Journal of Network and Computer Applications: Sandeep K. Sood
Journal of Network and Computer Applications: Sandeep K. Sood
a r t i c l e i n f o abstract
Article history: Cloud computing is a forthcoming revolution in information technology (IT) industry because of its
Received 16 January 2012 performance, accessibility, low cost and many other luxuries. It is an approach to maximize the
Received in revised form capacity or step up capabilities vigorously without investing in new infrastructure, nurturing new
25 May 2012
personnel or licensing new software. It provides gigantic storage for data and faster computing to
Accepted 3 July 2012
Available online 25 July 2012
customers over the internet. It essentially shifts the database and application software to the large data
centers, i.e., cloud, where management of data and services may not be completely trustworthy. That is
Keywords: why companies are reluctant to deploy their business in the cloud even cloud computing offers a wide
Cloud security range of luxuries. Security of data in cloud is one of the major issues which acts as an obstacle in the
Encryption
implementation of cloud computing. In this paper, a frame work comprising of different techniques and
Message authentication code
specialized procedures is proposed that can efficiently protect the data from the beginning to the end,
Virtualization
Secured socket layer i.e., from the owner to the cloud and then to the user. We commence with the classification of data on
the basis of three cryptographic parameters presented by the user, i.e., Confidentiality (C), Availability
(A) and Integrity (I).The strategy followed to protect the data utilizes various measures such as the SSL
(Secure Socket Layer) 128-bit encryption and can also be raised to 256-bit encryption if needed, MAC
(Message Authentication Code) is used for integrity check of data, searchable encryption and division of
data into three sections in cloud for storage. The division of data into three sections renders
supplementary protection and simple access to the data. The user who wishes to access the data is
required to provide the owner login identity and password, before admittance is given to the encrypted
data in Section 1, Section 2, and Section 3.
& 2012 Elsevier Ltd. All rights reserved.
1084-8045/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jnca.2012.07.007
1832 S.K. Sood / Journal of Network and Computer Applications 35 (2012) 1831–1838
2006). The various security concerns and upcoming challenges are Yahoo! to business applications. They also provided the concept of
addressed in (Daniel and Wilson, 2003; Dikaiakos et al., 2009) and ‘‘On demand’’ and ‘‘SaaS’’ with their real business and successful
also reviewed in terms of standards such as PCI-DSS, ITIL, and ISO- customers. IBM extended these concepts in 2001, as detailed in the
27001/27002. There are also architectural security issues which are Autonomic Computing Manifesto, which describes advanced auto-
changing according to various architectural design functioning over mation techniques such as self-monitoring, self-healing, self-config-
cloud computing. Since outsourcing is the main theme of cloud uring and self-optimizing in the management of complex IT systems
computing, there are two main concerns in this area: with heterogeneous storage, servers, applications, networks, security
mechanisms and other system elements that can be virtualized
1. External attacker (any unauthorized person) can get to the across an enterprise. Amazon.com played a key role in the devel-
critical data, as the control is not in the hands of the owner. opment of cloud computing by modernizing their data centers. It
2. Cloud service provider himself can breach the owner, as data is found that the new cloud architecture resulted in significant internal
to be kept in his premises. efficiency improvements and providing access to their systems by
way of Amazon Web Services in 2005 on a utility computing basis.
Any kind of security and privacy violation is critical and can 2007 saw increased activity with Google, IBM and a number of
produce dire consequences. As soon as cloud privacy issues are universities embarking on a large scale cloud computing research
further organized and strict regulations and governance for cloud project, around the time the term started gaining popularity in the
operation are in position, more and more business owners will mainstream press. In August 2008, Gartner Research observed that
feel safe to opt for cloud computing. ‘‘organizations are switching from company-owned hardware and
The proposed model has been structured by bringing together software assets to per-use service-based models’’. The projected
various techniques and utilizing them to perform the task of data shift to cloud computing will result in dramatic growth in IT
security in cloud. This combination of diverse methods operate as a products in some areas and in significant reductions in other areas.
wall stood together against the security challenges, which have been Despite all the hope of gaining maximum advantage from this cloud
constantly creating the loop holes in the efficient functioning and computing, it seems to have born with security and management
growth of the cloud. This model is described in such a way that it concerns, which time to time hinders its growth. For this, lot of
provides a complete view of processing the data at different levels. research work has been done to secure the data in cloud computing
The model uses encryption as the main fundamental protection (primary concern) from every perspective, but everything seems to
scheme and data sent to cloud is in encrypted form. Encryption is face a new challenge as soon as it is employed.
the conversion of data into encrypted form called a cipher text that Juels et al. (2007) described a formal Proof of Retrievability (POR)
cannot be easily understood by unauthorized person and can be model for ensuring the remote data integrity. Their scheme com-
decrypted by the authorized person having a valid decryption key. bines spot-checking and error-correcting code to ensure both
Apart from this, the model positively handles the security issues by possession and recovery of files on archive service systems.
employing strict authentication parameters, digital signatures, stor- Shacham and Waters (2008) built on this model and constructed a
ing encrypted data in cloud according to sensitivity rating, building random linear function based Homomorphic Authenticator. This
of index, using of MAC for integrity check and keyword search for enables unlimited number of queries and requires less communica-
data in cloud. Thus all these parameters result into a defined tion overhead. Bowers et al. (2008a) proposed an improved frame-
mechanism that encourages the proper functioning of cloud com- work for POR protocols that generalizes both Juels and Shacham’s
puting. In this computing model, owner sends the encrypted data to work. Later in their subsequent work, Bowers et al. (2008b)
cloud where it is stored in different sections depending on the extended POR model to distributed systems. However, all these
sensitivity rating and then the data can be retrieved by user from schemes are focusing on static data. The effectiveness of their
the cloud when requested. However, this is achievable only after schemes rests primarily on the pre-processing steps that the user
passing the authentication parameters and then searching the data conducts before outsourcing the data file. Any change to the
by the use of keyword obtained from the owner. contents of data file, even a few bits, must propagate through the
This paper is structured as follow: Section 2 summarizes the error-correcting code, thus, establishing significant computation and
related work for security of data. In Section 3, a model is proposed communication complexity.
which is designed to solve the security issue of cloud computing. Chor et al. (1995) proposed private information retrieval (PIR)
Section 4 provides the security analysis of the designed model. so that clients can access entries in a distributed table without
Section 5 compares functionality of proposed model with other revealing which entries they are interested in. The PIR literature
security models. Section 6 shows the evaluation procedure and usually aims for very strong information theoretic security
Section 7 concludes this paper. bounds, which makes it harder to find practical schemes. PIR
schemes often require multiple non-colluding servers, consume
large amounts of bandwidth, do not guarantee the confidentiality
2. Related work of the data, do not support private keyword searching and do not
support controlled searching or query isolation. The schemes
The cloud is a terminology with a long history in telephony, which (Cachin et al., 1999; Chor et al., 1998; Gertner et al., 1998;
has in the past decade, been adopted as a metaphor for internet based Kusilevitz and Ostrovsky, 1997) are important exceptions which
services, with a common depiction in network diagrams as a cloud allow removing some but not all these limitations.
outline. The underlying concept dates back to 1960 when John Recently, Wang et al. (2009) described a homomorphism
McCarthy opinion that ‘‘Computation may someday be organized as distributed verification scheme using Pseudorandom Data to
a public utility’’; indeed it shares characteristics with service bureaus verify the storage correctness of user data in cloud. This scheme
which date back to the 1960s. The term cloud had already come into achieves the guaranty of data availability, reliability and integrity.
commercial use in the early 1990s to refer to large Asynchronous However, this scheme was also not providing complete protection
Transfer Mode (ATM) networks. By the turn of the 21st century, the to user data in cloud computing, since pseudorandom data would
term ‘‘cloud computing’’ had started to appear, although major focus not cover the entire information.
at this time was on Software as a Service (SaaS). In 1999, sales- Prasad et al. (2011) and Sood et al. (2011) discussed different
force.com was established by Marc Benioff, Parker Harris. They security aspects in computing. Prasad et al. (2011) technique provides
applied many technologies of consumer web sites like Google and a new way to authenticate in 3-dimensional approaches. It provides
S.K. Sood / Journal of Network and Computer Applications 35 (2012) 1831–1838 1833
availability of data by surmounting many existing problem like denial cannot be trusted with data of high sensitivity. Hereby we can say
of services and data leakage etc. Additionally, it also provides more that the proposed model has been designed by keeping all these
flexibility and capability to meet the rising demand of today’s things in mind and surely in comparison to prior works, provides
complex and diverse network. But in this model, the data stored is all these required measures to protect data in a very efficient and
not in encrypted form and once the username and password is lost, organized manner.
the data can easily be retrieved by any unauthorized user.
Kamara and Lauter (2010) worked over public cloud infra-
structure and proposed a model which is well suited for preser- 3. Proposed model
ving integrity with the help of cryptographic primitives. This
technique is purely based on cryptographic storage services. In Proposed framework has been structured to provide complete
proposed procedure, when a user wants to send data to other security to the data throughout the entire process of cloud
user, they first generate a master key that encrypts their message. computing, be it in cloud or in transit. Thus, multiple mechanisms
The secret key for decryption is stored on receivers’ system for and available techniques are applied to shield the critical infor-
decrypting the same message. They use the concept of index mation from unauthorized parties. The proposed frame work is
encryption and tokens are generated with the knowledge of divided into two phases. First phase deals with process of
secret key. The searching method is not very efficient for transmitting and storing data securely into the cloud. Second
encrypted data. They discussed symmetric searchable encryption phase deals with the retrieval of data from cloud and showing the
(SSE) and asymmetric searchable encryption (ASE). These techni- generation of requests for data access, double authentication,
ques are used for encrypted data searching but increase complex- verification of digital signature and integrity, thereby providing
ity and make the system cumbersome. authorized user with data on passing all security mechanisms.
Wang et al. (2010) discussed the drawbacks of using ordinary
encryption techniques and suggested that these techniques are not 3.1. Phase 1(storing of data)
useful over cloud because for this user should have pre knowledge
about the encrypted cloud data. Their model is based on symmetric This phase deals with mechanisms and methods to store and
searchable encryption method. They gave design for existing crypto- secure the data from beginning and transmitting it securely to the
graphic primitive and order preserving symmetric encryption cloud in encrypted form. It is further divided into sub-sections
(OPSE). Security analysis shows its success rate for one to many (Classification, Index Building and encryption, Message Authenti-
mapping and for ranked keyword search. This model did not provide cation Code (MAC) which provide stepwise details of action on
any information about the security attacks, confidentiality and the data).
integrity. This model is not well suited for preserving security.
Popa et al. (2010) presents Cloud Proof, a secure storage 3.1.1. Classification
system for increasing security over cloud. In this model users As the data in the cloud is intended to be stored, an approach is
can detect violations of integrity, confidentiality, write serial introduced for storing the data in different sections in the cloud
ability and freshness. Model use cryptographic tools and engi- (public, private, limited access) basis of three cryptographic para-
neering efforts to obtain an efficient and scalable system which meters viz: Confidentiality, Availability and Integrity. These values
allow users to detect and prove cloud misbehavior. will be listed by the client himself and sensitivity rating (SR) will be
Cloud computing is a layered technology and the data in cloud calculated using the proposed algorithm shown ahead. The value of
computing has to go through different processing levels, so the C (confidentiality) is based on the level of privacy needed at each
security mechanism should be efficient and provided at each step, step of data processing, value of I (integrity) is based on how much
i.e., from owner to cloud and cloud to user or back to owner. Data accuracy of data, reliability of information and protection from
should not succumb to the attackers trying to retrieve or tamper unauthorized modification is required, and value of A (availability)
with it and not even the cloud provider should be able to harm is based on how frequently data is accessed and should available
the data in any possible manner, because cloud service provider immediately when requested Fig. 1.
Algorithm.
In the algorithm listed above, the primary job of the owner is build up an index is that, for each word W (keyword) of interest,
to categorize the data on the basis of cryptographic parameters list the documents that contains W. Building up an Index provides
viz: C, I and A. Here D [ ] represents the data and the user has to faster retrieval of files. To provide more security against revealing
give values of C, I and A. After applying the proposed formula as any sort of information to cloud we will encrypt the index also.
shown above, the value of Sensitivity Rating (SR) is calculated. This index will basically contain a list of keywords, with each
This ‘‘SR’’ value is used to allocate the data to one of the three keyword contains list of pointers to the documents where key-
sections in cloud, i.e., S3 [Public], S2 [Private] or S1[Owner’s word appears. The keywords are words of interest that a user may
Limited Access] as shown in Fig. 2. want to search later. Best practice is to build an index of clear
documents and then encrypt both document and index and store
the encrypted data onto the cloud. The index should be encrypted,
3.1.2. Index building and encryption by encrypting keywords as well as document pointers in each list
After the successful allotment of values to data, the data now in the index. After this we need to encrypt data. Now, to code the
needs to gear up for another processing mechanism. As the data data, the model uses encryption. Encryption is the process of
on cloud will be stored in encrypted form and searching over turning intelligible information into useless information. With
encrypted data is a complicated issue, so we need to build up an Secure Socket Layer (SSL) encryption, there is also a key that
index, using Index builder shown in Fig. 3, so that while retrieval, allows only an authorized person to be able to decode the
we can perform searching over encrypted data. Possible way to information. This model uses 128-bit SSL encryption to encrypt
the data as well as index as shown below.
Now when the data has been stored in cloud in secure manner, 2. Answer to security question
the retrieval of data should be supported with equally best OWNER 4. Request for data USER
CLOUD 3. Send user identity
possible mechanism and techniques. First the retrieval of data and Digital Signature
5. sends Digital Signature & Keyword
requires the user to register him with the owner/organization by
getting a username and a password as shown in Fig. 5. The user
Fig. 7. Authentication process.
will register to get its username and password at organization,
which will further forward the username to cloud to let it store
the username into its directory.
In this model, when the user requires accessing the data in Keyword + File
Digital signature
cloud, he sends a request along with the username to cloud. Cloud
check the request and if it is for public section (Section 3), then DECRYPTION
without authentication access is granted and user after retrieving
can decrypt this data by the public key provided in the section
only. If the request is for private section (Section 2) and limited Search Request ENC. File
access section (Section 1), authentication is necessary and cloud
looks for username provided by user into its directory of user-
names, provided by the owner. Fig. 6, given below, depicts the Download request
transmission of these requests. The sectional database used in this Search request
Search result Encrypted file
model will provide access under the following guidelines:
upper section. e.g.: if user has got access for data in limited
access section, then the data of same owner available in Fig. 7 shows the authentication and request of user to owner
private section will also be available to the user. for data.
For authentication, the user first sends the password to the
owner and on clearing this parameter, user is prompted with a
As seen in Fig. 7, if the username matches, cloud forwards the
security question from owner and after answering it correctly,
username to the owner/organization for authentication, here the
user is authenticated. The owner also sends the user identity
important thing is that primary authentication process is kept
along with the digital signature to cloud so that cloud will be sure
with the owner only because criticality of the process is such that
that owner has given the access of data to particular user identity
even the cloud cannot be trusted. Now as the company receives
for this session only. Afterwards user sends the request for data to
the username from cloud, it has to authenticate the username.
owner which in turn sends the owner’s ‘‘Digital Signature’’,
keyword of requested data and a master key to decrypt the data
User Client [Enrolled in Directory]
provided by cloud. On receiving the digital signature and keyword
from the owner, the user forwards the same to the cloud with a
search request for data corresponding to the keyword as shown in
Company/ Owner
Fig. 8. The cloud first verifies the digital signature, and if verified,
1. Register to get username with password for authentication
cloud processes the search request using the keyword. Basically
2. Send username to cloud for storing in directory searching over encrypted data provides easy retrieval of file and
without revealing any critical information to the cloud. As
Fig. 5. Registration process.
explained earlier we already have stored an encrypted index
containing a list of keywords and with each keyword list of
pointers to the document where keyword appears. Whenever the
1. Request to access data & sends username cloud gets a keyword to search over encrypted data, it finds a
CLOUD USER
match and then returns the user encrypted list of matching
2. Looks into directory positions from the index. The user can then decrypt the encrypted
entries by the decrypting key provided by owners of data and
CLOUD DIRECTORY
sends cloud download request to retrieve relevant file or docu-
ment which user was looking for. The cloud replies the user with
3. Authentication message to user
encrypted file it requested for and then user can decrypt the file
by decryption key of file F, already shared by owner with user.
If
username One possible advantage for this scheme is that the request could
matches
be embedded in other retrievals so that cloud might have
uncertainty about the correlation of the search request and the
retrieval request for cipher text. Fig. 8 shows the user generating
4. Redirects
the search request and retrieving encrypted file from the cloud.
COMPANY/ OWNER Now, as the user has received his data from cloud, the question
or doubt arises in the mind regarding its integrity. As this model
Fig. 6. Access request. uses MAC for integrity check, the user can assure him by deriving
1836 S.K. Sood / Journal of Network and Computer Applications 35 (2012) 1831–1838
D
proposed model uses double encryption, one being done by
ENC. File +
File
F
owner and other using SSL. The processing power needed, among
MAC (F”) C
K1 other things, would render most attackers ineffective. Hence this
COMPARE MAC
K1 approach not only safeguards data where it lives, but also helps
assure customers that data is secure while in transit.
Fig. 9. MAC comparison.
4.3. Threat from cloud service provider
the MAC of encrypted file received using secret key, already The cloud is the place where the data resides after being
shared with user and comparing it with MAC received along with transmitted by the owner. Suppose the data in cloud is safe from
the encrypted file. Here the verification process of data integrity any third party, as the cloud service provider will use strict
can be performed as shown in Fig. 9. measures to protect it. The cloud service provider can turn against
Fig. 9 demonstrates that when the encrypted file is retrieved the owner. As the data is not in the control of owner when in
by user, Key (k1) is used to derive its MAC^ and then compared cloud, anything can be possible or cloud service provider can
with the MAC received along with the encrypted data. If manage any leakage of data even by helping the rival parties. So,
MAC^¼MAC, i.e., MAC of encrypted data which was computed the cloud service provider (CSP) cannot be trusted blindly. For this
by owner and MAC^ of encrypted data computed by user is equal the best possible solution used in proposed model is encryption of
then it proves that data has not been tampered throughout the data stored in cloud. SSL Certificates as used in the proposed
whole data traversing. On the other hand, the user can use the key model encrypts private communications over the public Internet.
(k1) to decrypt the encrypted file received. With this step the Using public key infrastructure, SSL consists of a public key
process of retrieval concludes and we have seen that throughout (which encrypts information) and a private key (which decrypts
the whole working, the model has taken all the required precau- information), so that only the key owners can read it. 128-bit SSL
tions and measures to protect the data from possible attacks such encryption encrypts the data in such a way that it is nearly
as data leakage, unauthorized access and tampering of data etc. impossible for an attacker to decrypt it by a brute force attack.
Table 1
Functionality comparison.
Juels et al. (2007) Chor et al. (1995) Wang et al. (2009) Prasad et al. (2011) Proposed model
7. Conclusion
Julisch K, Hall M. Security and control in the cloud. Information Security Journal: A Popa RA, lorch JR, Molnar D, Wang HJ, Zhuang L , Enabling security in cloud storage
Global Perspective 2010;19(6):299–309. SLAs with cloudproof. Technical report. Microsoft Research May 2010.
Kamara S, Lauter K. Cryptographic cloud storage. Lecture Notes in Computer Shacham H, Waters B. Compact Proofs of Retrievability, Proceedings of Asiacrypt
Science 2010;6054:136–49. ’08, 5350, p. 90–107, 2008.
Kusilevitz E, Ostrovsky R. Replication is not needed: single database, computa- Sood SK, Sarje AK, Singh K. A secure dynamic identity based authentication
tionally-private information retrieval, In Proceedings of the 38th annual protocol for multi-server architecture. Journal of Network and Computer
symposium on foundations of computer science, IEEE, p. 364–373, 1997. Applications 2011;34(2):609–18.
Overby E, Bharadwaj A, Sambamurthy V. Enterprise agility and the enabling role of Wang C, Cao N, Li J, Ren K, Lou W. Secure ranked keyword search over encrypted
information technology. European Journal of Information Systems 2006;15(3): cloud data. Journal of the ACM 2010;43(3):431–73.
120–31. Wang C, Wang Q, Ren K, Lou W. Ensuring data storage security in cloud computing,
Prasad P, Ojha B, Shahi RR, Lal R. 3-dimensional security in cloud computing. quality of service, 2009, IWQoS IEEE 17th international workshop, p. 1–9, 2009.
Computer Research and Development (ICCRD) 2011;3:198–208.