Journal of Information Security, 2012, 3, 295-306
http://dx.doi.org/10.4236/jis.2012.34036 Published Online October 2012 (http://www.SciRP.org/journal/jis)
Towards a Comprehensive Security Framework of Cloud
Data Storage Based on Multi-Agent System Architecture
Amir Mohamed Talib, Rodziah Atan, Rusli Abdullah, Masrah Azrifah Azmi Murad
Faculty of Computer Science & IT, University Putra Malaysia UPM, Serdang, Malaysia
Received May 20, 2012; revised June 27, 2012; accepted July 7, 2012
ABSTRACT
The tremendous growth of the cloud computing environments requires new architecture for security services. Cloud
computing is the utilization of many servers/data centers or Cloud Data Storages (CDSs) housed in many different loca-
tions and interconnected by high speed networks. CDS, like any other emerging technology, is experiencing growing
pains. It is immature, it is fragmented and it lacks standardization. Although security issues are delaying its fast adop-
tion, cloud computing is an unstoppable force and we need to provide security mechanisms to ensure its secure adoption.
In this paper a comprehensive security framework based on Multi-Agent System (MAS) architecture for CDS to facili-
tate confidentiality, correctness assurance, av ailability and integ rity of users’ d ata in th e cloud is proposed. Ou r security
framework consists of two main layers as agent layer and CDS layer. Our propose MAS architecture includes main five
types of agents: Cloud Service Provider Agent (CSPA), Cloud Data Confidentiality Agent (CDConA), Cloud Data Cor-
rectness Agent (CDCorA), Cloud Data Av ailability Agen t (CDAA) and Cloud Data Integr ity Agen t ( CDIA). In order to ver -
ify our proposed security framework based on MAS architecture, pilot study is conducted using a questionnaire survey.
Rasch Methodology is used to analyze the pilot data. Item reliability is found to be poor and a few respondents and
items are identified as misfits with distorted measurements. As a result, some problematic questions are revised and
some predictably easy questions are excluded from the questionnaire. A prototype of the system is implemented using
Java. To simulate the agents, oracle database packages and triggers are used to implement agent functions and oracle
jobs are utilized to create agents.
Keywords: Cloud Computing; Multi-Agent System; Cloud Data Storage; Security Framework; Cloud Service Provider
1. Introduction
Computer in its evolution form has been changed multi-
ple times, as learned from its past events. However, the
trend turned from bigger and more expensive, to smaller
and more affordable commodity PCs and servers which
are tired together to construct something called “cloud
computing system”. Moreover, cloud has advantages in
offering more scalable, fault-tolerant services with even
higher performance [1]. Cloud computing can provide
infinite computing resources on demand due to its high
scalability in nature, which eliminates the needs fo r cloud
service providers to plan far ahead on hardware provi-
sioning [2].
Cloud computing integrates and provides different
types of services such as Data-as-a-Service ( DaaS), wh ich
allows cloud users to store their data at remote disks and
access them anytime from any place.
However, Determining data security is harder today,
so data security functions have become more critical than
they have been in the past [3]. However, there still exist
many problems in cloud computing today, a recent re-
search shows that cloud data storage security have be-
come the primary concern for people to shift to cloud
computing because the data is stored as well as process-
ing somewhere on to centralized location called “data
centers” or CDS. So, the clients have to trust the provider
on the availability as well as data security. Even more
concerning, though, is the corporations that are jumping
to cloud computing while being obliv ious to the implica-
tions of putting critical applications and data in the cloud.
Moving critical applications and sensitive data to a public
and shared cloud environment is a major concern for
corporations that are moving beyond their data center’s
network perimeter defense. The problem of verifying
correctness, confidentiality, integrity and availability for
CDS security becomes even more challenging [4]. CDS
systems are expected to meet several rigorous require-
ments for maintaining users’ data and information, in-
cluding high availability, reliability, performance, repli-
cation and data consistency; but because of the conflict-
ing nature of these requirements, no one system imple-
ments all of them together. For example, availability,
C
opyright © 2012 SciRes. JIS

A. M. TALIB ET AL.
296
scalability and data consistency can be regarded as three
conflicting goals. Security framework is proposed to fa-
cilitate the correctness, confidentiality, availability, and
integrity of user’ data cloud security. Data security on the
cloud side is not only focused on the process of data
transmission, but also the system security and data pro-
tection for those data stored on the storages of the cloud
side. From the perspective of data security, which has
always been an important aspect of quality of service,
cloud computing inevitably poses new challenging secu-
rity threats for a number of reasons:
 Firstly, cloud computing is not just a third party data
warehouse. The data stored in the cloud may be fre-
quently updated by the users, including insertion, de-
letion, modification, appending, reordering, etc. To
facilitate storage correctness under dynamic data up-
date is hence of paramount importance. However, this
dynamic feature also makes traditional integrity in-
surance techniques futile and entails new solutions
[4,5].
 Secondly, the deployment of cloud computing is pow-
ered by data centers running in a simultaneous, coope-
rated and distributed manner. Individual user’s data is
redundantly stored in multiple physical locations to
further reduce the data integrity threats [4]. Therefore,
distributed protocols for storage correctness assurance
will be of most importance in achieving a robust and
secure cloud data storage system in the real world [5].
 Thirdly, CDS systems offer services to assure inte-
grity of data transmission (typically through check-
sum backup). However, they do not provide a solu-
tion to the CDS integrity problem. Thus, the cloud
client would have to develop its own solution, such as
a backup of the cloud data items, in order to verify
that cloud data returned by the CDS server has not
been tampered with.
 Finally, there is lack of fine-grained cloud data access
control mechanism to security-sensitive cloud reso ur c es
[6].
To alleviate these concerns, a cloud solution provider
must ensure that cloud users can continue to have the
same security over their applications and services by
providing evidence to these cloud users that their or-
ganization and cloud users are secure.
In order to achieve these problems we proposed a com-
prehensive security framework based on MAS architec-
ture, our security framework has been built using two
layers: agent layer and cloud data storage layer. The
MAS architecture has five agents: Cloud Service Provid er
Agent (CSPA), Cloud Data Correctness Agent (CD C or A ) ,
Cloud Data Confidentiality Agent (CDConA), Cloud
Data Availability Agent (CDAA) and Cloud Data Inte-
grity Agent (CDIA).
The term “agent” is very broad and has different mea n-
ings to different researchers [7-9]. Genesereth et al. [7],
has gone so far as to say that software agents are applica-
tion programs that communicate with each other in an
expressive agent communication language.
A multi-agent system (MAS) consists of a number of
agents interacting with each other, usually through ex-
changing messages across a network. The agents in such
a system must be able to interact in order to achieve their
design objectives, through cooperating, negotiating and
coordinating with other agents. The agents may exhibit
selfish or benevolent behavior. Selfish agents ask for
help from other agents if they are overloaded and never
offer help. For example, agents serving VIP (Very Im-
portant Person) cloud users for CSP service never help
other agents for the same service. Benevolent agents al-
ways provide help to other agents because they consider
system benefit is the priority. For example, agents serv-
ing normal cloud users for CSP service are always ready
to help other agents to complete their tasks [6].
1.1. Security Goals in Cloud Computing
Traditionally, cloud computing has six goals namely con-
fidentiality, correctness assurance, availability, data in-
tegrity, control and audit. These six goals need to be ful-
filling in order to achieve an adequate security. This paper
focuses in the first four security go als:
1.1.1. Confi dentiali ty
In cloud computing, confidentiality plays a major part
especially in maintaining control over organ izations’ data
situated across multiple distributed cloud servers. Confi-
dentiality must be well achieved when employing a public
cloud due to public clou ds accessibility nature. Asserting
confidentiality of users’ profiles and pro tecting their data
that is virtually accessible, allows for cloud data security
protocols to be enforced at various different layers of
cloud applications [10].
Data access control issue is mainly related to security
policies provided to the users while accessing the data. In
a typical scenario, a small business organization can use
a cloud provided by some other provider for carrying out
its business processes. This organization will have its
own security policies based on which each user can have
access to a particular set of data. The security policies
may entitle some considerations wherein some of the
employees are not given access to certain amount of data.
These security policies must be adhered by the cloud to
avoid intrusion of data by unauthorized users [11].
1.1.2. Correctness Assurance
Goal of correctness assurance in cloud computing is to
ensure cloud users that their cloud data are indeed stored
appropriately and kept intact all the time in the cloud to

A. M. TALIB ET AL. 297
improve and maintain the same level of storage correct-
ness assurance even if cloud users modify, delete or ap-
pend their cloud data files in the cloud [4].
1.1.3. Availability
Availability is one of the most critical information secu-
rity requirements in cloud computing because it is a key
decision factor when deciding among private, public or
hybrid cloud vendors as well as in the delivery models
[10]. The SLA is the most important document which
highlights th e trepidation of availability in cloud serv ices
and resources between the CSP and client. Therefore, by
exploring the information security requirements at each
of the various cloud deployment and delivery models,
vendors and organizations will have confidence in pro-
moting a secured cloud framework.
1.1.4. Dat a Integrity
Integrity of the cloud data has to deal with how secure
and reliable the cloud computing data. This could mean
that even if cloud providers have provided secure back-
ups, addressed security concerns, and increased the like-
lihood that data will be there when you need it. In a cloud
environment, a certification authority is required to cer-
tify entities involved in interactions; these include certi-
fying physical infrastructure server, virtual server, envi-
ronment, user and the net work devices [12].
2. Literature Review
Some argue that cloud user data is more secure when
managed internally, while others argue that cloud pro-
viders have a strong incentive to maintain trust and as
such employ a higher level of security. However, in the
cloud, your data will be distributed over these individual
computers regardless of where your base repository of
data is ultimately stored. Industrious hackers can invade
virtually any server. There are the statistics that show
that one-third of breaches result from stolen or lost lap-
tops and other devices. Besides, there also some cases
which from employees’ accidentally exposing data on the
Internet, with nearly 16 percent due to insider theft [13].
Wang et al. [4], stated that data security is a problem
in cloud data storage, which is essentially a distributed
storage system. And explained their proposed scheme to
ensure the correctness of user’s data in cloud data storage,
an effective and flexible distributed scheme with explicit
dynamic data support, including block update, delete, and
append relying on erasure correcting code in the file dis-
tribution preparation to provide redundancy parity vectors
and guarantee the data dependability. Their scheme could
achieve the integration of storage correctness insurance
and data error localizatio n, i.e., when ever data corrup tion
has been detected during the storage correctness verifica-
tion across the distributed servers, Could almost guaran-
tee the simultaneous identification of the misbehaving
server(s) through detailed security and performance ana-
lysis.
Takabi et al. [14], proposed a comprehensive security
framework for cloud computing environments. They pre-
sented the security framework and discuss existing solu-
tions, some approaches to deal with security challenges.
The framework consists of different modules to handle
security, and trust issues of key components of cloud
computing environments. These modules deal with issues
such as identity management, access control, policy inte-
gration among multiple clouds, trust management be-
tween different clouds and between a cloud and its users,
secure service composition and in tegration, and semantic
heterogeneity among policies from different clouds.
Yu et al. [15], formulated architecture of cloud that
consists of two separated spaces that are the User Space
and Kernal Space. These spaces connected through the
network interface and provide different levels of interac-
tion with in the cloud. The cloud’s Kernal Space is used
to regulate a physical allocation and access control. The
cloud’s User Space contains processes that are directly
used by the cloud users.
Du et al. [16], presented the design and implementa-
tion of RunTest, a new service integrity attestation sys-
tem for verifying the integrity of dataflow processing in
multitenant cloud infrastructures. RunTest employs ap-
plication-level randomized data attestation for pinpoint-
ing malicious dataflow processing service providers in
large-scale cloud infrastructures. They proposed a new
integrity attestation graph model to capture aggregated
data processing integrity attestatio n results. By analyzing
the integrity attestation graph.
Venkatesan and Vaish [17], proposed an efficient multi-
agent based static and dynamic data integrity protection
by periodically verifying the hash value of the files st ored
in the enormous date storage. Their proposed data inte-
grity model is based on the multi-agent system (MAS).
The reason for embedding the agent concept is known,
that is the agent is hav ing capability of autonomous, per-
sistence, social ability and etc. The proposed MAS ar-
chitecture has multiple agents to monitor and maintain
the data integr ity also the architecture in cludes three enti-
ties (respectively customer, service provider and the data
owner).
Looking at the wider technological perspective of
MAS and security in CDS environment has been studied
by Talib et al. [5] proposed a security framework based
on MAS architecture to facilitate security of CDS. Al-
though the illustrative MAS architecture is no t given, the
above should describe the security framework for CDS.
However, this model does not consider the technological
perspective of CDS. Therefore, the main motivation for
this study is to formulate a more detailed security frame-

A. M. TALIB ET AL.
298
work based on MAS architecture for collaborative CDS
environment. The long-term goal of this study is to for-
mulate a tool to support MAS tasks within collaborative
CDS environment. As such, the security framework shall
place more emphasize on the technological perspective.
3. Methodology
Currently, there is a lack of formal a security framework
for collaborative CDS environment [4,5], and there are
no hard and fast rules on how to formulate a security
framework. The investigation of the problems and then
analyzed the formulation of the proposed framework is
taking into account the problems identified from the sur-
vey result. This is very important to make sure the pro-
posed framework is met the objective and the limitation.
So in which there three steps are taken in the methodo-
logy, first conducted a survey and analyzed it, second
analyzed the security framework and lastly the process of
the formulation of the security framework.
A survey was conducted in selected 15 respondents (2
respondents from Information Security Department from
MIMOS Berhad, 7 respondents from Information Secu-
rity Group (ISG) from Faculty of Computer Science and
Information Technology (FSKTM), UPM, 3 security ex-
perts and 3 programmers from different companies) par-
ticipated in this research (pilot study). Thirty three ques-
tionnaires were distributed to the respondents, and fifteen
questionnaires were returned. The questionnaire data
were verified and was analyzed using Rasch Model. The
result of the survey contributed to the formulation of the
proposed security framework.
However, use of Rasch to analyze and validate ques-
tionnaires for theoretical constructs in other technical
fields is still lacking. Whilst the usage of Rasch often
deals with competency evaluation on people or objects,
the usage could also be extended to evaluate another
critical element of research—the research instrument
construct validity [18]. The pilo t data were tabulated and
analyzed using WinSteps, a Rasch tool.
The main components derived from the questionnaire
are: information security concept and understanding, cloud
computing concept and understanding, software agent
concept and understanding, cloud computing security and
CDS based on MAS.
A new security framework shall be synthesized as fol-
lows:
 Structured cloud data, which includes in CDS. There
are many potential scenarios where data stored in the
cloud is dynamic, like electronic documents, photos,
or log files etc.
 The collaborative CDS environment elements are
derived.
 Cloud users and CSPs are considered the main part of
this framework, in which they have to make a SLA
between them in term of facilitating the services by
the CSP and renting these services to the cloud users.
 Agents will act as a tool to facilitate t he security policies.
The proposed security framework based on MAS ar-
chitecture is formulated especially to facilitate the confi-
dentiality, correctness assurance, availability and inte-
grity of CDS and consists of four main componen ts: lay-
ers, cloud users, CSPs and data flow. The layers consist
of the collaboration tools of agents and CDS.
4. Security Fram e work
Figure 1 shows a schematic representation of security
framework. The framework has been built by using two
layers.
The functionality of those layers can be summarized as
follows [4,19]:
 Agent layer: This layer has one agent: the User In-
terface Agent. User Interface Agent acts as an effec-
tive bridge between the user and the rest of the
agents.
 Cloud data storage layer: Cloud data storage has
two different network entities can be identified as
follows:
 Cloud user: Cloud users, who have data to be stored
in the cloud and rely on the cloud for data computa-
tion, consist of both individual consumers and organi-
zations.
 Cloud service provider (CSP): A CSP, who has sig-
nificant resources and expertise in building and manag-
ing distributed cloud storage servers, owns and ope-
rates live cloud computing systems.
5. MAS Architecture
In MAS architecture, we proposed five types of agents:
Cloud Service Provider Agent (CSPA), Cloud Data Con-
fidentiality Agent (CDConA), Cloud Data Correctness
Agent (CDCorA), Cloud Data Availability Agent (CD AA)
and Cloud Data Integrity Agent (CDIA) as illustrated in
Figure 2.
The rest of agents are described as follows:
5.1. Cloud Service Provider Agent (CSPA)
Is the users’ intelligent interface to the system and allow
the cloud users to interact with the security service envi-
ronment. The CSPA provides graphical interfaces to the
cloud user for interactions between the system and the
cloud user. CSPA act in the system under the behavior of
CSP. CSPA has the following actions [6,19]:
 Provide the security service task according to the au-
thorized service level agreements (SLAs) and the or ig i-
nal message content sentby the CDCorA, CDConA,

A. M. TALIB ET AL.
299
Figure 1. Proposed security framework.
User 1 Us
Data Center 1
Data Center 2
er 2
.
B1,B2,B3…Bn
CDConA
CDAA
CDCorA
CSPA
CDIA
Cloud Data Blocks
Figure 2. Proposed MAS architecture.
CDAA and CDIA.
 Display the security policies specified by CSP and the
rest of the agents.
 Designing user interfaces that prevent the input of
invalid cloud data.
 Receive the security reports and/or alarms from the
rest of other agents to respect.
 Translate the attack in terms of goals.
 Monitor specific activities concerning a part of the
CDS or a particular cloud user.
 Creating security reports/alarm systems.
5.2. Cloud Data Confidentiality Agent
(CDConA)
This agent facilitates the security policy of confidential-
ity for CDS. Main responsibility of this agent is to pro-
vide a CDS by new access control rather than the exist-
ing access control lists of identification, authorization
and authentication. This agent provides a CSP to define
and enforce expressive and flexible access structure for
each cloud user [6]. Specifically, the access structure of
each cloud user is defined as a logic formula over cloud
data file attributes, and is able to represent any desired
cloud data file set. This new access control is called as:
 Formula-based cloud data access control (FCDAC).
This agent is also notifies CSPA in case of any fail
caused of the techniques above by sending security re-
ports and/or alarms.
Formula-Based Cloud Data Access Control (FCDAC)
and also named as a SecureFormula it’s an access policy
determined by our MAS architecture, not by the CSPs.
It’s also define as access is granted not based on the
rights of the subject associated with a cloud user after
authentication, but based on attributes of the cloud user.
In our system, CDConA provide access structure of each
cloud user by defining it as a logic formula over cloud
data file attribute. SecureFormula is an additional confi-
dentiality layer used by our system to verify that the
cloud users’ login page is a genuine.
If you are a cloud user, you are required to register
first to the system and write your valid email and enter
your SecureFormula during your first login. Your Se-
cureFormula will be sent to your email. Be ensured that,
your SecureFormula is not your password. Do not set
your SecureFormula to be the same as your password!
Sign in from your computer [6]:
1) Enter your Cloud User ID;
2) Verify that your SecureFormula image is correct;
3) Confirm by entering your password.
Our confidentiality layer guaranteed that, even if your
password is cor rect and your SecureFormula is inco rrect,
then you will not be able to login.
The architecture of CDConA consists of five modules,
as shown in Figure 3. Cloud Communication Module
provides the agent with the capability to exchange in-
formation with other agents, including the CDConA,
CDCorA, CDAA, CDIA and CSPA. Cloud Register
Module facilitates the registration fun ction for CDConA.
Cloud Request Management Module allows the agent to

A. M. TALIB ET AL.
300
Existing
access control
CDConA Cloud Reasoning
Module
Cloud Reque s t
Management
Cloud Resour c e
Management
Clo ud Re g is te r
Module
Cloud Communication Module
Cloud Coordination
Module
Cloud Communication Module
Confidentially
Policy Rules
Knowledge
Base
CSPA
CDCorA
Figure 3. CDConA architecture.
act as the request-dispatching center. Cloud Resource
Management Module manages the usage of the cloud
resources. Cloud Reasoning Module is the brain of the
CDConA. When the request management module and
resource management module receive requests, they pass
those requests to reasoning module by utilizing the in-
formation obtained from the knowledge base and the co n-
fidentiality policy rule.
5.3. Cloud Data Correctness Agent (CDCorA)
This agent facilitates the security policy of correctness
assurance for CDS. Main responsibility of this agent is to
perform various block-level operations and generate a
correctness assurance when the cloud user performs up-
date operation, delete operation, append to modify opera-
tion or insert operation. This agent notifies CSPA in case
of any fail caused of the techniques above by sending
security reports and/or alarms.
The architecture of the CDCorA consists of four mo-
dules, as shown in Figure 4. Cloud Communication
Module provides the agent with the capability to ex-
change information with CSPA. Cloud Coordination
Module provides the agent with the following mecha-
nisms. If the data is updated then the data encryption is
performed. If the data is deleted then the data encryption
is performed. If the data is Append then the data encryp-
tion is performed. If the data is inserted then the data
encryption is performed. Cloud Reasoning Module cal-
culates the necessary amount of cloud resources to com-
plete the service based on the required service level
agreements (SLA) by utilizing the information obtained
from the knowledge base and the correctness assurance
policy rule. Cloud Services Module performs the block-
level operations of encryption and decryption when the
cloud user update, delete, append and inser t his/h e r data.
In CDS, there are many potential scenarios where data
stored in the cloud is dynamic, like electronic documents,
photos, or log files etc. Therefore, it is crucial to consider
the dynamic case, where a cloud user may wish to per-
form various block-level operations of update, delete and
Cloud Reasoning
Module
Cloud Service
Module
Correctness
Policy Rules
Knowledge
Base
CDCorA
CSPA
CDConA
Figure 4. CDCorA architecture.
append to modify the data. Our proposed correctness
assurance protocol is not going to be genuine if there is
absent of SecureFormula. So in case of: Update operation:
The cloud user needs to enter his/her SecureFormula plus
00, Delete operation: The cloud user needs to enter
his/her SecureFormula plus 01, Append operation: The
cloud user needs to enter his/her SecureFormula plus 10
and Modify operation: The cloud user needs to enter
his/her SecureFormula plus 11.
5.4. Cloud Data Availability Agent (CDAA)
This agent facilitates the security policy of availability
for CDS. Main responsibility of this agent is to receive
and display the security issu es th at offer by its sub -ag ents
of CDDPA and CDRA. CDAA facilitate two new tech-
niques of file distribution preparation and file retrieval.
This agent is also notifies CSPA in case of any fail c au s ed
of the techniques above by sending security reports and/
or alarms.
Cloud data availability is to ensure that the cloud data
processing resources are not made unavailable by mali-
cious action. Our MAS architecture is able to tolerate
multiple failures in cloud distributed stor age systems.
To ensure the availability, we explain the notions of
global and local cloud attack blueprints. To detect intru-
sions, the CDAA receives a set of goals representing the
global cloud attack blueprints. To recognize this global
cloud attack blueprint, it must be decomposed in local
cloud sub-blueprints used locally by the different agents
distributed in the CDS. In general agents can detect only
local cloud attacks because they have a restricted view of
the CDS. So, we make a distinction between a global
cloud attack blueprint and local cloud sub-blueprints. A
global cloud blueprint is an attack blueprint, derived
from the security policies specified at a high level by the
CSPs, that the MAS must detect and the detection of this
blueprint will be notified only to CDAA. A local cloud
blueprint is a blueprint derived from the global cloud
blueprint but that must be detected by local agents. For a
CDAA over-viewing the global cloud attack blueprint the
probability of an attack is equal to 1, while for the local

A. M. TALIB ET AL. 301
agent it is below 1.
Cloud Reasoning
Module
The architecture of the CDAA consists of three modules,
as shown in Figure 5. Cloud Communication Module
provides the agent with the capability to exchange infor-
mation with CDAA and CSPA. Cloud Servers Modules
provides the agent with the following mechanisms: 1)
Disperse the data file redundantly across a set of distri-
buted servers; and 2) Enable the cloud user to reconstruct
the original data by downloading the data vectors from
the servers. Cloud Reasoning Module provides the CD AA
with the specific misbehaving server(s) and server col-
luding attacks by utilizin g the information obtained from
the knowledge base and th e availability policy rule.
Cloud Comm u nication Module
5.5. Cloud Data Integrity Agent (CDIA)
This agent facilitates the security policy of integrity for
CDS. It is used to enable the clo ud user to reco nstruct the
original cloud data by downloading the cloud data vec-
tors from the cloud servers. Main responsibility of this
agent is backing up the cloud data regularly from “Cloud
Zone” and sending security reports and/or alarms to
CPSA when [20]:
 Human errors when cloud data is entered.
 Errors that occur when cloud data is transmitted from
one computer to another.
 Software bugs or viruses.
 Hardware malfunctions, such as disk crashes.
Our proposed integrity layer named as “CloudZone”.
In CloudZone, we introduce the first provably-secure and
practical backup cloud data regularly that provide recons-
truct the original cloud data by downloading the cloud
data vectors from the cloud servers.
“CloudZone” Requirements
 “CloudZone” only backs up the MS SQL databases. It
does not back up other MS SQL files such as program
installation files, etc.
 “CloudZone” does not support component-based back-
up.
 “CloudZone” does not use Visual SourceSafe (VSS)
for backup and restor e.
 “CloudZone” supports backup and recovery of Win-
dows Oracle 10 g.
With “CloudZone” Cloud Backup, you can select
any of the f ol lowing as bac kup objects :
 Oracle Server 10 g running on Windows.
 Microsoft SQL Serv er 2000, 2005 and 2008.
 Microsoft Exchange Server 2003 and 2007.
The architecture of the CDIA consists of three modules,
as shown in Figure 6. Cloud Communication Module
provides the agent with the capability to exchange in-
formation with CDIA, CDConA, CDCorA, CDAA and
CSPA. Cloud Resources Management Modules provides
Cloud Servers
Module
Availability
Policy Rules
Knowledge
Base
CSPA
CDA
CDIA
Figure 5. CDAA architecture.
Cloud Reasoning
Module
Cloud Communication Module
CDIA
Cloud Resource
Management
Module
Integrity
Policy Rules
Knowledge
Base
CSPA
CDA
Figure 6. CDIA architecture.
the agent with the following mechanisms. If the CDIA
registered as CDIA-VIP then back-up of the data is per-
formed successfully. If the CDIA did not register as
CDIA-VIP, it asks the cloud user to back-up the data
manually. Cloud Reasoning Module sh ows the r easons of
in case the result of the back-up the data is failed by utiliz-
ing the information obtained from the knowledge base
and the integrity policy rule [20].
6. Implementation
Ganawa Security as a Service (GSecaaS) has been im-
plemented (~30.000 lines of JAVA code) with Oracle 11
g. The implementation was based on structure- in-5 MAS
architectures described above. We briefly describe the
GSecaaS implementation to illustrate the role of the
agents and their interaction. To simulate the agen ts, Ora-
cle database packages and triggers are used to implement
agent functions and Oracle jobs are utilized to create
agents. Each agent is considered as an instance of the
agent in the environment that can work independently,
and can commu nicate with other agents in order to fulfill
its needs or fulfill the others requests. To demonstrate the
feasibility of the proposed system, a prototype is imple-
mented using Java and PHP.
At the interface layer, the interaction of the system
with the cloud user is based on a set of dialogues. These
dialogues are implemented using Java and PHP. An exam-
ple of an interface is shown in Figure 7.

A. M. TALIB ET AL.
302
7. Pilot Study spread of person responses is = 3.29 logit is fair. This is
due to extreme responses by a participant. However, Re-
liability = 0.82 and Cronbach Alpha = 0.94 indicates
high reliable data and hence the data could be used for
further analyses.
7.1. Result
The pilot data were tabulated and analyzed using Win-
Steps, a Rasch tool. The results of Person and Item sum-
mary statistics and measures are tabulated in Tables 1
and 2.
On the questionnaire items, the summary of 15 mea-
sured questionnaire items (Table 2) reveals that the
spread of data at 2.36 logit and reliability of 0.74 are
good and fair, respectively.
The results of the survey are analyzed in three parts;
data reliability, fitness of respondent and items data and
determ i nation of component groups cut-off points. Details on each measured items are listed in Table 3.
The acceptable limits are 0.4 < Acceptable Point Mea-
sure Correlation < 0.8 and 0.5 < Outfit Mean Square <
1.5, and –2.0 < Outfit z-standardized value < 2.0). The
previous pilot study is therefore proven helpful in mak-
ing the questionnaire more reliable.
7.1.1. Data Reliability
Summary statistics for respondents (persons) and items
(questions) are depicted in Tables 1 and 2, respectively.
15 respondents returned the survey questionnaire. Out of
which, Rasch identified an extreme score which will later
be excluded from further analysis. 7.1.2. Fi t ne ss of Respo nd ent Data and Questionna ire
Items Data
From the summary of measured persons (Table 1) , the
A Person-Item Differential Map (PIDM) is used to reveal
the “easiest” and “hardest” questions answered by res-
pondents. Based on the summaries and PIDM, a few ob-
servations could be concluded. Person SU1 is at the
leftmost of the person distribution. Rasch provides the
Person Item Distribution Map (PIDM), which is similar
to histogram (Figure 8). PIDM allows both person and
items to be mapped side-by side on the same logit scale
to give us a better perspective on the relationship of per-
son responses to the items. PIDM indicates a higher Per-
son Mean (0.64) compared to the constrained Item Mean.
This indicates tendency to rate higher importance to the
prescribed questionnaire items.
Figure 7. An example of interaction window with a cloud
user (confidentiality laye r ).
Table 1. Summary of measured persons.
Infit Outfit
Raw score Count MeasureModel errorMNSQ ZSTD MNSQ ZSTD
MEAN 133.8 42.8 0.49 0.27 1.02 –0.2 1.01 –0.2
S.D. 14.9 3.5 0.69 0.02 0.52 2.1 0.53 2
MAX. 167 45 2.64 0.34 3.14 6.4 3.37 6.7
MIN. 86 30 –0.65 0.25 0.28 –4.5 0.28 –4.4
Real RMSE 0.30 Adj. S.D. 0.62 Separation 2.10 Person reliability 0.82 Model RMSE 0.27 Adj. S.D. 0.64 Separation 2.35 Person reliability
0.85 S.E. of p erson mea n = 0 .11 Maximum extreme score: 1 Pe rson valid responses: 95.0%.
Table 2. Summary of measured items.
Infit Outfit
Raw score Count MeasureModel errorMNSQ ZSTD MNSQ ZSTD
MEAN 119.8 38.3 0.02 0.3 1 0 1 0.1
S.D. 16.7 3.2 0.64 0.08 0.12 0.6 0.15 0.7
MAX. 150 40 1.16 0.6 1.29 1.5 1.4 1.9
MIN. 88 29 –1.2 0.2 0.83 –1.3 0.74 –1.3
Real RMSE 0.32 Adj. S.D. 0.54 Separation 1.69 Item reliability 0.74 Model RMSE 0.27 Adj. S.D. 0.64 reparation 2.35 Item reliability 0.75
S.E. of item mean = 0.09.

A. M. TALIB ET AL. 303
Table 3. Items statistics—Measure order.
Item Raw ModelInfit Outfit Pt Mea
No. Item ScoreCountMeasureS.E. MNSQZStd MNSQ ZStd Corr.
A cloud data storage (CDS)
1 A1 roles 139 15 –1.18 0.3 0.98 0 1 0.1 0.31
2 A2 resources 127 15 0.12 0.22 0.88 –0.7 0.84 –0.8 0.44
3 A3 infrastructure 132 15 –0.32 0.26 0.88 –0.6 0.85 –0.7 0.43
4 A4 req analys is 128 14 –0.26 0.27 0.84 –0.8 0.81 –1 0.46
5 A5 sys analysi s 129 15 –0.4 0.23 0.97 0 1.13 0.6 0.37
6 A8 implementation 124 14 –0.01 0.27 0.84 –0.7 0.84 –0.8 0.48
7 A7 domain 138 15 –0.82 0.28 1.04 0.3 1 0.1 0.28
B cloud user
8 B1 behavior 106 11 –0.56 0.39 1.06 0.3 1.1 0.4 0.24
9 B2 awareness 111 12 –0.53 0.28 1.19 0.6 1.27 0.9 0.13
10 B3 usage 94 11 0.05 0.26 0.9 –0.2 0.93 –0.1 0.48
C cloud service provider (CSP)
11 C1 facilitate 150 15 –0.72 0.38 0.89 –0.6 0.8 –0.7 0.31
12 C2 encourage 90 15 0.92 0.24 1.27 1.2 1.25 1.1 0.38
13 C3 provide 89 15 0.99 0.23 1.06 0.4 1.07 0.4 0.49
14 C4 trust 107 14 0.09 0.27 1.03 0.2 1 0.1 0.43
D agent tools
15 D1 definition 112 13 0.17 0.43 0.89 –0.2 0.88 –0.2 0.47
16 D2 characteristic 95 12 0.76 0.46 0.95 0 0.91 –0.1 0.31
17 D3 communication 126 10 0.2 0.6 0.83 –0.2 0.74 –0.3 0.56
18 D4 prosperity 128 12 0.76 0.46 0.95 0 0.91 –0.1 0.49
19 D5 goal 132 12 0.76 0.46 0.95 –0.2 0.94 –04 0.76
E security goals in cloud computing
20 E1 confidentiality 145 15 –0.09 0.34 0.86 –1.3 0.79 –1.3 0.39
21 E2 correctness assurance 137 15 0.81 0.34 0.9 –0.9 0.87 –1 0.41
22 E3 availability 126 15 –0.25 0.35 0.9 –0.3 0.87 –0.4 0.49
23 E4 integrity 116 15 0.75 0.26 0.95 –0.2 0.96 –0.2 0.45
24 E5 data privacy 131 39 1.13 0. 3 5 1.1 0.8 1.14 0.9 0.26
25 E6 multi-tenancy 123 39 –0.5 0.39 1 0.1 1.05 0.3 0.4
26 E7 control 134 15 1.16 0.35 1 0 1.01 0.1 0.35
Mean 119.838.30.0 0.3 1.0 0.0 1.0 0.0 0.4
S.D. 16.7 3.2 0.6 0.1 0.1 0.6 0.2 0.7 0.1
PIDM is used to reveal the “easiest” and “hardest”
questions answered by respondents. Based on the sum-
maries and PIDM, a few observations could be con-
cluded. Person SU1 is at the leftmost of the person dis-
tribution. As the Customer Service Director, it’s lonely
up there and not many want to share with him informa-
tion, hence the pattern of answers. Item F1—“Cloud user
must pay in order to get the cloud services”, and
F5—“Agents have the ability to pass the parameters
among them” are on the rightmost and leftmost of the
Item distribution, respectively. The question for F1 is on
“Cloud user must pay in order to get the cloud services”
Strategy and F5 is on “Agen ts have the ability to pass the
parameters among them” strategy. We believe that re-
spondents might not understand the terms “Cloud user
must pay in order to get the cloud services” and “Agents
have the ability to pass the parameters among them” in
cloud computing concept and software agent concept.
Layman-terms were used to better represent the questions.
In this case, questions F1 and F5 were rephrased to
F1—“both of these strategies the respondent must totally
agreed”. Determining the “Easy” qu estions is not as easy
as portrayed in the Person-Item Variable map. It was
envisaged that question F1 and F5 were revised.
7.1.3. Compo nent Group Cut-Off Poin t s
There are no hard and fast rules on how to determine
which of the less important components should be ex-
cluded from the framework. The components are sorted
into descending logit values. The list is then distributed

A. M. TALIB ET AL.
304
Figure 8. Person-item distribution map.
to four experts from software engineering fields, and thr ee
cloud computing security experts.
7.2. Discussion
Based on the overall experts’ judgments, the following
components are selected to be excluded from the model
(Table 3):
 C2 Encourage/CSPs must encourage cloud users to
use their trusted CDS.
 D1 CSPA—Provide the security service task accord-
ing to the au thorized service level agreements (SLAs)/
different area.
 E5 Data privacy/different area.
 E6 Multi-tenancy/different area.
 E7 Control/different area.
Based on the above reduced components, the revised
framework is depicted in Figure 1 and its MAS archi-
tecture in Figure 2. Based on the Pilot study results, the
revised security framework based on MAS architecture is
directly driven from the initial framework. This is be-
cause the most of the components are common and used
to identify the respondent in the questionnaire.
The proposed security frameworks to facilitate secu-
rity of CDS are based on Wang et al. [4], Talib et al. [5],
Takabi et al. [14], Yu et al. [15], Du et al. [16] and
Venkatesan and Vaish [17], they all runs in six main
parts layers, functions, security goals, infrastructures, ap-
proaches, technologies and applications and overlaps on
some specific components are architectures and collabo-
rations. The major comparison on the major components
of all above frameworks is depicted in Table 4.
8. Conclusion
In this paper, we investigated the problem of data secu-
rity in cloud computing environment, to ensure the con-
fidentiality, correctness assurance, availability and inte-
grity of users’ data in the cloud; we proposed a security
framework and MAS architecture to facilitate security o f
CDS. This security framework consists of two main lay-
ers as agent layer and cloud data storage layer. The pro-
pose MAS architecture includes five types of agents:
CSPA, CDConA, CDCorA, CDAA and CDIA. To for-
mulate the security framework for collaborative CDS
security, the components on MAS, cloud user and CSP

A. M. TALIB ET AL. 305
Table 4. Comparisons between the frameworks.
Item/Framework Wang et al. [4] Talib et al. [5]Takabi et al. [14]Yu et al. [15]Du et al. [16]Venkatesan and Vaish [17]
Layer Y Y Y NA Y NA
Function Y Y Y NA Y Y
Security goal Y Y Y Y Y Y
Infrastructure Y Y Y Y Y Y
Approach NA Y Y Y Y Y
Technology Y Y Y Y Y Y
Application Y NA Y Y Y NA
Architecture NA Y NA NA Y Y
Collaboration Y Y Y Y Y Y
are compiled from various literatures. An initial model of
modified MAS components for collaborative CDS secu-
rity is proposed. The relationships between these com-
ponents are used to construct the questionnaire, which
were tested in a pilot study. Rasch model was used in
analyzing pilot questionnaire. Item reliability is found to
be poor and a few respondents and items were identified
as misfits with distorted measurements. Some problema-
tic questions are revised and some predictably easy ques-
tions are excluded from the questionnair e. A prototype of
the system (GSecaaS) is implemented using Java and
PHP. The use of this system has shown how the system
could be used to facilitate the security of the CDS.
REFERENCES
[1] M. Zhou, R. Zhang, W. Xie, W. Qian and A. Zhou, “Se-
curity and Privacy in Cloud Computing: A Survey,” Pro-
ceedings of the Sixth International Conference on Seman-
tics Knowledge and Grid (SKG), Beijing, 2010, pp. 105-
112.
[2] C. S. Aishwarya, “Insight into Cloud Security Issues,”
UACEE International Journal of Computer Science and
Its Applications, 2011, pp. 30-33.
[3] J. W. Rittinghouse and J. F. Ransome, “Cloud Computing:
Implementation, Management, and Security (Chapter 6),”
2009.
[4] C. Wang, Q. Wang, K. Ren and W. Lou, “Ensuring Data
Storage Security in Cloud Computing,” IEEE, Vol. 186,
No. 978, 2009, pp. 1-9.
[5] A. M. Talib, R. Atan, R. Abdullah and M. A. A. Murad,
“Formulating a Security Layer of Cloud Data Storage
Framework Based on Multi-Agent System Architecture,”
TGSTF International Journal on Computing, Vol. 1, No.
1, 2010, pp. 120-124.
[6] A. M. Talib, R. Atan, R. Abdullah and M. A. A. Murad,
“Towards New Access Data Control Technique Based on
Multi Agent System Architecture for Cloud Computing in
Software Engineering and Computer Systems Part II,” In:
V. Snasel, J. Platos and E. El-Qawasmeh, Eds., Springer
Series: Communications in Computer and Information
Science 189, Springer-Verlag, pp. 268-279.
[7] M. R. Genesereth and S. P. Ketchpel, “Software Agents,”
Communication of the ACM, Vol. 37, No. 7, 1994, pp.
48-53.
[8] E. H. Durfee, V. R. Lesser and D. D. Corkill, “Trends in
Cooperative Distributed Problem Solving,” IEEE Transac-
tions on Knowledge and Data Engineering, 1989, pp. 63-
83.
[9] H. Mouratidis, P. Giorgini and G. Manson, “Modelling
Secure Multi-Agent Systems,” ACM, 2003, pp. 859-866.
[10] S. Ramgovind, M. M. Eloff and E. Smith, “The Manage-
ment of Security in Cloud Computing,” Information Se-
curity for South Africa (ISSA), Sandton, Johannesburg,
2010, pp. 1-7.
[11] K. D. Bowers, A. Juels and A. Oprea, “HAIL: A High-
Availability and Integrity Layer for Cloud Storage,” 2009.
http://eprint.iacr..org/2008/489.pdf
[12] D. Zissis and D. Lekkas, “Addressing Cloud Computing
Security Issues,” Future Generation Computer Systems,
Vol. 28, No. 3, 2010, pp. 583-592.
[13] J. Yang and Z. Chen, “Cloud Computing Research and
Security Issues,” International Conference on Computa-
tional Intelligence and Software Engineering (CiSE),
2010, pp. 1-3.
[14] H. Takabi, J. B. D. Joshi and G. J. Ahn, “SecureCloud:
Towards a Comprehensive Security Framework for Cloud
Computing Environments,” 34th Annual IEEE Computer
Software and Applications Conference Workshops, 2010,
pp. 393-398.
[15] H. Yu, N. Powell, D. Stembridge and X. Yuan. “Cloud
Computing and Security Challenges,” ACM, 2012, pp.
298-302.
[16] J. Du, W. Wei, X. Gu and T. Yu, “RunTest: Assuring
Integrity of Dataflow Processing in Cloud Computing In-
frastructures,” ASIACCS’10, Beijing, 13-16 April 2010,
pp. 293-304.
[17] S. Venkatesan and A. Vaish, “Multi-Agent Based Dy-
namic Data Integrity Protection in Cloud Computing,”
2011, pp. 76-82.
[18] A. A. Aziz, A. Mohamed, A. Zaharim, S. Zakaria, H. A.
Ghulman and M. S. Masodi, “Evaluation of Information
Professionals Competency Face Validity Test Using
Rasch,” Proceedings of the 4th Pacific Rim Objective

A. M. TALIB ET AL.
306
Measurement Symposiom (PROMS), 2008, pp. 396-403.
[19] A. M. Talib, R. Atan, R. Abdullah and M. A. A. Murad,
“Security Framework of Cloud Data Storage Based on
Multi Agent System Architecture: Semantic Literature
Review,” Computer and Information Science, Vol. 3, No.
4, 2010, p. 175.
[20] A. M. Talib, R. Atan, R. Abdullah and M. A. A. Murad,
“CloudZone: Towards an Integrity Layer of Cloud Data
Storage Based on Multi-Agent System Architecture,” ICOS,
2011, pp. 127-132.