See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337790853

Task Scheduling based on Modified Grey Wolf Optimizer in Cloud Computing

Environment

Conference Paper · December 2019

DOI: 10.1109/ICTCS.2019.8923071

CITATIONS READS
9 140

3 authors:

Abdullah Issa Alzaqebah Rizik M. H, Al-Sayyed

The World Islamic Science and Education University University of Jordan
28 PUBLICATIONS 177 CITATIONS 70 PUBLICATIONS 337 CITATIONS

SEE PROFILE SEE PROFILE

Raja M.T Masadeh

The World Islamic Science and Education University
31 PUBLICATIONS 362 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Grey Wolf Optimization Applied to the 0/1 Knapsack Problem View project

A Survey on Automatic Web Extraction Information: An Efficiency and Accuracy Issues of Crawling Technique View project

All content following this page was uploaded by Raja M.T Masadeh on 04 August 2020.

The user has requested enhancement of the downloaded file.

 Task Scheduling based on Modified Grey Wolf
Optimizer in Cloud Computing Environment
Abdullah Alzaqebah Rizik Al-Sayyed Raja Masadeh
Computer Science Department, Information Technology Department, Computer Science Department,
The World Islamic Sciences and King Abdullah II School for The World Islamic Sciences and
Education University Information Technology, Education University
Amman, Jordan University of Jordan Amman, Jordan
Abdullah.zaqebah@wise.edu.jo Amman, Jordan raja.masadeh@wise.edu.jo
r.alsayyed@ju.edu.jo

Abstract—Task scheduling is considered as one of the to maximize resource utilization, minimize both makespan
most critical problems in cloud computing environment. The and cost to optimize the scheduling in cloud environments.
main target of task scheduling includes scheduling jobs on
virtual machines as well as improves performance. This Cloud task scheduling is known as an NP-complete
study employed Grey Wolf Optimization (GWO) algorithm problem [13]. More precisely, the required time for
with modifications on the fitness function by making it detecting the solution changes by the problem size [14].
handles multi-objectives in single fitness; the makespan and Cloud task scheduling is categorized into two classes
cost are the objectives included in the fitness in order to solve namely; meta-heuristic and heuristic algorithms. Heuristics
task scheduling problem. The main target of this technique is algorithms problem-specific strategy; it cannot be used to
to reduce both cost and makespan. CloudSim tool is used to answer open problems. On the other hand, the meta-
evaluate the objectives of the proposed method. The heuristics algorithm can be used (or applied) to solve a
simulation results showed that the proposed method wide range of problems in reasonable time.
(Modified Grey Wolf Optimizer - MGWO) has better
performance than both the traditional Grey Wolf Recently; Meta-heuristic algorithms are the most
Optimization Algorithm (GWO) and Whale Optimization applied techniques for task scheduling because they find
Algorithm (WOA) with makespan based fitness in terms of the optimal solutions or near-optimal solutions in
makespan, cost and degree of imbalance. reasonable time. Moreover, they detect the solutions by
employing the random choices. The most suitable example
Keywords—GWO, MGWO, WOA, Fitness, Makespan, and of a meta-heuristic algorithm is a Genetic Algorithm (GA)
Cost which is adopted by many studies to solve task scheduling
problem (TSP) in several manners. In literature studies
I. INTRODUCTION [15-18], the required time for mapping tasks into resources
Due to the availability of big data as well as the on- is increased when the number of jobs is increased.
demand operation in cloud computing (CC), the
requirements of CC environments have increased in recent In this research, we proposed cloud task scheduling
years. CC [1, 2] allows the clients to access the available which is based on the multi-objective model and Grey
and suitable resources such as internet applications, Wolf Optimization (GWO) algorithm to minimize both
storages, and servers [3]. The main role of the cloud cost and makespan in the cloud environments. CloudSim
service provider is to handle and manage client requests tool is used to evaluate the proposed technique.
(services) over the Internet [4]. The CC environment The organization of the paper is described as follows:
presents various services to clients. The most important Section II contains the related work, while section III
services are Platform as a Service (PaaS) [5], Infrastructure describes the GWO algorithm in details. Section IV
as a Service (IaaS) [6], Expert as a Service (ExaaS) [7] and outlines the suggested work. Simulation results are
Software as a Service (SaaS) [8][9]. The cloud clients have presented in section V. Finally, section VI concludes this
various tasks, and these tasks are implemented and research.
achieved at the same time by the available resources in the
cloud. The performance of CC can be developed by II. RELATED WORK
mapping tasks into resources in an optimized manner. One Many researchers tried to solve cloud task scheduling
of the most critical operations of the cloud is task using different techniques. Most of them employed meta-
scheduling which generates great influence on the entire heuristic algorithms such as GA, ACO, GWO, and WOA
cloud by impacting the Quality of Service (QoS) [10, 11]. in order to solve one of the main problems of cloud
The CC task scheduling preserves the balance over the environment which is task scheduling problem (TSP) as
entire system load. Each job demands response time, well as to find the optimal distribution of available
memory and computing time in several scales. In resources. However, there are still some issues in this
additions; the CC has the distributed resources. research area [2,19].
The efficient task scheduling method must minimize A novel algorithm is proposed which is based on neural
the makespan of the application [12]. Therefore, there is a network (NN) in order to classify the tasks queues which
need for algorithms to schedule the cloud tasks of users occur on any resource as well as to grant priorities to a
which optimally assign tasks into resources as well as variety of tasks [20]. NN is considered as an artificial
reduce the makespan. However, there are other criteria intelligence system which can discover and distinguish a
playing role in cloud task scheduling such as cost and pattern. Also, it can learn by instance and adapt to novel
utilization. Multi-objectives task scheduling algorithm has
978-1-7281-2882-5/19/$31.00 ©2019 IEEE
concepts and knowledge. Employing NN will be high the proposed technique can greatly minimize the total
potential to optimize mapping of tasks into virtual execution time to find the available cloud resources as well
machines (VMs) in CC environments. as significantly develop efficiency.
Few researchers employed GWO algorithm to solve the Some studies employed a Genetic Algorithm (GA) to
problem. Multi-Objectives cloud independent task propose novel cloud scheduling techniques. A new
scheduling based on mean GWO is presented [21]. The scheduling strategy and assists in appropriate and dynamic
primary objectives of the proposed algorithm [21] are to resource utilization are proposed in Kumar, P. et al. work
reduce both makespan and power consumption. Based on [30]. In other words, an improved GA is introduced which
simulation results, they proved that the suggested Mean of combined Min-Min and Max-min techniques in traditional
Grey Wolf Optimization algorithm has better results than GA. Based on simulation results, the proposed strategy
other traditional GWO and PSO algorithms. While [22] outperformed the traditional GA in terms of makespan.
employed the GWO method in order to solve dependent Suggested enhancement of GA is introduced by Wang, T.
tasks in CC environments. Makespan, cost, and resource et al. [31] which achieved independent task scheduling
utilization are taken into consideration. The experimental with minimizing makespan and balancing the entire system
results showed that the proposed algorithm has better load. The experimental results proved that the suggested
performance than the other existing techniques. algorithm can reduce the makespan and balance the system
load efficiently.
Some studies used Whale Optimization Algorithm
(WOA) to solve TSP. The study of sharma, m. et al. [23] III. GREY WOLF OPTIMIZATION (GWO) ALGORITHM
focused on both minimizing energy consumption and
makespan for cloud independent task scheduling. Grey Wolf Optimization (GWO) algorithm is
Experiments are performed over a variable number of tasks considered one of the most recently nature-inspired meta-
and VMs. Based on simulation results, the suggested heuristic optimization algorithm that is proposed by [32].
technique provided superior results than Min-min Moreover, it mimics the foraging and hunting behavior of
algorithm in terms of makespan and consumed energy. grey wolves. The most distinguished of grey wolves is
Another cloud task scheduling technique is suggested their social hierarchy; where they live in a pack that
based on WOA and multi-objective model that is called consists of 5-12 wolves. Each pack has alpha, beta, delta,
W-Scheduler [24]. The main objectives of W-Scheduler and omega members. Alpha is represented as a leader
are reducing makespan and budget cost. In addition, the which is responsible for take the decisions. Beta is a
simulation results of W-Scheduler are outperformed other consultant to the leader (alpha) which helps alpha to make
existing compared algorithms. Another multi-objectives decisions. Delta wolves are described as subordinate that
WOA is proposed in study of Reddy, G. N et al. [25] in submits to the upper levels (alpha and beta) but they
order to schedule independent tasks in CC environments. dominate the lower level which is called omega.
Energy consumption, makespan, resource utilization and Hunting behavior of grey wolves is split into stages as
quality of services are taken into accounts. Simulation follow [32-37]:
results proved that the suggested algorithm has better
performance compared with the existing techniques. • Tracking, chasing and approaching prey.
Masadeh, R. et al. [26] proposed a new metaheuristic • Pursuing, encircling and harassing the prey
optimization algorithm which is called Vocalization until it stops moving.
behavior of humpback Whale Optimization Algorithm
(VWOA). VWOA mimics the vocalization behavior of • Attack towards the prey.
humpback whales in nature. Also, the researchers
introduced cloud task scheduling technique which is based The mathematical model of the GWO algorithm is
on the VWOA and multi-objective model that is focused provided as follows:
on makespan, cost, resource utilization, and energy 1- Encircling prey: during the hunt phase, the grey
consumption. The simulation results showed that the wolves encircle the prey which is mathematically
proposed technique has better performance than other modeled as following equations Eq.1 and Eq.2:
algorithms.
Many researchers utilized Ant Colony Optimization
(ACO) to solve TSP in CC environment. Cloud task =| . ( )− ( ) (1)
scheduling algorithm is proposed based on load balancing
and ACO algorithm (LBACO) [27]. This algorithm ( + 1) = ( )− . (2)
balanced the entire system load, in turn, minimizing
makespan. Simulation results showed that the results of the
suggested strategy are provided superior results than First- Where t indicates to the current iteration, → and → are
Come-First-Served (FCFS) and traditional ACO coefficient vectors while is denoted as the position
algorithms. Another solution is proposed in study of
Tawfeek, M. A. et al. [28] that take into consideration the vector of the prey and → represents the position vector of a
makespan and degree of imbalance. Moreover, the grey wolf. In addition, → and → are computed using the
experimental results demonstrated that the suggested
following Eq. 3 and Eq. 4.
strategy outperformed Round-Robin (RR) and FCFS
techniques. Dependent tasks scheduling based on ACO and
two-way ants strategies is introduced in the work of Zhou,
Y. et al. [29]. The experimental results demonstrated that =2 . − (3)
 = 2. (4) the broker is to optimize some needed parameters such as
makespan, Cost, resource utilization and energy
Where r1 and r2 represent random vectors in [0, 1] and consumption by assigning the tasks to VMs to satisfy the
→ is linearly decreased from 2 to 0. [32] optimization function.
2- Hunting: This phase is guided by the leader alpha The scheduling process is based on some parameters;
and the consultant's beta and delta wolves which the scheduler needs information about the resources during
have enough knowledge about the position of prey. the tasks execution process. The Resource Information
Thus, the rest of the wolves should update their Server (RIS) is responsible about feeding the scheduler
locations according to the location of the best about theses information by summarizing the data center
agent that is mathematically modeled as following information such as CPU, memories and all other
equations Eq.(5,6 and 7): information about the contained VMs. On the other hand,
the scheduler assigns the tasks to the resources based on
this information with respect to optimize the given
= . − , = . − , = . − (5) parameters [38].

= − . , = − . , = − . (6) In this research, GWO is employed as the scheduler

engine of the cloud tasks according to their optimization
( + 1) = ( + + )/3 (7)
behaviour with respect of consumed time and it is recently
proposed by Mirjalili (2014). GWO scheduler algorithm
starts with random individuals (solutions) then evaluate
3- Exploitation and exploration (Attacking prey and these solutions according to the fitness values and update
search for prey): The prey that chased and attacked the search agents' positions in order to create other
by wolves is considered the ability of wolves solutions. Moreover, according to the evaluation process
catching the prey. More precisely, the ability of (Fitness Function), the algorithm creates near-optimal or
wolves can lead to global optima; which is the optimal solution by keeping the best fitness value
ability of exploitation. Since the value of A plays a solutions. In this research; the modification is make the
significant role; in case |A|<1, the grey wolves are fitness function contains multi-objectives instead of single
obliged to assault the prey. In case |A|>1, the grey one, the objectives cost and makespan are the used
wolves are forced to go away from the prey and objectives inside the fitness function in GWO in order to
looking for another one. Algorithm 1 shows the evaluate each solution; Thus, MGWO based on Multi-
pseudocode of the GWO algorithm. [32] Objectives function and Grey Wolf Optimization
Algorithm. for that we call it MGWO.
Algorithm 1: Pseudocode of GWO algorithm
A. Performance Metrics
Begin
1- Makespan: is the overall execution time that
1. Initialize Population needed to accomplish the tasks in the CC
2. Initialize a, A and C environment. The minimum value of makespan
3. Calculate the fitness of each search agent) means better efficiency in CC and this is done by
4. X = the best search agent making the scheduler assigns tasks to prior VM
5. X = the second-best search agent according to the task's information and the RIS
6. information. Assume the ET is the execution time
X = the third best search agent
7. of (tn) task on (Vmm) VM as following, {t1, t2,
While (t< maximum number of iteration) …… tn} are tasks, {Vm1, Vm2, …….Vmm} are
8. For each search agent VMs and the execution time is {ET1, ET2, …..
9. Update the position of the current search agent ETn}, Eq.8shows the makespan fitness function
by equation (7). [26].
10. End for
11. Update a, A and C
12. Calculate the fitness of the current search agent. = { } (8)
13. Update Best Solution.
14. Update X_α, X_β and X_δ
15. t= t+1 2- Cost: is the execution cost of execution task on a
16. End While specific VM, this cost relies on the length of the
17. Return X task (TaskSize), the cost of transfer task to the
specific VM and the storage of that VM. Eq. 9
End shows the cost equation and Eq. 10 illustrates the
fitness of cost metric [26, 29].
IV. PROPOSED WORK
Task manager which is called cloud broker is key = (9)
responsible for collecting and controlling the tasks
submitted by cloud users. More precisely, the management = { …… } (10)
process is distributing the incoming tasks to the available
resources (VMs) in the cloud datacenter. The main aim of
 3- Evaluation of Fitness Function: in this paper, two
performance metrics makespan and cost are
included in the fitness function of the MGWO
scheduler which aims to minimize the fitness value
and this is the modification of traditional use of
GWO. The fitness equation presented in Eq. 11
where ti represents the i'th task from the tasks list.

=( + ) (11)

V. SIMULATION RESULTS
Fig.2: Makespan of various numbers of tasks when the number of VMs
The proposed algorithm is simulated using CloudSim is 2
tool; where its platform based on Java. All these
experiments are validated on a personal computer with
Intel Core i-7 processor, 16 GB RAM, and Windows 8.1
operating system. The proposed MGWO differs from
GWO by modifying the core fitness function by make it
considering multi-objectives instead of just a single
objective which is the makespan. The outcomes of
employed modified GWO (MGWO) are compared with
the original GWO and existing WOA technique since the
WOA is a recently proposed optimizer by Mirjalili (2016)
with various numbers of independent tasks (200, 400, 600,
800 and 1000) and different numbers of VMs (1,2, 4 and
8); in terms of makespan, cost and degree of imbalance.
Fig.3: Makespan of various numbers of tasks when the number of VMs
The simulation results showed minimum cost and total is 4.
execution time compared with other selected algorithms. In
this simulation, each scenario is executed 10 times and
then the average is calculated and taken into consideration.
The average makespan for executed tasks using MGWO,
GWO and WOA is illustrated in Fig.1 – Fig.4. It is obvious
that MGWO has better performance than existing WOA
and traditional GWO in terms of makespan because of
using cost with makespan in the fitness function to
evaluate the solutions which make better scheduling
process which directly effect on the overall makespan. In
addition, when the number of VM equals one, all
algorithms form same results in both makespan and cost
since there are no other resources to schedule tasks into it.
The cost which represents the execution cost of running Fig.4: Makespan of various numbers of tasks when the number of VMs
an independent task on a particular VM. Moreover, it is 8
depends on the task's length, VM's storage and cost of
transmitting task to a particular VM, Moreover, due to the
simulation settings are almost same so as clearly shown in
the results there is no significant difference in term of cost
metric. Fig.5- Fig.9 showed the cost of a different number
of executing tasks on various numbers of VMs.

Fig.5: Scheduling cost of a various number of tasks when the number of

VMs is 1.

Fig.1: Makespan of various numbers of tasks when the number of VMs

is 1
 Fig.9: Degree of Imbalance of GWO, MGWO, and WOA on 8 VM
Fig.6: Scheduling cost of a various number of tasks when the number of
VMs is 2.
VI. CONCLUSION
Various meta-heuristic algorithms are employed in
order to develop task scheduling methods for CC
environment. In this work, a new task scheduling based on
GWO (MGWO) is introduced by modifying the fitness
function and make multi-objective in single fitness instead
of using the single makespan objective. The major target of
independent task scheduling based on both cost and
makespan is executed in the CloudSim. The performance
of the proposed technique is compared with traditional
GWO and WOA. The simulation results provided good
outcomes in reducing makespan, cost, and degree of
imbalance.

Fig.7: Scheduling cost of a various number of tasks when the number of

VMs is 4. REFERENCES

[1] Mell, P., & Grance, T. (2011). The NIST definition of cloud
computing..
[2] JoSEP, A. D., KAtz, R., KonWinSKi, A., Gunho, L. E. E.,
PAttERSon, D., & RABKin, A. (2010). A view of cloud
computing. Communications of the ACM, 53(4).
[3] He, H., Xu, G., Pang, S., & Zhao, Z. (2016). AMTS: Adaptive
multi-objective task scheduling strategy in cloud computing. China
Communications, 13(4), 162-171.
[4] Lin, X., Wang, Y., Xie, Q., & Pedram, M. (2014). Task scheduling
with dynamic voltage and frequency scaling for energy
minimization in the mobile cloud computing environment. IEEE
Transactions on Services Computing, 8(2), 175-186.
[5] Navimipour, N. J., Rahmani, A. M., Navin, A. H., &
Hosseinzadeh, M. (2015). Expert Cloud: A Cloud-based
framework to share the knowledge and skills of human resources.
Fig.8: Scheduling cost of a various number of tasks when the number of Computers in Human Behavior, 46, 57-74.
VMs is 8. [6] Malawski, M., Juve, G., Deelman, E., & Nabrzyski, J. (2015).
Algorithms for cost-and deadline-constrained provisioning for
While the Degree of Imbalance (DI) is measured the scientific workflow ensembles in IaaS clouds. Future Generation
Computer Systems, 48, 1-18.
imbalance among VMs [36] using the following Eq. 5:
[7] Navimipour, N. J. (2015). A formal approach for the specification
and verification of a trustworthy human resource discovery
( )= (5) mechanism in the expert cloud. Expert Systems with Applications,
42(15-16), 6112-6131.
[8] Keshanchi, B., Souri, A., & Navimipour, N. J. (2017). An
Where Tmax represents the maximum execution time of improved genetic algorithm for task scheduling in the cloud
VMs, Tmin and Taverage are denoted the minimum and environments using the priority queues: formal verification,
average execution time, respectively. Fig.9 illustrated DI's simulation, and statistical testing. Journal of Systems and Software,
experiments which are performed for a different number of 124, 1-21.
independent tasks on 8 VMs. Moreover, it clearly showed [9] Alkhanak, E. N., Lee, S. P., & Khan, S. U. R. (2015). Cost-aware
that the MGWO achieved lowest level of degree of challenges for workflow scheduling approaches in cloud
computing environments: Taxonomy and opportunities. Future
imbalanced which means better scheduling balancing. Generation Computer Systems, 50, 3-21.
 [10] Rimal, B. P., Jukan, A., Katsaros, D., & Goeleven, Y. (2011). [30] Kumar, P., & Verma, A. (2012). Independent Task Scheduling in
Architectural requirements for cloud computing systems: an Cloud Computing by Improved Genetic Algorithm. International
enterprise cloud approach. Journal of Grid Computing, 9(1), 3-26. Journal, 2(5).
[11] Rimal, B. P., Choi, E., & Lumb, I. (2009, August). A taxonomy [31] Wang, T., Liu, Z., Chen, Y., Xu, Y., & Dai, X. (2014, August).
and survey of cloud computing systems. In 2009 Fifth International Load balancing task scheduling based on genetic algorithm in
Joint Conference on INC, IMS and IDC (pp. 44-51). Ieee. cloud computing. In 2014 IEEE 12th International Conference on
[12] Navin, A. H., Navimipour, N. J., Rahmani, A. M., & Dependable, Autonomic and Secure Computing (pp. 146-152).
Hosseinzadeh, M. (2014). Expert grid: new type of grid to manage IEEE.
the human resources and study the effectiveness of its task [32] Mirjalili, S., Mirjalili, S. M., & Lewis, A. (2014). Grey wolf
scheduler. Arabian Journal for Science and Engineering, 39(8), optimizer. Advances in engineering software, 69, 46-61.
6175-6188. [33] Masadeh, R., Alzaqebah, A., Hudaib, A., & Rahman, A. A. (2018).
[13] Ullman, J. D. (1975). NP-complete scheduling problems. Journal Grey Wolf Algorithm for Requirements Prioritization. Modern
of Computer and System sciences, 10(3), 384-393. Applied Science, 12(2), 54.
[14] Xu, Y., Li, K., He, L., & Truong, T. K. (2013). A DAG scheduling [34] Masadeh, R., Hudaib, A., & Alzaqebah, A. (2018). WGW: A
scheme on heterogeneous computing systems using double hybrid approach based on whale and grey wolf optimization
molecular structure-based chemical reaction optimization. Journal algorithms for requirements prioritization. Advances in Systems
of Parallel and Distributed Computing, 73(9), 1306-1322. Science and Applications, 18(2), 63-83.
[15] Singh, S., & Kalra, M. (2014, November). Scheduling of [35] Masadeh, R., Sharieh, A., & Sliet, A. (2017). Grey wolf
independent tasks in cloud computing using modified genetic optimization applied to the maximum flow problem. International
algorithm. In 2014 International Conference on Computational Journal of Advanced and Applied Sciences, 4(7), 95-100.
Intelligence and Communication Networks (pp. 565-569). IEEE. [36] Yassien, E., Masadeh, R., Alzaqebah, A., & Shaheen, A. (2017).
[16] Kaur, K., & Kaur, A. (2015). Optimal Scheduling and Load Grey wolf optimization applied to the 0/1 knapsack problem.
Balancing in Cloud using Enhanced Genetic Algorithm. International Journal of Computer Applications, 169(5), 11-15.
International Journal of Computer Applications, 125(11). [37] Alzaqebah, A., & Abu-Shareha, A. A. (2019). Ant Colony System
[17] Wang, T., Liu, Z., Chen, Y., Xu, Y., & Dai, X. (2014, August). Algorithm with Dynamic Pheromone Updating for 0/1 Knapsack
Load balancing task scheduling based on genetic algorithm in Problem. International Journal of Intelligent Systems and
cloud computing. In 2014 IEEE 12th International Conference on Applications, 11(2), 9.
Dependable, Autonomic and Secure Computing (pp. 146-152). [38] Menascé, D. A., Saha, D., Porto, S. C. D., Almeida, V. A., &
IEEE. Tripathi, S. K. (1995). Static and dynamic processor scheduling
[18] Lakshmi, R. D., & Srinivasu, N. (2016). A dynamic approach to disciplines in heterogeneous parallel architectures. Journal of
task scheduling in cloud computing using genetic algorithm. Parallel and Distributed Computing, 28(1), 1-18.
Journal of Theoretical & Applied Information Technology, 85(2).
[19] Singh, P., Dutta, M., & Aggarwal, N. (2017). A review of task
scheduling based on meta-heuristics approach in cloud computing.
Knowledge and Information Systems, 52(1), 1-51.
[20] Maqableh, M., & Karajeh, H. (2014). Job scheduling for cloud
computing using neural networks. Communications and Network,
6(3), 191-200.
[21] Natesan, G., & Chokkalingam, A. (2018). Task scheduling in
heterogeneous cloud environment using mean grey wolf
optimization algorithm. ICT Express, 1-5.
[22] Khalili, A., & Babamir, S. M. (2017). Optimal scheduling
workflows in cloud computing environment using Pareto‐based
Grey Wolf Optimizer. Concurrency and Computation: Practice and
Experience, 29(11), 1-11.
[23] Sharma, M., & Garg, R. (2017, December). Energy-aware whale-
optmized task scheduler in cloud computing. In 2017 International
Conference on Intelligent Sustainable Systems (ICISS) (pp. 121-
126). IEEE.
[24] Sreenu, K., & Sreelatha, M. (2017). W-Scheduler: whale
optimization for task scheduling in cloud computing. Cluster
Computing, 1-12.
[25] Reddy, G. N., & Kumar, S. P. (2017, October). Multi objective task
scheduling algorithm for cloud computing using whale
optimization technique. In International Conference on Next
Generation Computing Technologies (pp. 286-297). Springer,
Singapore.
[26] Masadeh, R., Sharieh, A., & Mahafzah, B. A. Humpback Whale
Optimization Algorithm Based on Vocal Behavior for Task
Scheduling in Cloud Computing.
[27] Li, K., Xu, G., Zhao, G., Dong, Y., & Wang, D. (2011, August).
Cloud task scheduling based on load balancing ant colony
optimization. In 2011 Sixth Annual ChinaGrid Conference (pp. 3-
9). IEEE.
[28] Tawfeek, M. A., El-Sisi, A., Keshk, A. E., & Torkey, F. A. (2013,
November). Cloud task scheduling based on ant colony
optimization. In 2013 8th international conference on computer
engineering & systems (ICCES) (pp. 64-69). IEEE.
[29] Zhou, Y., & Huang, X. (2013, November). Scheduling workflow in
cloud computing based on ant colony optimization algorithm. In
2013 Sixth International Conference On Business Intelligence And
Financial Engineering (pp. 57-61). IEEE

View publication stats