Mobile Agents-Based Framework for Dynamic Resource Allocation in

Cloud Computing

Saﬁa Rabaaoui

1 a

, H

ela Hachicha

1,2

and Ezzeddine Zagrouba

1 b

University of Tunis El Manar, Higher Institute of Computer Science, Laboratory of Informatics, Modeling and Information

and Knowledge Processing (LIMTIC), 2 Rue Abou Rayhane Bayrouni, Ariana 2080, Tunisia

College of Computer Science and Engineering, University of Jeddah, Saudi Arabia

Keywords:

Cloud Computing, Multi Agent-System, Mobile Agent, Dynamic Resource Allocation, Cost, Makespan, Task

Scheduling.

Abstract:

Nowadays, cloud computing is becoming the more popular technology for various companies and consumers,

who beneﬁt from its increased efﬁciency, cost optimization, data security, unlimited storage capacity, etc.

One of the biggest challenges of cloud computing is resource allocation. Its efﬁciency directly inﬂuences the

performance of the whole cloud environment. Finding an effective method to address these critical issues

and increase cloud performance was necessary. This paper proposes a mobile agents-based framework for

dynamic resource allocation in cloud computing to minimize the cost of virtual machines and the makespan.

Furthermore, its impact on the best response time and task rejection rate has been studied. The simulation

shows that our method gave better results than the former ones.

1 INTRODUCTION

The cloud is a term used to refer to a network of re-

mote servers that are accessed over the Internet to

store, manage, and process data instead of relying on a

local server or personal computer (Yahya et al., 2020).

It provides on-demand access to a shared pool of com-

puting resources that can be rapidly provisioned and

scaled to meet users’ needs, including storage, pro-

cessing power, and applications. The cloud enables

users to store and access data and applications from

anywhere with an internet connection, eliminating the

need for local storage and computing infrastructure. It

offers ﬂexibility, scalability, and cost-effectiveness, as

users can pay for the resources they use rather than in-

vest in and manage their own hardware and software

infrastructure. One of the main challenges in cloud

computing is resource allocation due to the dynamic

nature of resources (Kumar et al., 2018). In fact,

with companies and consumers expanding their re-

quirements, the need for efﬁcient resource allocation

is also emerging. Also, the cost of cloud resources

represents a serious concern among many companies

and consumers. This is why cloud resource alloca-

https://orcid.org/0000-0001-5978-5613

https://orcid.org/0000-0002-2574-9080

tion is a topic of discussion, and this is to meet the

current demand in the best response time and reduce

cloud services costs. This issue is a challenge in

cloud computing systems. Computing services must

be highly dependable, scalable, and autonomous to

support ubiquitous access, dynamic discovery, and

composability (Buyya et al., 2008). Resource allo-

cation in cloud computing faces challenges, includ-

ing cost efﬁciency, response time, computational per-

formance, and scheduling tasks (Belgacem, 2022).

Users of cloud computing services target to accom-

plish tasks with the lowest costs possible. Research

in multi-agent systems (MAS) and mobile agents has

evolved considerably, and several distributed systems

have been deployed using these technologies. Indeed,

MAS is based on autonomous and intelligent agents

sharing a typical environment. These agents cooper-

ate and interact with each other to achieve a global

goal. While there are differences between cloud com-

puting and multi-agent systems, they are both tech-

nologies designed for distributed models. Thanks to

their advantages and features, many issues in differ-

ent ﬁelds can be solved by integrating or combining

cloud and multi-agent systems, as in (Qasim et al.,

2020). Also, in (Bei et al., 2022), the authors pro-

posed a fair and efﬁcient multi-resource allocation for

cloud computing. In addition, research has been per-

766

Rabaaoui, S., Hachicha, H. and Zagrouba, E.

Mobile Agents-Based Framework for Dynamic Resource Allocation in Cloud Computing.

DOI: 10.5220/0012390200003636

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 16th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2024) - Volume 3, pages 766-773

ISBN: 978-989-758-680-4; ISSN: 2184-433X

formed in recent years to provide architectures for us-

ing the cloud in MAS, such as (Venkateshwaran et al.,

2015), which presented a framework for agent ne-

gotiation while using the resources provided by the

cloud environment. Other research focuses on using

agent technologies to address cloud computing difﬁ-

culties. Among these, we mention the work of (Ma-

havidyalaya and Nadu, 2021), (Barkat et al., 2021),

and (Yahya et al., 2020), who recommended employ-

ing MAS to address cloud challenges. The authors

of (FAREH, 2015) presented an architecture based on

self-organizing agents to deal with the difﬁculty of

cloud service composition. Current researchers han-

dle these issues dependently without achieving an op-

timal solution. Therefore, this work focuses on dy-

namic resource scheduling for efﬁcient resource allo-

cation. Synchronously, we must maintain customer

service quality by minimizing the makespan and cost

of virtual machines. The contributions of our research

are:

• Create a dynamic resource allocation model to in-

crease resource utilization and user satisfaction by

minimizing task rejection rate.

• Propose a new Dynamic Resource Allocation ap-

proach with Mobile Agents DRAMA to minimize

the cost and makespan.

This paper is structured as follows. Section 2 is a list

of related works. Section 3 shows the problem state-

ment and formulation. Section 4 describes our pro-

posed solution. Experiment results and the implemen-

tation of a proposed prototype are reported in Section

5. Finally, Section 6 concludes the paper by identify-

ing our plans.

2 RELATED WORKS

This section will discuss some of the most relevant

frameworks that focus on using multi-agent systems

and mobile agents for resource allocation in cloud

computing. In (Soltane et al., 2018), authors have

proposed a cloud architecture based on a multi-agent

system exhibiting a self-adaptive behavior to address

dynamic resource allocation DRA. The principal fo-

cus of this work is to enhance energy consumption

while satisfying the quality of service QoS demanded

by users. This architecture consists of four agents: an

analyzer agent who identiﬁes the resources and ser-

vices required by users and builds speciﬁc queries.

The scheduling agent is responsible for allocating

resources users need and making the ﬁnal decision

about resource allocation. The controller agents track

the status of resources in the data center. The coordi-

nator agent supervises the whole process.

Wang et al.(Wang et al., 2016) have deﬁned a de-

centralized multi-agent-based Virtual Machine (VM)

allocation approach. The approach aims to allocate

VMs to Physical Machines (PMs) while minimizing

system energy costs. This approach allows dispatch-

ing a cooperative agent to each PM to assist the PM in

managing resources. Another solution based on agent

technology, called low-level resource distribution, is

proposed by (Bajo et al., 2016). This approach allows

the distribution of computational resources through-

out the entire cloud computing infrastructure, con-

sidering its complexity and associated computational

costs. In this system, agents are distributed over the

infrastructure. Each physical server in the cloud envi-

ronment contains a set of stationary agents in charge

of monitoring and making decisions that involve as-

signing or removing nodes for a particular service.

Each service offered to the cloud users is associated

with two agents, one for monitoring and the other

for control; both are responsible for ensuring compli-

ance with the previously established SLA agreement.

Other agents also ensure the proper operation of the

cloud computing system. On the other hand, some ap-

proaches have been proposed using mobile agents. In

(Singh et al., 2017), a new mechanism was proposed

that deploys various intelligent agents to reduce the

cost of virtual machines and resource allocation com-

plexity. This system deﬁnes four stationary agents

and one mobile agent, which searches the resources

from the available resource instances of a current data

center. The mobile agents can manage resource allo-

cation.

Further, Aarti Singh et al. (Singh and Malhotra,

2015) proposed using mobile agents for resource allo-

cation in cloud computing, focusing on cost optimiza-

tion. The end users send the resource request to the

cloud data center, where all the resources are avail-

able. Every cloud is associated with a cloud Mobile

Agent (MAc). Every MAc is responsible for all in-

formation on resources and their status, whether free

or allocated. Initially, a service request arrives at an

MAc and checks the available free resources to de-

cide whether the request can be served. After that, the

resource manager agent (RMA) will determine how

it would be allocated, i.e., which technique should be

applied to resources so that they will be adequately

distributed and the cost of VMs minimized. Finally,

resources are distributed to the user.

Belgacem Ali et al. (Belgacem et al., 2020) focus

on dynamic resource allocation. The authors present

a multi-objective search algorithm called the Spac-

ing Multi-Objective Antlion algorithm (S-MOAL) to

Mobile Agents-Based Framework for Dynamic Resource Allocation in Cloud Computing

767

minimize the makespan and the cost of virtual ma-

chines. However, they did not consider the system’s

energy consumption and fault tolerance issues. In

the work (Abdullah and Surputheen, 2022), the au-

thors present the Cooperative Agents Dynamic Re-

source Allocation and Monitoring in Cloud Comput-

ing CADRAM system. They include more than one

agent to manage and observe resources provided by

the service provider. The solution increases resource

utilization and decreases power consumption while

avoiding SLA violations. He presented an algorithm

to select a virtual machine called the node failure dis-

covery. Hence, VMs are allocated to the user based

on the type of the job.

3 PROBLEM STATEMENT AND

FORMULATION

This research focuses on the dynamic resource alloca-

tion in Infrastructure as a Service (IaaS). In the IaaS

model, computing resources are provisioned and de-

livered over the Internet in a virtualized environment.

IaaS provides users with virtualized infrastructure

components such as virtual machines, storage, and

networking capabilities. The cloud resources provider

makes virtual machines available to customers when

it sends requests to determine their resource require-

ments via the graphical user interface GUI. Each vir-

tual machine is conﬁgured with its own processors

and memory resources, which gives an overview of

a resource pool. These resources are dynamically re-

served and released. The ﬁrst aim of our research

is to develop a dynamic resource allocation system

that acts as an intermediary between cloud providers

and users. This system allows us to connect these

two partners: cloud users who want to choose the ap-

propriate resources that meet their needs and cloud

providers who seek to maximize their beneﬁts and al-

locate their resources efﬁciently. We decided to use

mobile agents and multi-agent systems that are widely

used to design and develop complex and distributed

systems. The cloud computing system has j Datacen-

ters:

DC = {DC

,DC

,...,DC

} (1)

Each data center comprises a set of physical nodes

(PNs):

PN = {PN

,PN

,...,PN

} (2)

Where p signiﬁed the number of physical nodes

in the data center, and many virtual machines (VMs)

reside on each physical node. Each VM is a resource

modeled as a set:

V M = {V M

,V M

,...,V M

} (3)

Where m signiﬁed the number of virtual machines

running on PN at a given time.

To handle this problem, we consider that the input

is a set of tasks and a set of VMs, where:

T = {T

,...,T

} (4)

Represent a set of tasks. Output obtaining a better

mapping of T

to V M

, (T

,V M

) in order to reduce

the cost and the makespan. We consider a parallel-

machine scheduling problem with n Tasks and m ma-

chines: The makespan is the time taken to complete

all the Tasks when optimally allocated across the ma-

chines. The equation can be represented as:

M(t) = max(m

,...,m

) (5)

Where m

,...,m

represent the completion times

of each machine.

m(n) =

∑

) (6)

Where p

represents the processing time of each Task

n assigned to machine m. The cost of VM utilization

is denoted by C is the sum prices of running all the

tasks on the VM, as shown in (7).

C(t) =

∑

i=0

(t) (7)

(t) = Cost

CPU

+Cost

ram

+Cost

bandwidth

(8)

4 PROPOSED FRAMEWORK

The proposed dynamic resource allocation with mo-

bile agents (DRAMA) model permits to handle with

the main issues: minimizing the makespan, the cost,

and the rejection rate of tasks. As depicted in Fig. 1,

our architecture is structured into three layers, each

with speciﬁc roles and responsibilities within the ar-

chitecture.

1. User Layer: it provides an interface to access the

cloud services through which users specify their

resource allocation needs.

2. Resource Allocation Layer represents an inter-

layer between the upper layer (users) and the

lower layer (data center infrastructure).

3. Data Center Infrastructure Layer: it provides re-

sources as services. It consists of a physical layer

(physical machines, hosts) and a virtual layer (Vir-

tual Machines, VMs).

ICAART 2024 - 16th International Conference on Agents and Artiﬁcial Intelligence

768

Figure 1: The Architecture’s Layers.

The proposed architecture shown in Fig. 2 in-

volves a set of stationary and mobile agents interact-

ing and collaborating to achieve a common goal. In

this architecture, we have integrated the concept of

cloud providers. Within the same provider, it is pos-

sible to have many data centers which are connected

through a network.

For this reason, we have deﬁned a new class of

mobile agents (Data Center Management Agents) that

has a mission to migrate from one data center to an-

other related to one cloud provider. Also, we have as-

signed each user request to one mobile agent (Alloca-

tion Agent) that migrates from one provider to another

to make allocation decisions and deﬁne the allocation

plan. In the following, we describe each agent’s roles

and internal architecture in our architecture.

Request Agent RA: is a stationary agent that acts

as an intermediary between the user and the system.

RA is responsible for retrieving the data sent from the

user interface and formulating the user’s request. Af-

ter receiving the requests, RA organizes the requests

(tasks) based on a task scheduling algorithm to use the

available resources correctly. Every task is scheduled

by a scheduling algorithm in such a way that each task

is forwarded to the allocation agent. It is also account-

able for building speciﬁc queries (Request Q). The re-

quest Q delivered by the request agent is composed of

three elements, as follows:

Q = (R,Cp,Qos) (9)

Where:

R = (r

,...,r

) (10)

Equation (10) represents resources and services the

user needs (such as data storage and virtual ma-

chines).

Cp = (cp

,cp

,...,cp

) (11)

Cp is the capacity corresponding to each resource;

Qos = (Qos

,Qos

,...,Qos

) (12)

Qos represents the Qos for each resource. We

represent the combination of resources and their

corresponding requested capacities and Qos to build

the suitable V M

as:





r1 cp1 Qos1

r2 cp2 Qos2

r3 cp3 Qos3





(13)

The Request Agent uses the Task Scheduling al-

gorithm 1 to assign the tasks to allocation agents. The

priorities for each task are assigned based on the job

size and inter-arrival time. Then, the tasks are con-

verted into priority-based tasks represented by:

T Q = (T Q

,T Q

,...,T Q

) (14)

Then, they will search for the priority queue.

Input : Tasks T={t1,t2,,,,,,,tn};

Output : connections by the Allocation Agent;

for all t1,t2,,,,,tn in Task scheduler do

Consider for all Task length Sx;

if the current task length = next task length

then

Sort the request according to their

arrival time At;

Add Task T to Qe based on At;

EnQueue Qe ;

else

Sort the request according to their

length;

EnQueue Qe;

end

while Qe!= Null do

Create Allocation Agent AA ;

Add Task TQ to Allocation Agent AA;

end

Algorithm 1: Scheduling Algorithm.

Input : Tasks T={t1,t2,,,,,,,tn};

Output : best offer (available resources ) ;

1. AA migrates to CSA (provider representative)

to select the appropriate resources for the Task;

2. Receives responses from each CSA (sets of

available resources);

3. It ﬁlters and selects the best offer;

Algorithm 2: Allocation Agent Algorithm.

Allocation Agent AA: is a mobile agent that aims to

ﬁnd the best offer appropriate to the user’s request in

the data-center infrastructure. It receives the task and

migrates to all cloud providers to select the appropri-

ate resources for the received request. It participates

Mobile Agents-Based Framework for Dynamic Resource Allocation in Cloud Computing

769

Figure 2: The Architecture’s Cloud Providers and Agents.

in the system and aims at obtaining maximum bene-

ﬁt for the user (optimization of the cost-performance

ratio, minimization of the cost).

Cloud Supervisor Agent CSA: It represents the

provider in the cloud environment. This agent is re-

sponsible for cooperating with the allocation agents

and assigning the data-center management agent the

mission to ﬁnd the virtual machine corresponding to

the user’s needs. The cloud supervisor agent offers an

offer (list of resources to be allocated, prices, etc.) to

respond to a given task. To do this, it cooperates with

the Data-Center Management Agent in order to ex-

ploit the resources available in the different data cen-

ters.

Input : needed resources;

Output : available resources ;

1. CSA sends the request to DCMA;

2. It receives all available resources;

3. It proposes an offer (list of resources to be

allocated);

Algorithm 3: Cloud Supervisor Agent Algorithm.

Data Center Management Agent DCMA: It discov-

ers resource capacities on physical nodes and man-

ages virtual machines in DC. This agent has visibility

into all resources in DC from the same provider. It

maintains the current state of physical machines and

tracks resource status. It is responsible for the coor-

dination between the different service agents. When

it receives a Task from the cloud supervisor agent,

it migrates to all service agents to discover the re-

source capabilities of the different virtual machines,

then prepares the best answer with the minimum cost

and transfers it to the Cloud Supervisor Agent.

Input : CPU, RAM, storage, cost

cpu

, cost ram,

cost

storage

,Bandwidth, cost

bandwidth

;

Output : TotalCost;

for all VM1,VM2,,,,,VMm do

TotalCost

CPU

= CPU * cost

cpu

;

TotalCost

ram

= ram * cost

ram

;

TotalCost

storage

= storage *

TotalCost

bandwidth

= Bandwidth *

cost

bandwidth

;

TotalCost = total

ost

pu+ total

ost

am +

total

ost

torage+ total

ost

andwidth

end

Algorithm 4: TotalCost Algorithm.

Service Agent SA: is responsible for maintaining the

capacity of physical nodes and handling virtual ma-

chines in the data center. It is responsible for dis-

covering the current state of each machine (capac-

ity used, free capacity, state). It is also responsible

for maintaining information on the status of resources

and characteristics. It also has the mission to man-

age resources (allocation, freeing). For each physical

machine exists one Service Agent.

5 SIMULATION AND RESULTS

This section presents the performance and result of

the proposed framework throughout the implementa-

tion of three prototypes. We have used the CloudSim

simulator platform (Calheiros et al., 2011) to simu-

late the cloud infrastructure and the JADE platform

to implement our multi-agent system. The cloud data

center is created by using a series of PMs. Each data

center with multiple numbers of hosts and their cor-

responding VMs is initiated. In the ﬁrst prototype,

ICAART 2024 - 16th International Conference on Agents and Artiﬁcial Intelligence

770

we implement an allocation system without a multi-

agent system. In the second prototype, we implement

our proposed framework (Rabaoui et al., 2021).

In this architecture, we have only one mobile

agent that migrates from one data center to another

without considering the cloud provider. In the third

prototype, we implement our proposed architecture

using two mobile agents: allocation agents that mi-

grate from one cloud provider to another and data cen-

ter management agents that migrate from one service

agent to another within the same cloud provider.

Table 1 provides the cloud sim speciﬁcation of

the proposed system. Different metrics, such as

makespan, execution time, Refusal Rate and cost are

used to analyze the simulation results. The proposed

approach (DRAMA) improves system efﬁciency and

is compared with some existing methods in the liter-

ature, including the First-Come First-Served (FCFS)

algorithm and the Preference Based Task Scheduling

(PBTS) algorithm.

Table 1: Simulation Speciﬁcation.

Parameter Value

The number of Hosts 4

The number of Tasks 30-300

Length of task 1600-3400

Memory 540

Bandwidth 25,00,00

Storage 500 GB

MIPS/PE 500

We use our proposed method. At t=0, a set of tasks

arrive simultaneously. Then, jobs are prioritized by

the task scheduling concept employed in our frame-

work. The prioritized tasks are given to the allocation

agents. AA will migrate and send the same received

job to all cloud supervisor agents (CSA1, CSA2, etc.)

listed in its directory. CSA forwards the task to one

of the Data Center Management agents (DCMA) that

it holds.

DCMA migrates to Service Agents (SA) to re-

trieve virtual machine states. The service agent sends

the current status of each virtual machine to the data

center management agent. DCMA compares the re-

sponses from the service agents to select the best so-

lution that meets the user’s requirements and informs

the cloud supervisor agent. The CSA receives the

best-selected response and sends its offer to the al-

location agent.

AA receives the offers of the CSA, ﬁlters them,

and sorts them according to the price. It then selects

the best suggestion that meets the user’s needs. It

sends an allocation request to the CSA if it receives

an allocation conﬁrmation and sends the answer to

the RA. The AA gets an error message from the se-

lected CSA if it is not conﬁrmed. The AA returns an

allocation request to the next CSA in its sorted list of

offers. It enters a loop until it receives an allocation

conﬁrmation from a CSA, or it determines its circle

and fails to allocate a resource, in which case it sends

an error message to RA.

Fig.3 shows the makespan according to the num-

ber of tasks. This measurement reveals to us what

time is needed to receive a response from the system.

The simulation result gives a shorter Makespan when

compared with the other two taken algorithms (FCFS

and PBTS).

Figure 3: Comparison of makespan.

The examination of the proposed approach in

terms of cost is presented in Fig. 4. It is apparent

from the comparison of the three scheduling algo-

rithms that the proposed method, DRAMA, decreases

the value of the cost due to the completion of each

task in an appropriate virtual machine that ensures a

minimum price. Therefore, the proposed method al-

lows a better match between the tasks and the virtual

machines.

Figure 4: Costs for different numbers of tasks.

A comparative graph shown in Fig. 5 shows our

proposed method’s lesser response time than existing

methods.

The examination of the proposed approach in

terms of the Refusal Rate is presented in Fig. 6.

Re f usalRate = Total

Tr j

/Total

T sb

(15)

Mobile Agents-Based Framework for Dynamic Resource Allocation in Cloud Computing

771

Figure 5: Comparison of response time.

Where: Total

Tr j

is the total number of tasks re-

jected, and Total

T sb

is the total number of tasks sub-

mitted. It indicates how many tasks are dismissed to

the total number of submitted jobs and how many are

accepted. It shows that our proposed methods have

signiﬁcantly lower refusal rates, indicating that they

get more tasks than rejected ones.

Figure 6: Refusal Rate.

6 CONCLUSIONS

In this paper, we proposed a Mobile agents-based

framework for a Dynamic Resource Allocation ap-

proach with Improved task scheduling in cloud com-

puting environments. Our system has two classes of

mobile agents and three classes of stationary agents:

mobile agents play a crucial role in choosing the

most suitable offer for user demand. This agent en-

sures a secure exchange and greatly minimizes traf-

ﬁc on the network. The stationary agents manage lo-

cal information. The performance evaluation results

can signiﬁcantly improve the makespan, the response

time, and service quality (minimizing VM cost). This

method combines the dynamic change of the cloud

system state and the task scheduling to solve various

issues, which makes it effective.

Three experiments were used to evaluate the per-

formance of the proposed method, DRAMA. The ﬁrst

shows good results regarding the tasks’ makespan,

cost, and rejection rate. However, there is still much

work to be done. For our future work, we plan to

present a formal model of our approach deﬁning the

query for resource allocation and user preferences.

We also plan to describe in detail the scheduling al-

gorithm and give additional experiments to evaluate

other performance aspects and the security for the re-

source allocation.

REFERENCES

Abdullah, M. and Surputheen, M. M. (2022). Cadram-

cooperative agents dynamic resource allocation and

monitoring in cloud computing. IJCSNS, 22(3):95.

Bajo, J., De la Prieta, F., Corchado, J. M., and Rodr

ıguez,

S. (2016). A low-level resource allocation in an agent-

based cloud computing platform. Applied Soft Com-

puting, 48:716–728.

Barkat, A., Kazar, O., and Seddiki, I. (2021). Framework

for web service composition based on qos in the multi

cloud environment. International Journal of Informa-

tion Technology, 13:459–467.

Bei, X., Li, Z., and Luo, J. (2022). Fair and efﬁcient multi-

resource allocation for cloud computing. In Interna-

tional Conference on Web and Internet Economics,

pages 169–186. Springer.

Belgacem, A. (2022). Dynamic resource allocation in cloud

computing: analysis and taxonomies. Computing,

104(3):681–710.

Belgacem, A., Beghdad-Bey, K., Nacer, H., and Bouz-

nad, S. (2020). Efﬁcient dynamic resource allocation

method for cloud computing environment. Cluster

Computing, 23(4):2871–2889.

Buyya, R., Yeo, C. S., and Venugopal, S. (2008). Market-

oriented cloud computing: Vision, hype, and reality

for delivering it services as computing utilities. In

2008 10th IEEE international conference on high per-

formance computing and communications, pages 5–

13. Ieee.

Calheiros, R. N., Ranjan, R., Beloglazov, A., De Rose,

C. A., and Buyya, R. (2011). Cloudsim: a toolkit for

modeling and simulation of cloud computing environ-

ments and evaluation of resource provisioning algo-

rithms. Software: Practice and experience, 41(1):23–

50.

FAREH, M. E.-k. (2015). Une approche bas

ee agents pour

l’allocation des ressources dans le Cloud Computing.

PhD thesis, Universit

e Mohamed Khider-Biskra.

Kumar, D., Mehrotra, D., and Bansal, R. (2018). Meta-

heuristic policies for discovery task programming

matters in cloud computing. In 2018 4th Interna-

tional Conference on Computing Communication and

Automation (ICCCA), pages 1–5. IEEE.

Mahavidyalaya, K. and Nadu, T. (2021). A novel multi-

agent approach to control service level agreement

violations in cloud computing. Turkish Journal of

Computer and Mathematics Education, 12(12):1432–

1438.

ICAART 2024 - 16th International Conference on Agents and Artiﬁcial Intelligence

772

Qasim, A., Amin, H., Aziz, Z., and Khalid, A. (2020). Efﬁ-

cient performative actions for e-commerce agents. Ap-

plied Computer Systems, 25(1).

Rabaoui, S., Hachicha, H., and Zagrouba, E. (2021).

Multi-agent-based framework for resource allocation

in cloud computing. In Agents and Multi-Agent Sys-

tems: Technologies and Applications 2021: Proceed-

ings of 15th KES International Conference, KES-

AMSTA 2021, June 2021, pages 427–437. Springer.

Singh, A., Juneja, D., and Malhotra, M. (2017). A

novel agent based autonomous and service compo-

sition framework for cost optimization of resource

provisioning in cloud computing. Journal of King

Saud University-Computer and Information Sciences,

29(1):19–28.

Singh, A. and Malhotra, M. (2015). Agent based resource

allocation mechanism focusing cost optimization in

cloud computing. International Journal of Cloud Ap-

plications and Computing (IJCAC), 5(3):53–61.

Soltane, M., Cardinale, Y., Angarita, R., Rosse, P., Rukoz,

M., Makhlouf, D., and Okba, K. (2018). A self-

adaptive agent-based system for cloud platforms. In

2018 3rd International Conference on Pattern Analy-

sis and Intelligent Systems (PAIS), pages 1–8. IEEE.

Venkateshwaran, K., Malviya, A., Dikshit, U., and Venkate-

san, S. P. (2015). Security framework for agent-based

cloud computing. IJIMAI, 3(3):35–42.

Wang, W., Jiang, Y., and Wu, W. (2016). Multiagent-based

resource allocation for energy minimization in cloud

computing systems. IEEE Transactions on Systems,

Man, and Cybernetics: Systems, 47(2):205–220.

Yahya, Z. B., Ktata, F. B., and Ghedira, K. (2020). Cloud

ma-morbac: A cloud distributed access control model

based on mobile agents. Int. J. Web Appl., 12(1):1–15.

Mobile Agents-Based Framework for Dynamic Resource Allocation in Cloud Computing

773