CLOUD COMPUTING SECURITY IN UNRELIABLE CLOUDS USING RELIABLE RE-ENCRYPTION Chandrala DN 1, Kulkarni Varsha 2 1 Chandrala DN, M.tech IV sem,department of CS&E, SVCE, Bangalore 2 Kulkarni Varsha, Asst. Prof. Department of CS&E, SVCE, Bangalore ABSTRACT In this paper, we propose a time-based re-encryption scheme using attribute-based encryption (ABE). This scheme is best suited for efficient data retrieval from the cloud, which enables the cloud to automatically re-encrypt the data based on the internal clock. It prevents the revoked users from decrypting the data using their old decryption keys. The naive solution is built on Attribute Based Encryption, an efficient data retrieval scheme that allows fine-grain access control based on internal clock of the server. Keywords-Attribute-based encryption, cloud computing, proxy re-encryption 1. INTRODUCTION Cloud is a distributed system where there are many cloud servers. In cloud environment the data owner s data is stored on multiple cloud servers. The cloud computing usage is rapidly growing because of the cost savings from outsourcing data to cloud service provider (CSP). A technique to protect the data from untrusted CSP is, the data owner to encrypt the outsourced data [1][2]. A Flexible encryption scheme like attribute based encryption (ABE) [3][4] is used, that provides fine-grained access control. ABE is first introduced by Sahai and Waters, provides a mechanism that ensures even when the storage is compromised, the loss of information will be minimal. ABE binds the access control policy to the data and users instead of having a server mediating access to files. In ABE, it encrypts the data using an access structure with different attributes. The user issues their attribute keys instead of decryption keys for specific files. Attributes of the user should satisfy the access structure to decrypt a particular file. For example, a file is encrypted using access structure {(a b) c } it means that either a user with attributes a and b, or a user with attribute c, can decrypt the file.in a secure cloud computing, the data will be stored in the encrypted form and issue the decryption keys to the authorized user. The problem in cloud lies in revoking access rights from users, whose permission is taken back, will still retain the earlier keys and they can still decrypts data in the cloud. A solution to this is to let the data owner to re-encrypt the data immediately, so that the revoked users cannot decrypt the data by using their old decryption keys. While distributing new keys to the remaining authorized users, it may lead to performance bottleneck where there is frequent users revocation. An alternative solution is to apply the proxy re-encryption (PRE) technique. PRE takes the advantage of the abundant resources from cloud by delegating the cloud to re-encrypt data [8], this approach is also known as command-driven re-encryption scheme, where cloud servers perform re-encryption while receiving commands from the data owner. Fig.1: A Typical Cloud Environment In Fig. 1, which should be propagated to CS1, CS2, and CS3? Due to a network outage, CS2 did not receive the command, and did not re-encrypt the data. At this time, if revoked users query CS2, they can obtain the old cipher text, and can decrypt it using their old keys.
A better solution is to allow each cloud server to independently re-encrypt data without receiving any command from the data owner. In this paper, we propose a secure re-encryption scheme in untrusted cloud. It is a time based re-encryption scheme, it enables cloud server to automatically re-encrypt the data based on server internal clock. The secure re-encryption in untrusted cloud combines the data with an access control and an access time. Each user is issued with a key that are associated with an attribute and attribute effective time. The data can be decrypted by the users using the keys with attribute effective time. The data owner and the CSP share a secret key, in which each cloud server can re-encrypt data by updating the data access time according to the cloud environment. A cloud is a distributed system, where a data owner s data is replicated on multiple servers for fast availability. The cloud as a distributed system experiences failures commonly, such as server crashes and network outages. So the re-encryption commands sent by the data owner may not propagate to all the cloud servers in a timely fashion. The proposed scheme is reliable re-encryption in unreliable clouds (R3 scheme for short) it is a timebased re-encryption scheme that enables the cloud server to automatically re-encrypt the data based on its internal clock. In proposed scheme the data is associated with access structure and access time. The keys generated and issued to users are associated with attributes and attribute effective time. The data is decrypted by the users using the keys that are associated with attributes satisfying the access structure, and attrib ute effective time. The main contributions of this paper are as follows: We propose an automatic, time-based, proxy re-encryption scheme suitable for cloud environment with unpredictable server crash and network outages. We extend an ABE scheme by incorporating timestamps to perform proxy re-encryption. Our solution does not require perfect clock synchronization among all the cloud servers to maintain correctness. 2. RELATED WORK Many researchers have proposed storing encrypted data in the cloud to define against the CSP [1][2]. Under this approach, users are revoked by having a third party to re-encrypt data such that previous keys can no longer decrypt any data [8]. The solution by [8] for instance, lets the data owner issues a re-encryption key to an untrusted server to re-encrypt the data. Their solution utilizes PRE [8], which allows the server to re-encrypt the stored cipher text that can only be decrypted using a different key. During the process, the server does not learn the content of the cipher text or the decryption keys. A Hierarchical Attribute-Based Encryption (HABE) model is combining a HIBE system and a CP- ABE system, to provide fine-grained access control and full delegation. Our scheme relies on time to re-encrypt data. However, in a cloud, the internal clock of each cloud server may differ. There have been several solutions to this problem. For instance, proposed a probabilistic synchronization scheme, for reading remote clocks in networks subject to unbounded random message delays. The method can be used to improve the precision of both internal and external synchronization algorithms. Message delay used to synchronize the clocks of their host processors, time server processors communicate among themselves by sending messages via communication networks. 3. PRELIMINARY A) System model The system consists of following subsystems. Data owner: Data owner will upload the files to the cloud and can update the contents of the file. It uses ABE to generate the keys for encrypting the file and stores in the cloud. Data User: The Data user requests for the particular file from cloud, the ABE generates keys for decryption and the user downloads the file. Cloud Storage Server: The encrypted data and keys will be stored in the cloud server, based o n server time it re-encrypts the file and stores in cloud.
B) Design model The main design goal is to protect the data in the cloud and providing security to data. The security requirements of R3 scheme are: 1. Access control correctness: A data user with valid keys can only decrypt the file. 2. Data consistency: The data user requests for a particular file F, should get the same contents in same time slice. 3. Data confidentiality: The contents of the file should be known only to the user with valid keys. 4. Efficiency: The cloud server should not re-encrypt the file without any data user request. C) Adversary model There are two types of adversaries in this system: CSP and malicious user. The CSP adversary is honest but it is curious, it executes the protocol correctly but it tries to get additional information about the stored data. It tries to learn the contents of the file that he is unauthorized to access. The adversary possesses invalid keys with incorrect attributes or time slice. The CSP and malicious adversaries may exist together. 3.1 Basic R3 Scheme In basic R3 scheme, the ideal condition is considered when the data owner and all the cloud servers share a synchronized clock and there should not be any transmission and queuing delays while executing read and write commands. A. Intuition At first, the data owner generates a shared secret key to the CSP, the data owner encrypts file with appropriate attribute structure and time slice and the data owner will uploads the file to t he cloud. The CSP will send copy of the file to many cloud servers which stores an encrypted file F with A and TSi. When a user queries the cloud server, the cloud server first uses its own clock to determine the current time slice and Assumes that the current slices is TSi+K, then the cloud server will automatically re-encrypted F with TSi+K without receiving any command from the data owner. During the process, the cloud server cannot learn or gain the contents of the cipertext and the decryption keys. Only users with keys satisfying A and TSi+K can decrypt the file F. B. Protocol Description The proposed R3 scheme relies on the following functions. 1)Setup ->(Public Key, Master Key, s) : At TS0, the data owner publishes the system public key and the system master key remains as secret. And send the shared secret keys to the cloud. 2)KeyGerenate(Public Key, Master key, s,public key of user, Attribute, Time) -> (Secrete Key of user, {Secrete Key period of user, Attribute}): If data owner wants to grant data user attributes with valid time period, the data owner generates Public key of user and {Secrete Key period of user, Attribute} using the system public key, the system master key, the shared secret key, user s public key, user s attributes and eligible time. 3)Encryption(public key, AccessStructure, s, TSt, File)-> (cipher text) :At time slice Tst, the data owner encrypts file with access structure, and produces cipher text using the system public key, time slice, and plain text file. 4)Decryption(Public Key, Cipher text, Secret key of user {secret key time of user, aij} 1<=j<=n ) -> File : At particular time slice TSt and the user U, who possesses version t attribute secret keys on all attributes in CCi, recovers the file by using the system public key, the user identity secret key, and the user attribute secret keys. 5)Re-Encryption(cipher text, s, TSt+k)-> Ct+k A : When the cloud server wants to return a data user with the file at TSt+k, it updates the ciphertext from CtA to Ct+KA using the shared secret key. The basic R3 description is divided into three stages: data owner initialization, data user read data and the data owner write data. 1) Data owner initialization: The data owner is an interface for the file uploaders. First it runs the Setup function to initiate the system. If the data owner wants to upload a file F to the cloud means, at first it defines an access
control A for file F, and determines the current time slice TSi. At last it runs the Encryption function with A and TSi to output the ciphertext. When the data owner wants to grant a set of attributes in a period of time to data user, it runs the KeyGenerate function with attributes and effective time to generate keys to the user. 2) Data user read data: Data user is an interface for the file downloader, If the data user wants to access file F at time period of TSi, then the user sends a read command R(F) to the cloud server. After receiving the read command R(F),the cloud server runs the Re-Encryption function to re-encrypt the file with time period TSi. On receiving ciphertext, the user runs the Decryption function using keys that satisfy A and TSi to recover F. 3) Data owner write data: When the data owner wants to write file F at TSi, then it will send a write command to the cloud server as W(F, seqnum), where seqnum is the order of the write command, it is necessary for ordering when data owner issues multiple write commands that has to take place in one time slice. After receiving the write command, the cloud server will commit it at the end of time period TSi. 4.SECURITY ANALYSIS The Encrypt algorithm in the Time scheme is the same as the Encryption algorithm in HABE that has been proven to be semantically secure. Therefore, we consider that the time scheme is secure if the following propositions hold: Proposition1. The keys produced by the KeyGenerate algorithm are secure. Proposition2. The cipher text produced by the Re-Encrypt algorithm is semantically secure. Proposition3. Given the root secret key and the original cipher text, the CSP can know neither the underlying data, while executing re-encryption. Algorithm: Extended R3 (Asynchronized clock with delays) While Receive write commands W (F, ti+1, seqnum) do If Current time is earlier than ti+1 + α then Build Window i for file F Commit write command in Window i at ti+1 + α Else Reject the write command Inform the data owner to send write command earlier While Receive a read request R (F, T Si) do If Current time is later than ti+1 + α then Re-encrypt the file in Window i with T Si Else Hold on the read command until ti+1 + α Access control correctness: It is clear that the correctness of access control is most vulnerable when a TS changes. Let us take an example of the case where Alice has keys with effective time up to TSi, and Bob has keys with effective time starting from TSi+1.Where assuming that the data owner updates file F to F such that a user querying the file at TSi should obtain F, and a user querying the file at TSi+1 should obtain F. The property of access control correctness fails if Alice is able to read F, or if Bob is able to read F. Data consistency: This property requires users that query within the same TS must receive the same data. Let us assume that both Alice and Bob have valid keys for the appropriate time slices, and we now want to show that so long as both Alice and Bob query within the same time slice, they must obtain the same data. Data confidentiality: In R3 scheme, we only store encrypted data in the cloud. Since this scheme preserves the data confidentiality operations from HABE scheme, and retain the same confidentiality properties, the cloud without knowledge of keys cannot learn any useful information about the stored data. Data efficiency: The cloud server does not re-encrypt a file until a data user requests that file. In the function Re-Encrypt, when k>1, the cloud server can combine the re-encrypt operations until receiving a file access request.
5.EXPERIMENTAL RESULT This section provides brief description about the implementation of proposed system and result. Implementation of proposed system is carried out using Java. Snapshot 1 explains, the key generation for the file which is uploaded into the cloud. It shows the random keys generated for encryption in data owner module.snapshot 2 shows the list of files that are uploaded to the cloud by the Data owner, user can select the particular file and he can download the file. Snapshot 3 shows the decryption of the file at the user end and the snapshot 4 shows the list of the files that are stored in the cloud, which is Amazon S3 cloud. Snapshot 1: Keys generation for Cloud cloud Snapshot 2: List of files in Snapshot 3: Decryption of file Snapshot 4: Files stored in Amazon S3 cloud 6. FUTURE ENHANCEMENT In future, the data owner can issue a valid user a special seed value, where the user can generate keys on his own. The challenge is to prevent the users from generating additional keys beyond their authorization. 7. CONCLUSION In this paper, we proposed the time based scheme to achieve fine grained access control based on the server s internal clock. Our scheme does not rely on the cloud to reliably propagate re-encryption commands to all servers to ensure access control correctness and our solution remains secure without perfect clock synchronization as long as we bound the time difference between the server and the data owner.
REFERENCES [1] S. Kamara and K. Lauter, Cryptographic cloud storage, Financial Cryptography and Data Security, 2010. [2] M. Armbrust, A. Fox, R. Griffith, A. Joseph, R. Katz, A. Konwinski, G. Lee, D. Patterson, A. Rabkin, and I. Stoica, A view of cloud computing, Communications of the ACM, 2010. [3] A. Sahai and B. Waters, Fuzzy identity-based encryption, Advances incryptology EUROCRYPT, 2005. [4] V. Goyal, O. Pandey, A. Sahai, and B. Waters, Attribute-based encryption for fine-grained access control of encrypted data, in Proc. Of ACM,2006. [5] Qin Liu, Chiu C. Tan, Jie Wu, and Guojun Wang Reliable Re-encryption in Unr Unreliable Clouds in IEEE Globecom 2011 proceedings. [6] N. Antonopoulos and L. Gillam, Cloud Computing: Principles, Systems and Applications, Springer Publishing Company, 2010. [7] A. Boldyreva, V. Goyal, and V. Kumar, Identity-based encryption with efficient revocation, in Proc. of ACM CCS, 2008. [8] G. Wang, Q. Liu, and J. Wu, Hierarchical attribute-based encryption for fine-grained access control in cloud storage services, in Proc. Of ACM CCS (Poster), 2010. [9] S. Yu, C. Wang, K. Ren, and W. Lou, Achieving secure, scalable, and fine-grained data access control in cloud computing, in Proc. of IEEE INFOCOM, 2010. [10] F. Cristian, Probabilistic clock synchronization, Distributed Computing, 1989.