Implementation of a Novel Peer-to-Peer Reputation-Based Trust Management Model in a Cloud Service Provisioning Environment

Implementation of a Novel Peer-to-Peer Reputation-Based Trust Management Model in

a Cloud Service Provisioning Environment

Nosipho Dladlu and Obeten O. Ekabua

Department of Computer Science

North-West University, Mafikeng Campus

Private Bag X246, Mmabatho 2735 South Africa {nosipho.dladlu, obeten.ekabua}@nwu.ac.za

Abstract - In recent times, coupled with the emergence of

the new cloud paradigm, a number of security related

issues have emerged. Therefore, trust and reputation

management has become unavoidable challenge in this new

cloud computing paradigm. Cloud computing represents a

shared and statistical service model that allows software,

platform and infrastructure to be shared as service, with

high degree of openness and autonomy. Consequently these

high degree of openness and autonomy impose the need for

surveillance of the services, and profer nearly an ideal

condition for a spread of unauthentic files. Although peer-

to-peer (P2P) models and their applications are popular

depending on their domain of applicability, protecting

peer’s security remain a challenge. As a result, this

research paper presents a novel Peer-to-Peer reputation-

based trust management model in a cloud provisioning

environment. To validate the proposed model, a simulation

environment with different peers and service provider was

used in CloudAnalyst environment tool.

Keywords: Peer- to-Peer, Cloud Computing, Reputation,

Trust Management, Provisioning Environment

1. Introduction

Cloud computing refers to “application and service” offered

over the internet [1]. These services are offered from data

centers all over the world, which collectively are referred to

as the cloud. It is defined as “a type of parallel and

distributed system consisting of a collection of

interconnected and virtualized computers that are

dynamically provisioned, and presented as one or more

unified computing resources based on service-level

agreements established through negotiations between the

service provider and consumers” [1]. Cloud computing

environment is increasing with every passing day and the

more it advances the more insecure it becomes. This cloud

computing environment is characterized by lack of central

controlling authority. This kind of computing usually

consists of illegitimate partners or peers which may publish

intentionally false or misleading information.

Consequently, the responsibility of ascertaining which

information should be trusted, falls squarely on each

participating agency, otherwise known as peers, making

trust management a challenge. More so, cloud computing consist of three levels on which

services offered to the consumer varies according to the

abstraction level of the service. The cloud computing model

services include Platform as a Service, Infrastructure and

Software as a Service, as illustrated in Figure1.

The first and lowest level is Infrastructure as a Service

(IaaS) and services are supplied in a form of hardware,

Figure 1 Cloud Computing as Service [6]

example Amazon EC2 [2, 3]. The second level is Platform

as a Service (PaaS). In this level cloud consumers do not

have to handle virtual machines. Instead a software

platform for hosting application such as web applications

which is already installed in an infrastructure and offered to

the consumers. Examples of PaaS are Google App Engine

and Aneka [4, 5]. The third level is Software as a Service

(SaaS) this is where an application is offered to consumers.

In this level consumers do not have to handle virtual

machines and software platforms that host the application

[3]. Reputation-based trust management for P2P offers an

environment where users rate reliability of parties they

communicate with and share information with their peers

[10]. These trust management systems are mainly used in

electronic markets for assessing the participants. In such

environment they are proved to be effective as number of

participants is large and the system is running a sufficient

amount of time [12]. But not everywhere reputation system

are more effective, there are still lot of issues in reputation-

based trust system.

However, paying attention into trust management is

important, since it is one of the most important issue in a

Peer-to-Peer system. Service Providers (CS) and consumers

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 210

are more concerned about trust in a cloud computing

environment. Although trust management is the most

important issue in cloud computing there are some

challenges regarding peer’s evaluation in P2P network in a

cloud environment. These challenges include: dishonest

feedback from malicious peers, collusions and complex

strategy fraud.

Therefore based on such challenges, there is a need to

develop P2P reputation-based trust management model to

solve those challenges within the cloud environment. The

main focus of this paper is to evaluate the recommendation

of peers to see if they are trustworthy or not.

2. Related Work

Several researchers offered trust and reputation-based

models to approach the problem of spreading malicious or

inauthentic file such as CuboidTrust [14], EigenTrust

[15,16], MGTrust [17], PeerTrust [18], GridPeerTrust [19],

AntRep [20, 21], PowerTrust [23] and others.

CuboidTrust model was proposed in 2007 and was defined

as global reputation trust management model that builds

four relations among three trust sources including

trustworthiness of peer’s (where there was a reporting

feedback), contribution of peer to peer system and quality

of resource [14]. It applies power iteration to compute the

global trust value of each peer. However, this model

introduces the concept of pre-trusted peers that cannot be

applicable in all cases. Direct and indirect trusts in

CuboidTrust model are not given a different treatment, they

don’t have different mark but are treated as the same.

EigenTrust model assigns a unique global trust value to

each peer in a P2P file sharing network, and this approach

is based on peers history of uploads and this is achieved by

a decrease in the number of downloads of inauthentic files

[15,16]. However both CuboidTrust and EigenTrust model

concept introduces pre-trusted peers. Although they are

useful models, they are not applicable in all cases because

not all sets of peers can be trusted always.

MGTrust model is a new global trust model based on

recommendations. This is a concept of nearness of fuzzy

set, where the term nearness is used in fuzzy set and

describes the similarity between two fuzzy subsets [17].

This model shows how to compute trust value in distributed

way and test the convergence property of MGTrust as well.

However MGTrust was better compared to EigenTrust and

DouWen’s model.

PeerTrust model was proposed in 2006. It is a model that

considers a coherent adaptive trust model for quantifying

and comparing the trustworthiness of peers based on a

transactions-based feedback system [18]. PeerTrust is a

reputation-based trust supporting framework, and it consist

of two features: Firstly, it introduces three basic trust

parameters and adaptive factors in computing

trustworthiness of peers such as feedback peer receives

from peers, the total number of transactions peer performs,

the credibility of the feedback sources, transaction context

factor and the community context factor. Secondly, it

defines a general trust metric to combine parameters.

However, PeerTrust does not differentiate between

confidence placed on peer when carrying out a task and

when giving recommendations. It assumes that peer with

higher trust value always gives more reliable feedback than

a peer with a lower trust value which might not be true.

PeerTrust model was improved in 2012, and the proposed

model was called GridPeerTrust which incorporated

PeerTrust in the grid environment along with improving

some drawbacks of PeerTrust model. GridPeerTrust is

designed to handle drawbacks of PeerTrust by changing

definition of satisfaction criteria and adding a decay

function in algorithm [19]. Where in the satisfaction

criteria, trust parameter deals with the number of desired

features fulfilled by the resource provider. In performing

grid to select a resource provider for performing grid

service the basic needs desired by resource consumer. The

algorithm is based on introducing decay function that is

updated with feedback trust calculation algorithm.

AntRep model [20, 21] was proposed by different authors,

it is a model where reputation evidences are distributed

over P2P network. These authors proposed the use of ant

system for building trust relationships in P2P network

efficiently and this model has ability to easily adapt to the

dynamic topology of P2P network. This is caused by the

use of colony system. The disadvantage of AntRep model is

that it provides a mechanism to distribute reputation

evidences not to assess evidence.

Another work in 2011 was designed to support customers in

reliably identifying trustworthy cloud providers [22]. A

multi-faceted Trust Management (TM) system architecture

for a cloud computing marketplace was proposed in order

to carry out the task in identifying trustworthy cloud

providers in terms of different attributes (e.g. security,

compliance and performance) assessed by multiple source

roots of trust information. TM aims at supporting customers

to identify trustworthy service providers as well as

trustworthy service providers to stand out.

PowerTrust model is another model that was proposed in

2007 and is defined as a robust and scalable P2P reputation

system, which leverages the power low feedback

characteristics found applicable in dynamically growing

P2P networks either structured or unstructured. PowerTrust

was developed as a model that builds a trust overlay

network (TON) on top of all nodes in a P2P system, where

every peer evaluates each other whenever a transaction

takes place between pair of them. PowerTrust has been

proved to be resistant against an individual malicious peer

attack and it is also resilient against malicious collectives

with camouflage and driving down the reputation of

reliable peers [23]. However PowerTrust is vulnerable to

malicious pre-trusted peers threats. Because of this power

nodes are considered as a fully reliable peers (as pre-

trusted peers in EigenTrust), so if the power peers become

malicious they can cause a great damage in the system.

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 211

Path Trust model was introduced in 2009 [24], it considers

the problem of using a reputation system to enhance the

member selection in virtual organization. They identified

different attacks, analyzed their impact and threats using

reputation system, however their system was not reliable.

Metadata model for IR was proposed in 2013, it focuses on

the development of reputation management system for

efficient selection of disaster management team. They

proposed Service Oriented Architecture to extract

information from information source and their reputation

score is calculated using an algorithm [23]. However the

reputation in this system is not credible, because the

information is just extracted from information source and it

is not checked.

3. Trust in Cloud Computing

According to the review and related research, trust

management is very important and necessary to ensure

trustworthiness among peers.

We introduce reputation-based trust model to establish

ranking of trustworthy peers in a private cloud, private

cloud is chosen simple because we want to distribute and

manage data within the environment. Private cloud is a

cloud-based service, data and process are managed within

the organization or cloud environment (CE). All the peers

in the CE have unique identity. We assume that peers in CE

interact with each other through transaction with the

mediator called Cloud Broker (CB). CB integrate cloud and

data center environment as the process layer. CB can

enforce a variety of application delivery-related policies,

are an architectural remedy to challenge of managing

distributed applications [25]. For example, an identity

broker mediates authentication and authorization processes

for cloud-deployment applications and devices as a means

to centralize and maintain control over access and accounts.

The trust model that we propose is based on the following

parameters:

• Storage Capacity

• Processing Capacity

• Data cost

• Link

The brief description of these parameters is shown in

Table1.

Table 1: Reputation-based Trust Management Parameters

Parameters Description

Storage Capacity The workload/resources stored by a

peer.

Processing

Capacity

The average workload processed by

the peer.

Data Cost The cost of resources that a peer

offers/ requests.

Link The number of times that a peer

interacts with other peers will

determine better Communication

link.

When the peer wants to communicate with other peers, it

selects the trusted peer to store the data based on the above

parameters. For instance, if the peerX is to be

recommended as a trustworthy peer it needs to have a better

communication link and in storage capacity parameter, the

peer chooses which peer to communicate with based on the

storage resources. This implies greater trust values, since

both storage capacity and link parameters increase the peers

capacity of transmitting and receiving data. If the peer

processing capacity is always utilized it takes longer to

attend any demands, therefore this consumes much time.

For instance, if peerX wants to purchase an item or wants to

receive an item it will place an order and wait on the queue

while the order is processed.

However, in this model a trust rank is established, allowing

peerX to determine whether it is possible to trust peerY, as

to perform storage operation or to receive an item in a cloud

environment. The trust value of peerY is determined by

considering the basic information first. When peerX needs

to communicate with peerY and does not know anything

about peerY it will request information to the CB and other

peers.

Figure 2 Trust Relation

Figure 2 represents the trust relation, where peerX review

the trust behavior of peerY through CB. The CB is

responsible of calculating whether the peer is trustworthy or

not based on the obtained results. Peer’s trustworthiness

indicates the quality of the peer’s services. It is often used

to establish the future behavior of the peer. Similarly, if a

peer is trustworthy, it is likely to provide good services in

future transactions. Unlike the peer which is not

trustworthy, it can provide the wrong information to other

peers and is recognized as a cheater, because of the

cheating probability.

CB is the one responsible for validating trust, it computes

trust based on the reputation which is based on the above

parameters. It consists of two table such as, direct and local

recommendation table to store the information sent by

peers. If the peer wants to check the reputation about other

peer it check the direct table first to see if the information

exists. The recommendation list table will be checked to see

if the is any information about the peer. If the is no

information in both tables about that particular peer, then

the request will be sent to the CB. The CB will broadcast

the information to other peers in the cloud environment

CB

PeerX PeerY

Request/Receive Request/Receive

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 212

(CE). After the service provider receives the feedback from

all peers in CE it will be able to compute if the peer is

trustworthy or not based on the reputation received from

other peers. Peers that send the feedback to the CS and

other peers, assign a greater trust value to the peers that

have great storage capacity, processing capacity, better link

and manageable data cost.

Figure 3 Recommendation Process

4. P2P Reputation-Based Trust Model

In this section we discuss p2p reputation-based trust

management model that we formulated.

In our model, when peerX accepts the communication with

peerY by simply accepting services offered by peerY, after

that peerX will evaluate peerY’s services. By sending

feedback to the CB and other peers on the cloud

environment. The CB will validate or verify the trust value

sent by peers by calculations. This computes trust based on

the reputation and the reputation is based on the parameters

as defined in table 1.

The trust between two peers is modeled using the following

equations:

𝑇𝑟 = ∑𝐴𝐶𝐴

𝑖

∞

𝑖=1

, 𝑇𝑟 𝜖 [0,1] (1)

Where,

𝑇𝑟 – probability trust in cloud environment

𝑖 – interaction

ACA – available attributes in cloud environment

The trust values are calculated from queries between peers

of CE. This allows peers to obtain necessary information

for final trust calculations.

The final trust calculation takes place in mediator of peers,

which is the CB. To calculate final trust value of a peer, it is

attributed by the CB of the cloud. The final trust is

represented by the following equation, which is derived

from equation 1:

𝑇𝑟(𝑥,𝑦) = ∑ 𝐹𝐶𝐴𝑦

𝑖

𝐶𝐴−1

(2)

𝑇𝑟(𝑥,𝑦) = (𝐶𝑥 ×40)+(𝑃𝑥 ×40)+(𝐷𝑥 ×40)+(𝐴𝑥 ×40)

𝑖 (3)

Where,

𝐹𝐶𝐴𝑦

– represent the final trust value of peerY in CE

𝑇𝑟(𝑥,𝑦) - is the trust of x in y in CE

𝐶𝑥 - represent storage capacity

𝑃𝑥 - represent processing capacity

𝐴𝑥 - represent link

𝐷𝑥 - represent data cost

5. Experimental Setup

The model was implemented using simulation tool in a

cloud service environment with peers, such as peerX and

peerY in private cloud. The peers are reachable to each

other and all peers have a unique identity. CloudAnalyst

simulator is used, since CloudAnalyst helps developers

with the insights on how to distribute applications among

cloud infrastructure and value added services such as

optimization of application performance. CloudSim

simulation environment also produces the interaction

between infrastructure providers as Service (IaaS) and their

customers [24]. CloudAnalyst is developed on top of

CloudSim, with new extensions introduced such as

Application users, Internet, Simulation defined by time

period and Service Brokers.

The configuration of machine used in this work is shown in

table 2.

Table 2: Configuration of Machine

Values

Memory RAM Size 4.00 GB

System Type 64-bit OS

Processor Intel Core i7-2620M

After we configure and define the simulation environment

of CloudAnalyst with the weights of metric assigned as

illustrated in section 4. The performance of the calculation

of trust of a peer can be done.

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 213

Figure 4 Region Showing Map

The experimental design map is shown in figure 2 above.

Where we have, two (2) geographically located user bases

named peerX and peerY were created with one (1) data

center named CB as shown in table 4 and table 3.

Table 3a Datacenter (Service Broker)

Table 3b Datacenter

Shown in table 3a and 3b is the parameters attributed by the

administrator in order to calculate trust of peer. Where

Storage Capacity and Processing Capacity/Speed with

weights 80%, the link 10% and the remaining of 10% to

Data transfer. Knowing that any peer can have its trust

value ranging from 0 to 1, and knowing that these values

vary in time. The reason why storage capacity and

processing capacity weights are higher than link and data

transfer is because these features are the responsibility for

ensuring the integrity and reliability. In table 3a the data

center variable like OS, and Physical HW Units are

constant. The physical machine of data center (CB) uses

x86 architecture and is running on Linux operating and Xen

virtual machine manager. The output results obtained from

these input parameters are discussed in section 6

Table 4: User Bases (Consumers)

The table above illustrates the consumers named as peerX

and peerY in different regions.

6. Results and Discussion

We looked at four aspects in calculating trust of the peer

such as storage capacity, processing capacity, link and data

cost as discussed in section 4 and 5. A larger storage and

processing capacity have a greater weight in a choice of

more reliable peers, because they show integrity. Each user

base is linked to CB, it is here that all the peers are

weighted in, after they have submitting the

recommendation about each other. The CB consists of

different visual machines (VM), in our experiments we

allocated two VM for each datacenter and the CB is able to

identify which machine performs well compared to the

other. We rated each VM according to the table 5.

Table 5: References values for Trust rate

Trust Average

Amount

Meaning Description

Greater than 500 Fully trust Completely

trustworthy

Between 100 and 500 Average Peer is trustworthy

Less than 100 Distrust Peer is completely

untrustworthy

According to the weights attributed it is possible to

calculate the trust of a peer. Analyzing the results of the

simulations, it is possible to identify the trust level of each

peer. The table above illustrates the parameters used in

order to show the category where each peer falls in, after

getting results from cloudanalyst simulation tool.

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 214

Figure 5 Simulation window

Figure 5 above represent the simulation process, showing

all the peers in cloud environment. It shows the threshold of

virtual machine in CB and peers (User bases) after the

administrator attributed parameters. In our proposed

scenario some machine have zero performance of cloudlet,

this is because they did not fulfill the checking condition of

reliable machine to perform task, compared to baseline

machine. CB1 do not satisfy the trust level desirable simple

because it did not perform any cloudlet, and the processing

cloudlet is zero. This simple means that CB1 is considered

as untrustworthy, even though it got response from other

peers it didn’t respond.

6.1 Processing Capacity

According to the parameters attributed by the administrator

the processing capacity for peerY is higher than the one of

peerX. After simulation we have found out that peerY has

much greater processing capacity compared to peerX and in

terms of trustworthiness it is trustworthy than peerX as

shown in the figure below.

Figure 6 Processing Capacity graph

0

200

400

600

Trust Avg prossesingcapacity

Avg

tru

st a

mo

un

t

parameters

Processing Capacity

peerX

peerY

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 215

6.2 Storage Capacity

Figure 7 below illustrate the storage of peerY and peerX.

Figure 7 Storage capacity

Consequently, when it come to storage the trust average of

peerY is higher than that of peerX. This shows that when it

comes to storage capacity peerY is responsoble to store

much data which makes peerY reliable and trustworthy.

6.3 Link and Data cost

The link and data cost determines who the peer is

interacting or connected with other peer based on

reputation. Figure 3 shows how each peer interacts (sending

and reciving resources) with other peers and the data cost of

each CB.

Figure 8 Trust worthiness

From the experiments we carried out, it is significant that

peerY more is trustworthy than peerX.

7. ConclusionThe trust model for peer-to-peer reputation-based trust

management in a cloud provisioning environment is

reported in this paper. Our model illustrates how the peers

evaluate and recommend other peers. Also, presented is a

mathematical model with parameters and variables that are

implemented in CloudAnalyst simulation tool. Setting and

varying the parameters is a function of the administrator.

The main contribution of this research paper is therefore the

development and implementation (as a proof of concept) of

a novel peer-to-peer reputation-based trust management

model, to enhance the provisioning of services in a cloud

environment where issue of trust and reputation is a

challenge.

8. References

[1] R. Buyya, C. S. Yeo, S. Vehugopa, J. Broberg and I.

Brandic. “Cloud computing and emerging IT platforms:

Vision, hope and reality for delivering computing as 5th

utility.” Future Generation Computer Systems, vol. 25,

pp.599-616, June. 2009.

[2] R. M. Vozmediano, R.S. Montero, I. M. Lloiente.

“Multicloud deployment of computing cluster for loosely

MTC applications.” IEEE Transactions on parallel and

distributed systems, vol. 22, pp. 924-930, June 2011.

[3] B. Wiekremasingne, R. N. Calheiros and R. Buyya.”

CloudAnalyst: A cloudsim-based visual modeler for

analyzing for cloud computing environments and

applications,” in Proc.24th IEEE International Conference

on Advanced Information Networking and Application

(IANA), 2010, pp. 446-452.

[4] R. Prodan, M. Sperk, S. Osterann. “Evaluating high-

performance computing on Google App Engine.” IEEE

Journal on Software, vol. 29, pp. 52-58, June.2012.

[5] C. Vecchiola, S. Pandey, R. Buyya. “High-

Performance cloud computing: A view of scientific

applications,” in Proc.10th International Symposium on

Pervasive Systems, Algorithms and Networks (ISPAN),

2009, pp. 4-16.

[6] K. Hwang and D. Li. “Trusted cloud computing with

secure resources and data colouring.” IEEE Internet

Computing Journal, vol. 14, pp. 14-22, Sep.2010.

[7] G. Suryanarayana, R. Erenkrantz, S. A. Hendrickson

and R. N. Taylor. “An architectural style for trust

management in decentralized applications,” Proceeding of

the 4th working IEEE Conference on Software Architectural

(WICSA), 2004, pp. 1-10.

[8] Hai Ren. “Comparison of trust model in P2P

systems,” TKK-T Seminar on Network Security, 2006, pp.

1-5.

[9] A. Samreen, S. Hussain. “Trust management and

inceptive mechanisms for P2P network: Survey to cope

challenges,” in Proc. IEEE International Multitopic

Conference (INMIC), 2008, pp. 301-306.

[10] A. A. Selcuk, E. Uzen and M. R. Pariente. “A

reputation-based trust management system for P2P

networks,” in Proc.IEEE International Symposium on

Cluster computing and Grid (CCGrid), 2004, pp. 251-258.

0

200

400

600

Trust Avg Storage

Avg

tru

st a

mo

un

t

parameter

Storage Capacity

peerX

peerY

peerX peerY

0

0.5

1

1.5

peers

Tru

st r

atio

(m

b)

Trust Worthiness

Trust Avg

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 216

[11] T. Grandson, S. Morris. “A survey of trust in internet

application.” in Proc. IEEE Communications Surveys and

Tutorials, vol. 3, pp. 2-16. Dec.2000.

[12] P. Resmic, Z. Richard, E. Friedman and K.O

Wabura. “Reputation Systems.” Communication of ACM,

vol. 43, pp.1-10, Dec. 2000.

[13] S. Moalla, S. Hamdi and B. Defude. “A new trust

management model in P2P systems,” in Proc. IEEE 6th

International Conference on Signal-Image Technology and

Internet based systems, 2010, pp. 241-246.

[14] R. Chen, X. Chao, L. Tang, J. Hu and Z. Chen.

“CuboidTrust: A global reputation-based trust model in P2P

networks,” in Proc. 4th ATC International Conference,

2007, pp. 201-215.

[15] S. Kmvar, M. Schhlosser and H.M. Garcia.

“EigenTrust algorithm for reputation management in P2P

networks,” in Proc. of International World Wide Web

(WWW), 2003, pp. 1-8.

[16] T. Nishikawa and S. Fujita. “An effective risk

avoidance scheme for the EigenTrust reputation system

management system,” in Proc. 1st International Conference

on Network and Computing (ICNC), 2010, pp. 36-43.

[17] W. Wang, X. Wang, S. Pan and P. Liang. “A new

global trust based on recommendation for P2P networks,”

in Proc. International Conference on New Trends in

Information and Service (NISS), 2009, pp. 325-328.

[18] W. Wang, G. Zeng and L. Yuan. “Ant-Based

reputation evidence distribution in P2P networks,” in Proc.

5th International Conference on Grid and Cooperative

computing, 2006, pp. 129-132.

[19] D. Kaur and J. S. Gupta. “Proposed P2P reputation-

based model to secure grid,” in Proc. International

Conference on Recent advances Trends in Information

Technology (iRAFIT), 2012, pp. 19-24.

[20] M. Dingo, V. Manienzo and A. Colorni. “The Ant-

System optimization and colony of cooperating agents.”

IEEE Transaction on System, vol. 26, pp.30-40, June. 1996.

[21] O. Cordon, F. Herrera and T. Stutzle. “A review on

the Ant colony optimization mataheuristic: Basics, models

and new trends.” Mathware and Soft Computing Journal,

vol. 9, pp. 141-175, 2002.

[22] F. G. Marmol and G. M. Perez. “Security threats

scenarios in trust and reputation models for distributed

systems.” Computer and Security Journal, vol. 28, pp.545-

556, May.2009.

[23] S. Javaid, A.Majeed and H. Afzal. “A Reputation

Management System for Efficient Section of Disaster

Managment Team.” In Proc. International Conference on

Advance communication Technology (ICACT), 2013,pp.

829-834.

[24] C.N. Rodrigo, R. Ranjan, A. Beloglazov, De. Rose,

A. F. Buyya. Rajkumar. “CloudSim: A Toolkit for

Modeling and Simulation of Cloud Computing

Environments and Evaluation of Resource Provisioning

Algorithms, Software: Practical and Experience (SPE)”

Wiley Press, vol 41, pp. 23- 50, ISSN: 00380644, New

York USA, 2011.

ISBN: 978-0-9891305-1-6 ©2013 SDIWC 217