Homomorphic Token and Distributed Erasure-Code for cloud

Transcription

1 Homomorphic Token and Distributed Erasure-Code for cloud 1B.Anjani Kumar, 2K.Hari Prasad, 3C.Subash Chandra 1,3Dept. of CSE, Kakinada Institute of Engineering & Technology, Korangi, AP, India 2P.G.Department of CS, Ideal College of Arts & Sciences Vidyut Nagar, Kakinada, AP, India ABSTRACT: Cloud users store their data in the cloud environment which is privileged by the service provider. Cloud users can access their cloud storage. Here we propose a flexible distributed storage integrity auditing mechanism, utilizing the homomorphic token and distributed erasure-coded data. The proposed design allows users to audit the cloud storage with very lightweight communication and computation cost. The proposed scheme is highly efficient and resilient against Byzantine failure, malicious data modification attack, and even server colluding attacks. Keywords - Cloud computing, Data Integrity, Dependable Distributed Storage, Error Localization, Data Dynamics, Byzantine. I.INTRODUCTION Cloud computing is a recent prominent preference because to handle users data and applications. It offers computing through pool of resources over internet rather than on users local machines. In other words we can say that cloud computing is a new delivery model with a variety of scalable hosted services over internet. Cloud services offered are basically categorized into IaaS Infrastructure -asservice), PaaS Platform -as-a-service), SaaS software-as-a-service). There are several emerging trends in this domain with integration of services and technologies. With an increase in network bandwidth, a provision of flexible nature, reliable network connections leads to high quality remote service. Enterprises treat and store their data as files and but they are stored as objects in cloud. So a new data model is required and maintained as part of its scalable architecture. Related to cloud data storage. Cloud Users: Customers, who use the cloud storage provision to store their individual or enterprise data. Users are privileged data access to the cloud server for data management. Cloud server: It is maintained and managed by the cloud service providers for providing the storage space and computation resources to the users remotely. Third party auditor: who has capability to audit the cloud server based on the user request, if the users may not have trust-worthy data access to cloud. In the cloud environment, cloud users store their data into the set of cloud servers. These servers are running in co-operated, simultaneous and distributed manner. Users can communicate with cloud service providers to retrieve their data. In addition to it, users can perform operations on the cloud such as insert, append, delete and update. By using erasure technique, we can reduce the redundancy of the data to tolerate errors and/or even system crashes whenever user s data grow rapidly. In some cases, the user may need to perform block level operations on his data. We have recently different system and security models which high light the remote data integrity. These are useful when users possessing local data and focus on single scenario. In this paper, we propose an effective and flexible distributed storage correctness and availability of data which also support dynamic operations. As a complementary approach, researches have also introduced distributed protocols for storage correctness across multiple servers. It provides efficient storage verification and data availability. In this paper we propose effective and flexible distributed storage verification with dynamic data support. During the file distribution preparation we are using erasure correcting code, by using this. Byzantine Servers: The services provided by the internet must be available to the user without interruptions. The interruptions caused by mainly due to the operator mistakes, software bugs and malicious attack. Due to these we see sudden behavior in the system known as Byzantine faults tolerance.. Fig.1. Cloud storage service architecture Page 1069

2 Our mechanism also secures the localization of data errors i.e. identify the malware of servers. It also concentrates on dynamic data operations. we give an extension to our proposed main scheme to support third party auditing, so that users can safely assign their auditing tasks to third party auditors TPA). Our proposed scheme can be summarized in the 3 aspects as follows given below: 1) Our proposed system achieves strong assurance of storage correctness and data error localization, compared to previous techniques. 2) Not only for data integrity our proposed scheme also supports efficient and secure dynamic operations on blocks including insertion, deletion, append, modify etc. 3) Analysis shows that our scheme is against Byzantine failure, malicious data modification attack and server colluding attacks. II. PROBLEM STATMENT 2.1 System model Thereafter, the user interacts with the cloud servers via CSP to access for application purposes or to retrieve his data. In some special cases, the user may require need to perform block level operations such as insert, delete, update and append the block. Beyond this in our paper, we focus on the fileoriented cloud application other than non-file application data like social networking. however, we introduce TPA, for any leakages of user's outsourced data. 2.2 Adversary Model Security Threats in the cloud data storages are from two different resources. On the one hand, a cloud server be not trusted, may be self-interested, and cause malicious attacks. Meanwhile they move the data from that has not been or rarely accessed to a lower tier of storage than agreed, it may also lead to attempt or hide data loss incident due to management errors,byzantine failures and soon. There are two types of adversary. Weak Adversary: The adversary is interested in corrupting the user s data files stored in different individual servers. Strong Adversary: It is the worst case scenario, in which we assume that the adversary model 2.3 Design Goals Of The System We design the perfect mechanisms to support data verification and dynamic operations in cloud: i)correctness of stored data: we must provide the appropriate data and kept that undamaged in the cloud at all times. ii)fast localization of error data: we must efficiently modify the server even if the server failure occurs. iii)dynamic operations support for data: It supports dynamic operations if the users modify the data such as insert, delete, append. iv)dependability: It reduces the effect of data errors or server failures and protect the data against the byzantine failures, malicious data modification attack and server colluding attacks. iv)light-weight communication: It enables users to perform checks whether the data stored correctly with minimum expenses. 2.4 Notations and its importance F-is the stored file in the cloud. F is denoted with the matrix with equal sized data vectors consists of l blocks. These blocks are represented in the galois field GF2 p ) for p=8 or 16. A-the dipersal matrix used for Reed-Solomon coding. Dispersal matrix is also known as covariance matrix. G-the encoded file matrix,which includes set of n=m+k vectors,each consisting of l blocks. f key.) -pseudorandom function which is defined as f={0,1} * * key->gf2 p ). key.) - pseudorandom permutation, which is defined as ={0,1} log 2 l) * key->{0,1} log 2 l). ver- a version number bound with the index for individual blocks. initially the version is zero for all data blocks. s ver ij -the seed for pseudorandom function which depends on the file name, block index i,server position j as well as the optional block version number is ver. III. PROPOSED SYSTEM Securing Cloud Storage Data In cloud storage user's store their data and don't maintain that data locally. Whatever be the data stored in the cloud server must be secure. The important issue is to effectively detect any unauthorized data modification and corruption. This is occur due to the server collaborate to dispute and/or random byzantine failures. To address these Page 1070

3 kind of problems, we are using the scheme foe securing the data in the cloud server. In first step we distribute for file to cloud servers by dividing into blocks. After that the homomorphic token is introduced. This belongs to the family of hash functions[17]. Challenge-response authentication is a protocol in that we have two parities. The first party send the challenge to the second party. The second party responds to that and send valid data to that challenge. In our scheme, the challenge-response protocol is used to verify correctness of stored data as well identifying the misbehaving servers. The procedure for error recovery and file retrieval is also discussed here. Finally, we describe how to extend it to third party auditor. 3.1 Distributing The File In the distributed systems erasure correcting code is used to recover the server colluding attacks and byzantine failures. In the cloud storage systems we store our file. Before going to store we first distribute the file and redundancy store across the n=m+k distributed servers. In this we are using m,k) Reed-solomon erasure- correcting code for creation of k parity vectors from data vectors. So that the original file can be constructed from any one of m+k vectors. By distributing m+k vectors on different servers if any of the failures occur, we can recover our data without losing. For efficient sequential I/O our layout is maintained as systematic. Let us assume that our file is F=F 1,F 2...F m ) and F i =f 1i,f 2i,...f li ) T i {1,...,m}).Here F=file we are distributing, T is the transpose denotes that each F i is represented as a column vector, and l denotes that data vector represented in blocks. All blocks are represented in GF2 p ).coming to our schema the elements blocks are represented in GF.It is represented as dispersal matrix A is derived from mxm+k) vandermonde matrix where jj {1,...n}) are distinct elements randomly picked from Galois field2 p ).After the sequence of elementary row transformations the desired matrix A can be written as Where I is a mxm identity matrix and p is the secret parity generation matrix with size mxk.note that A is derived from a vandermonde matrix.the user encoded file by multiplying F By A,that is denoted by G. G=F.A=G 1),G 2),...,G m),g m+1),...,g n) ) =F 1,F 2,...,F m,g m+1),...,g n) ), where Gj)=g 1 j),g 2 j),...g l j) ) T j {1,...,n}) where G m+1),...,g n) ) are parity vectors generated based on F. 3.2 Token Precomputation And Challenge For Token To acheive the correctness of the stored data and error localization.we depend on the technique that is verification of precomputed token:it is described as follows:if the user want to storage file in the cloud server. He distribute his file on the different server.before distributin his file on different server he first compute the short verifiction token for verification on each G j j {1,n}) vector. These token cover all l blocks.after precompution of tokens user store their data in different servers. If the user intersted to retrive a block of he want to make check the storage correctness. He challenges the server based on the challenging request. The server computes short signatures over the specified blocks and returns to the user.the signature value should match with the user precomputed tokens. The user precompute the tokens by using PRF f.),a PRP.), a challenge key k chal, and a master permutation key k PRP. The user do the following steps: 1.determine the random challenge value α i of the GF 2p) by α i =f kchal i) and permutation key K PRP. 2.compute set of r randomly-chosen indices based on { 1,, 1 }, where = q). 3.calculate token as: = *, where =. A=I/P)= Page 1071

4 After the generation of the token computation user keep it locally or store in the server in encrypted form.in this user stores locally. Algorithm 1:Token Precomputation 1: Procedure 2: Choose parameters, and function, 3: Choose the number of token; 4: Choose the number of indicates per verification; 5: Generate master key and challenge key ; 6: for vector, 1, do 7: for round 1, do 8: Derive = and from. 9: Compute = 10: end for 11: end for 12: Store all for s locally. 13: end procedure. Once the token computation is finished the final step is blind to the each parity vector in G m+1),...,g n) ). It is represented as by + s ij ), i {1,...,l},After the completion of the blinding process the user distribute the file to diffferent servers.s1,s2,...sn. 3.3 Verification Of Storage Correctness And Localization Of Errors Error localization is the used to identify the erros at server side. The challenge-response protocol is used to give challenge server and response from server,not only determine this but also identify the potential threats from servers.the procedure for the ith challenge-response protocol over n servers is described as follows: 1.The user computes the tokens and sends a challenge to each server α i as well as the ith permutation key to each servers. 2.The server storing vector G j) j {1,...n}) aggregates those rows specified by index into linear groups send back responses = * G j) [ q)],and where j {1,...,n}). 3.user receiving ) from servers and from that substract blind values from that. -, ).,where I q = q). 4.Then,the user verifies whether the received values remain a valid codeword present in secret matrix p.,..., ).P?,..., ). After the verification matching the valid code words,the user thinks that file exist in server without corruption.if the challenge is fail,then for that each response ) is computed exactly in the same way as token.thus we users can simply find which server is misbehaving by verifying the following n equations.? =, j {1,...,n}. Algorithm 2: Correctness verification and error localization. 1: procedure CHALLENGE I) 2: Re compute = and from ; 3: Send, 4: Receive from servers: to all the cloud serves; = 1 5: for + 1, ) do 6:, = 7: end for 8:if than,, ). ==,, 9: Accept and ready for next challenge. 10: else 11: for 1, ) do 12: If )! = than 13: return server is misbehaving Page 1072

5 14: end if 15: end for 16: end if 17: end procedure. 3.4 Retriving And Recovery Of File: The user can Reconstruct the orginal data by downloading the file.since our layout is systematic.byusing the parametersr,l,t) appropraitely we can guarantee the retrieve of file.when the corruption of the file occurs compare precomputed tokens and received response.the correction of blocks is done by erasure correction. Algorithm 3:Error Recovery 1: procedure %Assume the corruptions of blocks have been detected among % the specified rows; %Assume s k servers have been identified misbehaving 2: Download r rows of blocks from servers; 3:Treat s servers as erasures and recover the blocks. 4:Resend the recovered blocks to corresponding servers. 5: end procedure. 3.5 Toward Third party Auditing If the user does not have time to verify his data in the cloud service provider.he can optionally give his work to third party Auditor. By introducing the third party auditor the scheme does not bring any vulnerabilities.our scheme is similiar to third party Auditor with slight modification.the only difference is blinding parity vectors before distribute the file to each server. 1.Before distributing the file,user blinds each file block G 1),G 2),...,G m) ) by +s ij ), i {1,...,l}, where k j is the secret key for data vector Gj) j {1,...,m}). 2.Based on the blinded data vectors G 1),...,G m) ), the user generates k parity vectors G m+1),...g n) ) by using the secret matrix P. 3.The user calculates the ith token for server j as pervious scheme: ) = *G j) [I q ], where G j) [I q ]= and = i) GF2 p ). 4.The user sends the token set { { } }{1 i t, 1 j n },secret matrix P, permutation and challenge key k PRP, and K chal to TPA for auditing delegation. The correctness validation and misbehaving server identification for TPA is just similiar the previous scheme.the only difference s TPA does not take the blinding values in from server response R j) j {1,...,n}).TPA does not take away blinding values from R j) 's.because it can not perform following equation. ) -.,where I, q = q).so,tpa didnot get the data during the auditing process.the TPA doesnt collude the user information. IV. DYNAMIC OPERATION SUPPORT TO CLOUD STORAGE DATA The static file is suitable when we are storing libraries and scientific data sets. But for various world applications like electronic documents, photos,log files etc. The data is dynamic. The User is always interested to do various operations like update, delete, append. So our proposed scheme should support these operations. By using the secret matrix user first generate parities for data blocks to perform anyone of the dynamic operations. 4.1 Update operation In the cloud storage, user always interested to perform update operation on his data. Suppose if he wishes to change a block to new value, the old block is termed as f ij and the new value is named with f ij+ f ij. We say this operation as update. The user can construct update matrix as: = =,,, ) In the we specify zero elements to denote unchanged blocks. To get the updated information for data vectors and parity vectors. We must multiply.. = 1),..., G m), G m+1),... G n) ) = F 1,... F m, G m+1),., G n) ), where j) j { m+1,...,n}) denotes the updated information for parity vector G j). After updating of the file blocks. The next calculation of token verification. To maintain the same storage Page 1073

6 correctness in G j),unused token is every time exclude of old file and replace it with new block 4.2 Delete operation The delete operation is same as the update operation. This operation is done if the user wants to delete some files blocks stored in the cloud server. In this operation the deleted file blocks is represented with zero or some special predetermined symbols, where as in the update operations the unmodified blocks are represented with zero or some special symbols. After that we also modify the affected token values and corresponding to that blind the affected parity vectors which is similar to the update operation. we represent - f ij in F. 4.3 Append operation If the user wants to add some more blocks where the data stored in the cloud server. By appending operation the user add more blocks at a time not a single block. Appending the blocks is done by adding the blocks at the end of the file. If we have only l row in the matrix F. Then to we add m blocks, denoted as f l+1,1,f l+1,2,...,f l+1,m ).with the secret matrix p, the user can directly append the blocks for each parity server as f l+1,1,f l+1,2,...,f l+1,m ).P =,..., ) ). 4.4 Insert Operation The insert operation is similar to Append operation. Insert the file block at desired position by maintaining data block structure. By inserting the new file blocks, previous all blocks are shifted starting to index j+1.so insertion is difficult. So to maintain the block index we need additional structures. The operation is doing from logical append operation. V.EXPERIMENTS RESULTS AND FUTURE WORK We focused our scheme in terms of storage correctness, security, and efficiency. Our security analysis mainly focuses on on adversary model defined. Beyond this we also evaluate the efficiency by implementing both file distribution preparation and verification token pre computation. 5.1 Performance Evaluation We focus on the cost of file distribution preparation as well as token generation. preparing 1 GB file before outsourcing. In fig 2.a), we set the data vectors m constant at 10, while of decreasing the number of parity vectors k from 10 to 2. In fig. 2 b) total m+k parity vectors fixed at 22 and change the number of data vectors from 18 to 10. From the fig - 2 we can say that number k is the dominant factor for cost of both parity generation and parity blinding. The parity blinding cost is expected to further improve. 5.2 Future Enhancement The nodes which are connected to the server will be connect to the security layer. If the user wants to send his file to private cloud,he has to select the security algorithm based on privacy level of data. The security server will protect document and save in db. If any user wish to select his data, he is connected to same security server. VI. CONCLUSION: we investigate the problem of data security in cloud data storage, which is essentially a distributed storage system. To achieve the assurances of cloud data integrity and availability and enforce the quality of dependable cloud storage service for users, we propose an effective and flexible distributed scheme with explicit dynamic data support, including block update, delete, and append. By improving performance of Byzantine failure, malicious data modification attack, and even server colluding attacks VI1.REFERENCES [1] A. Juels and B.S. Kaliski Jr., PORs: Proofs of Retrievability for Large Files, Proc. 14th ACM Conf. Computer and Comm. Security CCS 07), pp , Oct [2] G. Ateniese, R. Burns, R. Curtmola, J. Herring, L. Kissner, Z. Peterson, and D. Song, Provable Data Possession at Untrusted Stores, Proc. 14th ACM Conf. Computer and Comm. Security CCS 07), pp , Oct [3] M.A. Shah, M. Baker, J.C. Mogul, and R. Swaminathan, Auditing to Keep Online Storage Services Honest, Proc. 11th USENIX Workshop Hot Topics in Operating Systems HotOS 07), pp. 1-6, [4] M.A. Shah, R. Swaminathan, and M. Baker, Privacy-Preserving Audit and Extraction of Digital Contents, Cryptology eprint Archive, Report 2008/186, Page 1074

7 ISSN Print) [5] G. Ateniese, R.D. Pietro, L.V. Mancini, and G. Tsudik, Scalable and Efficient Provable Data Possession, Proc. Fourth Int l Conf. Security and Privacy in Comm. Netowrks SecureComm 08), pp. 1-10, [6] Q. Wang, C. Wang, J. Li, K. Ren, and W. Lou, Enabling Public Verifiability and Data Dynamics for Storage Security in Cloud Computing, Proc. 14th European Conf. Research in Computer Security ESORICS 09), pp , [7] C. Erway, A. Kupcu, C. Papamanthou, and R. Tamassia, Dynamic Provable Data Possession, Proc. 16th ACM Conf. Computer and Comm. Security CCS 09), pp , [8] H. Shacham and B. Waters, Compact Proofs of Retrievability, Proc. 14th Int l Conf. Theory and Application of Cryptology and Information Security: Advances in Cryptology Asiacrypt 08), pp , [9] K.D. Bowers, A. Juels, and A. Oprea, Proofs of Retrievability: Theory and Implementation, Proc. ACM Workshop Cloud Computing Security CCSW 09), pp , [10] R. Curtmola, O. Khan, R. Burns, and G. Ateniese, MR-PDP: Multiple-Replica Provable Data Possession, Proc. IEEE 28th Int l Conf. Distributed Computing Systems ICDCS 08), pp , Mr B.Anjani Kumar is a Student of Kakinada Institute of Engineering & Technology. Currently he is persuing his M.TechCS) from this college. He received his Post Graduation M.ScCS) from Andhra university. Mr. K. Hari Prasad received MCA from Andhra University and M.Tech CSE) from JNTU Kakinada is working as Assistant Professor in P.G. Department of Computer Science, Ideal College of Arts & Sciences, Vidyut Nagar Kakinada. He has 6 years of teaching experience. He has participated in three National Seminars and work shop conducted by IBM and ORACLE, Published Two National & International Journal in the area of Wireless Sensor Networks, Optimization problems and Dataa Mining. His areas of interest are DBMS, Networking, Artificial Intelligence and Cloud Computing Mr. Subash Chandra Chadalavada, received M.Sc. in Physics with Electronics specialization from Andhra University and M.Tech. CSE) from JNTU Kakinada is working as Associate Professor in Department of Computer Science Engineering, Kakinadaa Institute of Engineering and Technology, Korangi. He is an active member of CSI & ISTE. He has 10 years of teaching experience. There are a couple of publications both national and International Conferences / Journals to his credit. His area of interest includes Information Security, Cloud Computing, Computational Photography and other advances in Computer Applications. Page 1075