which collusion attack to get the max profit. Digital fingerprinting should resist collusion attack to prevent redistributing of the fingerprinted copy [35] . In this work, we address a novel desynchronization fingerprinted multimedia distribution scheme using social network analysis to combat average collusion attack. This is the first time the desynchronized fingerprinted content is distributed with social network analysis. The related works are introduced in section 2, and followed by explain of social network in section 3. In section 4, we show the desynchronization distribution method and the performance of the scheme. In Section 5, we show the experimental results. Section 6 concludes this paper and show the future research. Copyright ⓒ 2016 SERSC 217 network analysis. Our approach is based on the subscriber-specific need of image and social network analysis. Each uniquely preprocessed copy of the host-signal prior to be assigned according social network analysis. The proposed scheme can be applied to multimedia social network, in which the proposed scheme can resist average collusion attack even if there is a large number of users in social network.