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

net/publication/358635809

An Analysis of a Meta-Heuristic Optimization Algorithms for Cloud

Computing

Conference Paper · February 2022

DOI: 10.1109/ISCON52037.2021.9702376

CITATIONS READS

5 308

2 authors:

Pillareddy Vamsheedhar Reddy K. Ganesh Reddy

VIT-AP University VIT-AP University
8 PUBLICATIONS 20 CITATIONS 76 PUBLICATIONS 318 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Pillareddy Vamsheedhar Reddy on 01 March 2022.

The user has requested enhancement of the downloaded file.

 2021 5th International Conference on Information Systems and Computer Networks (ISCON)
GLA University, Mathura, India. Oct 22-23, 2021

An Analysis of a Meta Heuristic Optimization Algorithms for

Cloud Computing
Pillareddy vamsheedhar Reddy Karri Ganesh Reddy
School of Computer science and engineering, School of Computer science and engineering,
VIT-AP University, VIT-AP University,
Amaravathi, India. Amaravathi, India.
[email protected] [email protected]
2021 5th International Conference on Information Systems and Computer Networks (ISCON) | 978-1-6654-0341-2/21/$31.00 ©2021 IEEE | DOI: 10.1109/ISCON52037.2021.9702376

Abstract— Cloud computing enables users to access pooled each other. According to Task Scheduling, virtualized
services via the internet on demand through a cloud platform resources are assigned to a specific task for a particular
that is extremely flexible and provides consumers with a pay- period of time. Scheduling has the goal of maximizing the
per-use model with variable SLA. The majority of cloud use of cloud resources. In cloud computing, task scheduling
services operate on virtual machines, and in this scenario, job is a major challenge. Due to the dynamic nature of cloud
scheduling is critical. Because of the restricted number of computing resources, traditional scheduling algorithms are
optimum parameters evaluated in task scheduling, current ineffective.
meta heuristic optimization methods in cloud computing are
insufficient for finding optimal solutions. Existing studies, such When scheduling in cloud computing, the following
as load balancing, energy cost per unit, and VM migration, metrics are most important.
lack dependability and availability factors, which are required
to create an effective task scheduling algorithm. Finally We did 1. Turnaround Time
a proportional analysis of prior algorithms and identified 2. Execution time
scope to research in the field of cloud scheduling algorithms. 3. Waiting Time
4. Cost
Keywords—Scheduling ,task scheduling,work flow
scheduling,VM migration,load balance,availability,reliability.

I. INTRODUCTION
A new and developing trend in the field of information
technology is known as cloud computing. On demand cloud
computing it is accessible via internet. The resources for
cloud computing are provided by using cloud services, that
are based on a pay per use model. Two entities are involved
in cloud computing: the client and the cloud provider. When
customers need resources, the cloud provider provides them.
Since the cloud computing environment demands more
efficient scheduling, this has become a priority. Customers
have a wide range of cloud computing services to choose
from. Amazon, Microsoft, and Google are the main cloud
providers in the market. As shown in Fig.1, its services are
classified into three types: software-as-a-service(SaaS),
platform-as-a-service(PaaS),and infrastructure-as-a-service
(IaaS).
Scheduling in Cloud Computing:
Scheduling is nothing more than assigning a task or job
to a computer. Resource management in the cloud relies The following paper is organized in a structured manner
heavily on scheduling. Scheduling is responsible for resource and discuss as follows. In Section II it gives a clear
sharing on a number of fronts. Several virtual machines can information about all literature works done, In Section III it
share a single physical server, and each virtual machine can provides research directions for upcoming researchers, and
run multiple applications, each of which may have multiple followed with conclusions and future scope.
jobs or tasks. Algorithms for scheduling must be efficient,
fair, and free of stifling. To increase throughput while II. LITERATURE SURVEY
reducing turnaround time, the scheduler's goal must be
We included in this section measures such as CPU
achieved. Programming is a polynomial solution to an NP-
utilization, waiting time, reaction time and memory
Hard Problem. In order to find optimal solutions, task utilization. All available parameters of algorithms were
scheduling employs heuristic and meta-heuristic techniques. studied in this section.
It is possible to schedule tasks in two different ways:
dependently and independently. Job that is dependent on
others In independent task scheduling, jobs are performed A. A Multiple Population Ant colonyApproach[1]:
without regard to order or sequence, whereas scheduling is A multi-objective ant-colony system is built on a multi-
nothing more than a collection of tasks that are dependent on population co-evolution, with two colonies addressing the

978-1-6654-0341-2/21/$31.00 ©2021 IEEE 1

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
two objectives. Furthermore, the method includes the three F. GWO in Cloud computing enviornment[6]:
approaches listed below to effectively deal with multi- The authors adapted a new novel GWO to address the
objective challenges: (1) a rule on pheromone update based task scheduling problems in the cloud computing
on a set of non-dominated archive solutions to guide each environment. The grey wolf optimizer is a nature-inspired-
colony in sufficient quest for its optimization goal; (2) an algorithm. The goal of this scheduling algorithm is to assign
additional strategic heuristic approach to prevent a colony an N number of tasks to N virtual machines in order to
from focusing only on its corresponding single optimization reduce make span, resource costs, and energy consumption.
goal by working with an update pheromone rule to balance
the quest for both targets; and (3) elite updating
G. MGWO in Cloud Computing Environment[7]:
Experimental data show that the strategy exceeds several
advanced multi-target optimization techniques as well as In this study, the authors modified the fitness function to
constrained optimization approaches. include multiple objectives rather than a single goal in order
to examine each solution. The aim of the modified grey wolf
B. Task Scheduling for meta heuristic optimization[2]: optimizer (MGWO) is to minimize the fitness function value.
It came up with two algorithms, MHO-S (meta heuristic
H. Cloud based application non live task migration
optimization static) and MHO-D (meta heuristic optimization
approach [8]:
Dynamic). The algorithm handles tasks by taking their
properties into account. In particular, this work focuses on the migration of
independent tasks using the TM-eFCFS and task migration
The MHO algorithm is capable of handling both static methods by assigning the task to each VM which executes its
and dynamic property-based tasks. The MHO quickly sets tasks faster and is able to pre-emotively take the next partly
VMs for static properties to perform simpler jobs, increasing executed task from the slower VM in a way that improves
the ability of the algorithm to handle dynamic attributes, time and use of resources.
therefore reducing memory, response time and execution by
using fewer VM's. I. Comparison of cloud environment using Cloud report
[9]:
C. Multi objective task scheduling algorithm[3]:
Unlike the other models, this algorithm focuses on
Two task scheduling algorithms for heterogeneous reducing resource utilization. The authors implemented three
systems were proposed in this algorithm:1) TBTS offline task scheduling algorithms, namely, space shared, time
algorithm, 2) SLA-LB online scheduling algorithm. The shared and dynamic workload. In Space Shared the tasks are
TBTS scheduling algorithm is a two-phase scheduling arranged in a queue based with respected to arrival time.
technique. To increase cloud efficiency and reduce make Based upon the arrival time the first task is assigned to the
span value, pre-processing processes are designed before VM. After the first task is completed, the scheduler moves
distributing tasks to virtual machines (VMs). In this case, the on to the next task in the sequence. In time sharing each task
tasks that are received are not automatically distributed to the is assigned with a particular amount of time ,is called
cloud. It is a single phase scheduling algorithm that quantum. The task which arrives first moves into the ready
dynamically arranges work depending on the needs of queue state and it is allocated to the VM. When the time
customers. SLS-LB is a single phase scheduling algorithm quantum expires, then the new task is assigned to the VM
that schedules tasks based on the needs of clients. which is in the ready queue and the previous task is placed at
the end of the ready queue like as a round robin. The tasks
D. Loin Optimization Algorithm for Cloud computing which are currently inactive VMs then the dynamic workload
enviornment [4] are equally distributed among those VMs to provide proper
The authors proposed a new task scheduling algorithm, resource utilization. Compared to above discussion it is
LOA (loin optimization algorithm), which is based on a discovered that time-shared scheduling consumes less
lion's behavior. The goal of this algorithm is to reduce the resource utilization than the other two models.
life span by randomly assigning VMs to jobs or tasks, as well
as to reduce costs by migrating tasks from one virtual J. New work flow whale optimizer in cloud
machine to another. In this case, we can improve the environment[10]:
algorithm's performance by reducing its resource usage. Using spiral action of a bubble net feeding device, WOA
is the greatest search optimizer for hunting beasts. The
E. Enhanced Load balancing meta heuristic [5]: implementation of the whale optimizer enhances workflow
The firefly algorithm inspired this approach, which has planning limitations and balances the load between various
been divided into three sets: i). Generation Population, resources in a cloud environment. The multi-objective
ii).Scheduled index and iii) optimization of the scheduled list optimisation criteria encompassing the make-span, deadline
done using the firefly algorithm. Initially, a population is hit and resource utilization parameter are used to evaluate
generated using a cloud network. Then, SI is computed and cloud-orientated WOAs. Further, in comparison with other
eventually the scheduled list is optimized using FA.The existing methodologies, the WOA suggested has better
authors proposed cloud scheduling methods to balance the outcomes.
load using FA & PSO.The main aim of this algorithm is to
balance the load of the entire system as well as reduce the K. W-Scheduler for task scheduling in cloud
make span time. computing[11]:
The W-scheduler uses the multi-objective and whale
optimization algorithms, along with the budget cost function,

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
to compute the fitness value. The total budget costs are P. Improved PICEA-g workflow scheduling in cloud
estimated by summing the CPU and memory costs. When environment[16]:
working with W-scheduler, the assumption is made that the This algorithm, called ch-PICEA-g, was developed by
present solution is the best and the system then selects the authors as an effective heuristic algorithm for multi-objective
optimal option. The algorithms perform better than PBACO, co-evolution. Ch-PICEA-g utilizes two different chaotic
SLPSO-SA, and SPSO-SA. systems (logistic and tent maps) to generate chaotic values
for population composition, followed by genetic operators
L. Task scheduling using ACO[12] for evolution. Additionally, an improved fitness function was
This algorithm concentrates the load on the virtual applied to the original PICEA-g to improve its performance.
machines, which has an effect on their make span. To This algorithm was rigorously tested by several experiments.
improve load balancing capability, the authors defined a load The outcomes show that this algorithm outperforms the
balance factor to choose a higher load balancing virtual competing algorithms in regards to various performance
machine with a high probability. metrics.

M. hybrid cuckoo in task schedduling[13]. Q. workflow scheduling using fuzzy dominance sort
This algorithm combines the GA and the cuckoo HEFT[17]:
algorithms. This hybrid cuckoo algorithm focuses on these With regard to IaaS clouds, the authors suggested a
parameters: time, resource utilization, and energy workflow scheduling scheme in which the makespan and
consumption. The objective of this algorithm is proper cost of workflows are both optimally targeted. In this
utilization of resources, to reduce the waiting time, and scenario, a heterogeneous earliest finish time (HEFT) based
reduce the chances of failure of the system. fuzzy dominance sort (FDS) algorithm was implemented.
This approach closely integrates the list scheduling heuristic
N. Reducing Energy Consumption in Heterogeneous Cloud with the dominance sorting method. Real-world and
Computing [14]: synthetic workflow experiments demonstrated the system's
effectiveness. Hyper volume and run up to hundreds of times
Considered job prioritization and processor selection in
faster than the state-of-the-art algorithms were dramatically
perspective of time constraints. It showed to be superior in
better on the cost-make span tradeoff fronts.
terms of deadline accomplishment and job scalability, while
HEFT provided a shorter overall execution time. Many
Workflow applications are hosted on the cloud. R. MOWS heuristic algorithm in cloud [18]:
Authors designed a systematic approach that takes into
O. Distributed particle swarm optimization (DGLDPSO) consideration both security requirements and collaboration
[15]: requirements as it relates to the secure tasks' placement in the
cloud. The framework devised a method for task scheduling,
DGLDPSO is an effective optimized technique for large-
and used that as a starting point for developing a heuristic
scale optimization and it is extended to large-scale cloud
algorithm. Workflows, a widely used cloud simulation tool,
workflow scheduling. An Adaptive Renumber Strategy
was used to conduct extensive experiments to evaluate the
(ARS) was established when DGLDPSO was used to large-
approach's effectiveness. This work from real-world systems
scale cloud workflow scheduling, enabling solutions to relate
revealed that the approach improved the overall system
to resource characteristics and giving the search experience
security across a wide range of workload characteristics,
an added benefit. To further investigate DGLDPSO's
specifically with regard to quality of security and security
performance, a series of experiments were performed on the
risk.
large-scale benchmark functions and the cloud workflow
scheduling instances. According to the comparison results, Additionally, the findings proved that the algorithm used
DGLDPSO was more effective or comparable to other state- to recognize and respond to threats in the cloud environment
of-the-art large-scale optimization and workflow scheduling was efficient.
algorithms.

TABLE I. SUMMARY OF META HEURISTIC SCHEDULING ALGORITHMS FOR CLOUD COMPUTING

Study Presented work Objective Tool Used

MOACS adapts concepts of MPMO techniques
two colonies with some number of Ants. To guide each colony to search, to
A Multiple Population Ant colony Approach
Two colonies named as avoid focusing on one optimization Cloud Sim
[1]
1.Tcolony and to improve solution Quality
2.Ccolony
To provides low execution time
MHO [2] They came up with two algorithms MHOS and
with utilization of minimum no. of Cloud Sim
MHOD
virtual Machines
They proposed two algorithms TBTS offline,
To reduce Make span and increases
Multi objective task scheduling algorithm [3] SLA-LB online scheduling Algorithm. Cloud Sim
cloud utilization
To minimize make span and degree
Loin optimization in cloud computing
Proposed based on LOA of imbalance and also to produce Cloud Sim
environment[4]
high resource Utilization

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
 This approach inspired by FA , it is developed in
three sets
An improved load balanced meta heuristic To minimize make span and degree
1.Genreation population Cloud Sim
scheduling in cloud [5] of imbalance
2.Scheduled index
3.Optimization of scheduled list done using FA
For Solving task scheduling
They have meta heuristic adapted innovative
GWO in cloud computing Environment [6] problems in cloud computing Cloud Sim
GWO
environment
Proposed MGWO and compared with GWO and
MGWO in cloud computing environment [7] To reduce make span and cost Cloud Sim
WOA.
To reduce the amount of Make
Cloud based application non live task A scheduling algorithm which incorporates task
Span while optimizing the Cloud Sim
migration approach [8] migration using TM-eFCFS method.
consumption of resources
Implemented three scheduling schemes,
It discovers that Time shared
Comparison of cloud environment using cloud specifically time-shared-scheduling, space-
Scheduling require less resource Cloud Sim
report [9] shared-scheduling, and dynamic workload-
utilization than other models
scheduling.
New workflow whale optimizer in cloud WOA search Optimizer to hunt the prey by spiral To fulfill deadline and budget
Cloud Sim
environment[10] behavior of “bubble-net Mechanism” constrains such as QoS.

By referring to the table1, we've defined the various scheduling algorithms we examined in the literature survey, as well as
the computational tools used to test these algorithms in the cloud.

TABLE II. COMPARISON OF SCHEDULING ALGORITHM

Electricity price per unit

Avg resource utilization

Degree of imbalance

Cloud utilization
Execution time

Avg Migration
Response time
Make span

Gain
SLA

cost
Algorithm

MHO[2] ‫ݱ‬ ‫ݱ‬ ‫ݱ‬

- - - - - - -
Multi objective task scheduling algorithm ‫ݱ‬ ‫ݱ‬ ‫ݱ‬ ‫ݱ‬ ‫ݱ‬
- - - - -
[3]
Loin optimization in cloud Computing ‫ݱ‬ ‫ݱ‬ ‫ݱ‬ ‫ݱ‬
enviornment[4]
- - - - - -

An improved load balanced meta heuristic ‫ݱ‬ ‫ݱ‬

scheduling in cloud [5]
- - - - - - - -
GWO in cloud computing Environment [6] ‫ݱ‬
- - - - - - - - -
MGWO in cloud computing environment [7] ‫ݱ‬ ‫ݱ‬
- - - - - - - -
Cloud based application non live task
migration approach [8]
‫ݱ‬ ‫ݱ‬ - - - - - - - - -
Comparison of cloud environment using cloud
report [9]
- ‫ݱ‬ - - - - - - - - -
New workflow whale optimizer in cloud
environment[10]
‫ݱ‬ ‫ݱ‬ ‫ݱ‬ - - - - - - - -
A Multiple Population Ant colonyApproach
[1]
‫ݱ‬ - - - - - - - ‫ݱ‬ - -

From Table 2 reveals that the majority of algorithms

currently in use aim to minimize Make span and maximize III. RESEARCH DIRECTIONS
resource utilization. There are other important parameters Based on the analysis of several scheduling algorithms,
such as wait time, response time, and memory utilization, but we classified them into various types, such as resource based,
scheduling algorithms tend to focus on the importance of load balancing based, Cost based, and nature-inspired
only a few of these parameters. Even if algorithms improve, algorithm based.
there is still a need for better ones, which measure multiple
metrics.

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
 generation and distribution plants, to move to the cloud
computing model. Additional security risks are associated
with the services offered by third-party cloud service
providers. Security concerns are heightened when user assets
(data, applications, and so on) are moved outside of the
administrative control of a shared environment in which a
large number of users are collocated.

IV. CONCLUSION AND FUTURE WORK

A. Classification of Resource-based scheduling algorithm:
Cloud computing has rapidly become a hot research area
To reduce the make span and average resource utilization when it comes to scheduling. Among heterogeneous and
in resource-based scheduling algorithms, various techniques homogeneous cloud environments, Several scheduling
have been developed. Resource migration or VM migration algorithms exist. Most existing algorithms, based on our
are not currently considered to be optimal parameters. Server comparison study, focus on make span time and resource
changes are dependent on workload changes. In some cases, usage. There is no mention of a single parameter like VM
the number of users that request something is low, and on the migration time or Resource Migration or electricity price per
other hand, the number of people who desire something is unit, or a combination of parameters like SLA and security
exceptionally high (for example, at the end of the month). between client and provider.
Adding servers manually is almost impossible. So, virtual
machine migration is set up to cater to workloads that change The continued improvement of existing and future
without the user having to intervene. scheduling algorithms is essential to developing a better
cloud-scheduling algorithm. To assign tasks to cloud
B. Degree Balanced Base Scheduling Classification: resources in the future, we will use a multi-objective
scheduler algorithm in which we will use the price of
There are a large number of existing algorithms that are
electricity per unit cost and security as the main metrics for
proposed for degree balance based scheduling, each with
assigning tasks to cloud resources
a different type of metric, such as the make span, the
resource utilization, the response time, or the waiting
time. As the cloud is spread over the globe and electricity REFERENCES
costs per unit vary from place to place, the price of power per [1] Z.G. Chen et al., “Multiobjective cloud workflow scheduling: a
unit can be regarded as a parameter of degree-balance-based multiple populations ant colony system approach,” IEEE Transactions
programming methods. This metric can help cloud providers, on Cybernetics, vol. 49, no. 8, pp. 2912-2926, August 2019.
as the programmer can automatically select the location with [2] Ziyath, S. Peer Mohamed, and S. Senthilkumar, “MHO: meta
heuristic optimization applied task scheduling with load balancing
the minimum cost per unit of electricity, allowing for more technique for cloud infrastructure services,”Journal of Ambient
efficient supply of resources, if the lowest cost per unit of Intelligence and Humanized Computing, pp. 1-10. July 2020.
electricity is prioritized. [3] Lavanya, M., B. Shanthi, and S. Saravanan,“Multi objective task
scheduling algorithm based on SLA and processing time suitable for
C. Nature Inspired Based Classification: cloud environment,” Computer Communications, vol. 151,pp. 183-
195,February 2020.
Algorithms that are inspired by nature are used to solve [4] Almezeini, Nora, and Alaaeldin Hafez,“Task scheduling in cloud
real-world problems that humans face. Many existing nature- computing using lion optimization algorithm,” Internatonal Journal of
inspired algorithms use make span as a metric to find the best Advanced Computer Science and applications vol.8,no.11,November
optimal solutions, which is common in the algorithmic 2017.
literature. However, in this case, we must take into account [5] Aruna, M., D. Bhanu, and S. Karthik,“An improved load balanced
other metrics, such as service level agreements (SLA). It is metaheuristic scheduling in cloud,” Cluster Computing,september
2017
critical that client benefit from the resources and service
[6] Bacanin, Nebojsa, et al.,“Task Scheduling in Cloud Computing
levels provided by cloud service providers, while cloud Environment by Grey Wolf Optimizer,”2019 27th
service providers themselves must improve the precision Telecommunications Forum (TELFOR). IEEE,pp.1-4,November
with which they describe their services in order to optimize 2019.
their own infrastructure. In cloud computing, the SLA is [7] Alzaqebah,Abdullah, Rizik Al-Sayyed, and Raja Masadeh,“Task
extensively used. It is a contract between cloud providers and scheduling based on modified grey wolf optimizer in cloud
consumers to offer details on service levels. computing environment,” 2019 2nd international conference on new
trends in computing sciences (ICTCS),IEEE ,pp.1-6,October 2019.
[8] Panwar, Neelam, Sarita Negi, and Man Mohan Singh Rauthan,“Non-
D. Cost Base Scheduling Classification: live task migration approach for scheduling in cloud based
The metrics examined for these criteria in existing applications,” International Conference on Next Generation
algorithms include task duration, average run time and Computing Technologies. Springer, Singapore, vol.828,pp.124-
137,June 2018.
execution costs. Cloud computing has the ability to provide
[9] Jena, Soumya Ranjan, et al,“Comparison of different task scheduling
on-demand resources over the internet which are cost algorithms in cloud computing environment using cloud
effective, easily managed, elastic and powerful and have the reports,” Smart Intelligent Computing and Applications. Springer,
potential to be revolutionary. Because of its efficient and Singapore,vol.159, pp.33-42,september 2019.
shared utilization, cloud computing enhances the capabilities [10] Thennarasu, S.R., Selvam, M. & Srihari, K, “A new whale optimizer
of hardware resources. Organizations and individual users for workflow scheduling in cloud computing environment,” J
are encouraged to move their applications and services to the Ambient Intell Human Comput ,january 2020.
cloud as a result of the features discussed above. It is even [11] Sreenu, K, Sreelatha, M,“W-Scheduler: whale optimization for task
being implemented in critical infrastructure, such as power scheduling in cloud computing,” Cluster Comput ,july 2017.

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
[12] K. Li, G. Xu, G. Zhao, Y. Dong and D. Wang, “Cloud Task application in cloud workflow scheduling,”IEEE Transactions on
Scheduling Based on Load Balancing Ant Colony Cybernetics, vol. 50, no. 6, pp. 2715-2729, 2019.
Optimization,” 2011 Sixth Annual Chinagrid Conference, pp. 3- [16] Peyman Paknejad, Reihaneh Khorsand and Mohammadreza
9,Agust 2011. Ramezanpour, “Chaotic improved PICEA-g-based multi-objective
[13] Aujla, Sumandeep, et al,“Task scheduling in cloud using hybrid optimization for workflow scheduling in cloud environment,” Future
cuckoo algorithm,” International Journal of Computer Networks and Generation Computer Systems, vol. 117, pp. 12-28, November 2020.
Applications (IJCNA) ,vol.2,pp. 144-150,june 2015. [17] Xiumin Zhou, Gongxuan Zhang, Jin Sun, Junlong Zhou, Tongquan
[14] Y. Chen, G. Xie and R. Li, “Reducing energy consumption with cost Wei and Shiyan Hu, “Minimizing cost and makespan for workflow
budget using available budget preassignment in heterogeneous cloud scheduling in cloud using fuzzy dominance sort based HEFT ,”Future
computing systems,” IEEE Access, vol. 6, pp. 20572-20583,April Generation Computer Systems, vol. 93, pp. 278-289, April 2019.
2018. [18] Farzaneh Abazari, Morteza Analoui, Hassan Takabi and Song Fu,
[15] Zi-Jia Wang, Zhi-Hui Zhan, Wei-Jie Yu, Ying Lin, Jie Zhang, Tian- “MOWS: multi-objective workflow scheduling in cloud computing
Long Gu and Jun Zhang, “Dynamic group learning distributed based on heuristic algorithm,”Simulation Modelling Practice and
particle swarm optimization for large-scale optimization and its Theory, vol. 93, pp. 119-132, May 2019.

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 11:11:19 UTC from IEEE Xplore. Restrictions apply.
View publication stats