Vol.48 (CIA 014), pp.103-107 http://dx.do.org/10.1457/astl.014.48.18 Watermar-based Provable Data Possesson for Multmeda Fle n Cloud Storage Yongjun Ren 1,, Jang Xu 1,, Jn Wang 1,, Lmng Fang 3, Jeong-U Km 4 1 Jangsu Engneerng Center of Networ Montorng,Nanjng Unversty of Informaton Scence & Technology, Nanjng 10044, Chna Computer and Software School, Nanjng Unversty of Informaton Scence & Technology, Nanjng 10044, Chna 3 Computer and Software School, Nanjng Unversty of Aeronautcs and Astronautcs, Nanjng 10044, Chna 4 Department of Energy Grd, Sangmyung Unversty, Seoul 110-743, Korea Abstract. In cloud computng, data owners host ther data on cloud servers and data consumers can access the data from cloud servers. Due to the data outsourcng, however, ths new paradgm of data hostng servce also ntroduces new securty challenges, whch requres an ndependent audtng servce to chec the data ntegrty n the cloud. However, the exstng solutons are not specfc to the multmeda data. Moreover copyrght protecton s not provded. In ths paper, we defne specally a provable data possesson model for multmeda fle, and present a framewor based on dgtal watermarng for multmeda data storage audt servce, n whch we defne the securty features of audt servce for multmeda data outsourcng and the correspondng propertes of dgtal watermarng. Keywords: Cloud computng, Data storage audtng, Provable data possesson 1 Introducton Cloud Computng has been envsoned as the next-generaton archtecture of IT enterprse. Storng the data n cloud envronment becomes natural and also essental. Cloud s a large shared resource pool and the users move towards them wth respect to ther needs. It offers amentes for data storage, data access and other computatonal capabltes n a relable manner. But, securty becomes the major concern for all enttes n cloud servces. In order to solve the problem of data audtng servce, many schemes are proposed under dfferent systems and securty models [1-13]. Exstng solutons are not specfc to multmeda data. Besdes, copyrght protecton s not provded. In ths paper, we present an effcent watermarng-based audt servce for mage outsourced storages. Our audt system uses self-embeddng watermarng to provde mage content audt and support authentcty and ntegrty for mage content n cloud computng. In addton we also provde the ablty of copyrght protecton for mage by usng self-embedded technology. ISSN: 87-133 ASTL Copyrght 014 SERSC
Vol.48 (CIA 014) Related Wor The message authentcaton code (MAC) s a nd of hash functon whch has been used for checng the data ntegrty for a long tme. Some solutons usng MAC for storage audtng servce have been proposed. Based on the pre-computed MACs stored on the verfer, the protocols proposed by Lllbrdge et al.[] and Naor et al. [3]can detect any data loss or corrupton wth hgh probablty. Shacham et al. [4] proposed a MAC-based batch verfcaton for multple data blocs. In 007 Atenese, et al [5] proposed a PDP model to solve the storage problems of fles. They dvded the fle nto blocs, and computed a homomorphc tag [6] for each bloc, completed the proof of the data ntegrty by samplng and verfyng the correspondence of the tags and blocs randomly. A.Juels, et al [7] proposed a provable data recovery (POR) model. Instead of taggng fle blocs, they nserted some sentnel blocs, and verfed the ntegrty of the fle by checng the correctness of sentnel blocs. For the sentnel blocs are one-tme labels, the number of tmes that the fle can do ntegrty verfcaton s lmted, related to the number of sentnel blocs. Havav Shacham and Brent Waters [4] proposed an mproved POR model under the securty model defned n [7], and had a very complete proof. They used tags smlar to [5], and appled to publc authentcaton. Kevn D. Bowers et al [8] and Yevgeny Dods et al [9] made some theory and applcaton extensons based on [4][7]. Zheng and Xu also present a dynamc POR model n [10].PDP model proposed n [5] only appled to prvate authentcaton. It meant that only the person who has the prvate ey can verfy the ntegrty of the fle. Atenese mproved PDP model to apply to publc authentcaton n [11]. They replaced the homomorphc tags n [5] wth homomorphc tags supported publc authentcaton [1]. 3 Self-Embeddng Watermar-Based PDP 3.1 Watermar generaton and embeddng Let the orgnal mage (Q) s N N. In cloud mage owners mae the followng calculaton to generate watermar for mage audtng. 1.The orgnal Q s dvded nto n n sub-bloc wthout overlap. Each sub-bloc s recorded as 1,,, ( N / n )..The least sgnfcant m bts of each pxel of sub-bloc s recorded as BtSet (, j,0 ), s the seral number the sub-bloc, and are set nto zero. And the new. The process can be mared as the equalty., j 1,,, m, BtSet (.) s settng functon. 104 Copyrght 014 SERSC
Vol.48 (CIA 014) 3. Computng sngular value decomposton (SVD) of.the generated sngular value s recorded as, s the seral number of sngular value, and 1,,, n. 4. Accordng to the followng formula to compute the norm of sngular value for,.e. Norm. N o rm n ( ) (1) 1 5.Extracton the party of the hghest order for robust watermar W. If the party of the hghest order for W 0 ;otherwse W 1, W s the -th bt of W. 6.Embeddng gettng sub-bloc Norm and produce orgnal s even, N o rm W to the least sgnfcant m bts of every pxel for wth watermarng. and s restructured and produce an mage ( Q ) ncludng watermar. The self-embeddng process can be expressed as: BtGet (, j, W ) From the above mentoned we now that the length of W s embedded to number of self-embedded watermarng s every pxel for () ( N / n ) bts, W s n pxels and every pxel s embedded by m bts. So the total bt W s mn and the embeddng capacty of the mn.because W s embedded nto the least sgnfcant m bts of, whch lead to mall change, the nvsblty of watermar algorthm s very good. Moreover the watermarng W s generated by the characterstcs of the orgnal mage and has the robustness aganst the attac, whch s called robust watermarng. After the above operaton, MDO wll be able to upload the mage Q wth watermarng to CSP for cloud storage servce. 3. Copyrght protecton of cloud mage servce from robust watermarng Image audtor MDA download mage Q from cloud servce provder. And then t extracts the robust watermarng to dentfy copyrght of the mage. 1. Image Q s dvded nto n n sub-bloc wthout overlap. Each sub-bloc s recorded as and 1,,, ( N / n )., s the seral number the sub-bloc, Copyrght 014 SERSC 105
Vol.48 (CIA 014).The least sgnfcant m bts of each pxel n sub-bloc s recorded as BtSet (, j,0 ) are set nto zero. And the new. The process can be mared as the equalty: 3.Computng sngular value decomposton of value s recorded as, j 1,,, m, BtSet (.) s a settng functon..the generated sngular, s the seral number of sngular value, and 1,,, n. 4.Computng the norm of sngular value for Norm n 1,.e. Norm. ( ) (3) 5.Extracton the party of the hghest order for watermar Norm and produce robust W. If the party of the hghest order for Norm s even, W 0 ; otherwse W 1. W s the -th bt of W. 6.Calculatng and analyzng the NC dentfy copyrght of W and W N C ( ( W W )) / ( W W ) 3.3 Integrty and authentcty verfcaton of mage content Image audtor download mage Q from cloud servce provder and mplement the followng operaton to verfy the mage content. 1. Image Q s dvded nto n n sub-bloc wthout overlap. Each sub-bloc s recorded as and 1,,, ( N / n )..The least sgnfcant m bts of each pxel n.e. L j ( r ) B tg e t ( A B, j ), s the seral number the sub-bloc, are extracted, j L r s ;where B tg et (.) s extracton potental functon. ( ) the bt sequence of the mnmum j bts for each pxel n the -th sub-bloc. j 1,,, m, and r 1,,, n. j 3. Contrastng the extracted robust watermar W and L ( r ). Image audtors verfy the consstence between them. If any one bt for every j and r s not consstent, the sub-bloc content has been altered; otherwse the mage Q s ntact. 106 Copyrght 014 SERSC
Vol.48 (CIA 014) 4 Conclusons In ths paper, we present an effcent watermarng-based audt servce for mage outsourced storages. We ntegrate mage content audt servce and copyrght protecton through a double functon self-embedded watermarng scheme, whch greatly reduced the consumpton of resources and s very sutable for multmeda data n cloud envronment. Acnowledgments. Ths wor was supported by the Industral Strategc Technology Development Program (10041740) funded by the Mnstry of Trade, Industry and Energy (MOTIE) Korea. It was also supported by the Natural Scence Foundaton of Jangsu Provnce (No. BK01461). Prof. Jeong-U Km s the correspondng author. References 1. Yang, K., Ja, X.: Data storage audtng servce n cloud computng: challenges, methods and opportuntes. The journal of World Wde Web. July 01, Volume 15, pp 409-48.. Lllbrdge, M., Elnety, S., Brrell, A., Burrows, M., Isard, M.: A cooperatve nternet bacup 3. Naor, M., Rothblum, G.N.: The complexty of onlne memory checng. In: Proceedngs of the 46th Annual IEEE Symposum on Foundatons of Computer Scence, FOCS 05, pp. 573 584. 4. IEEE Computer Socety, Washngton, DC, USA (005) H. Shacham and B. Waters. Compact proofs of retrevablty. In ASIACRYPT '08, pp. 90-107, 008. 5. G. Atenese, R. Burns, R. Curtmola, J. Herrng, L. Kssner, Z. Peterson, and D. Song. Provable data possesson at untrusted stores. In CCS '07, pp.598-609, 007. 6. R. Johnson, D.Molnar, D. Song, and D. Wagner.: Homomorphc sgnature schemes. In Proc. of CT-RSA, volume 71 of LNCS, pp. 44-6, 00. 7. A. Juels and B. Kals. PORs: Proofs of retrevablty for large fles. In CCS '07, pp.584-597, 007. 8. Bowers, K., Juels, A. and Oprea, A.: Proofs of retrevablty: Theory and mplementaton. Techncal Report 008/175, Cryptology eprnt Archve, 008. 9. Y. Dods, S. Vadhan, and D. Wchs. Proofs of retrevablty va hardness applcaton. In TCC, vol.5444 of LNCS, pp. 109-17, 009 10. Zheng, Q. and Xu, S.: Far and Dynamc Proofs of Retrevablty. CODASPY 11, February 1 3, 011, San Antono, Texas, USA. 11. G. Atenese, S. Kamara, and J. Katz. Proofs of storage from homomorphc dentfcaton protocols. ASIACRYPT 09, LNCC, 009. 1. D Boneh, B Lynn, H Shacham. Short sgnatures from the wel parng. ASIACCRYPT 001. LNCS, vol. 48, pp. 514-53, 001. Copyrght 014 SERSC 107