Multi-objective tasks scheduling using bee colony algorithm in cloud computing

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 12, No. 5, October 2022, pp. 5657~5666
ISSN: 2088-8708, DOI: 10.11591/ijece.v12i5.pp5657-5666  5657
Journal homepage: http://ijece.iaescore.com
Multi-objective tasks scheduling using bee colony algorithm in
cloud computing
Mehdi Salehi Babadi1
, Mohammad Ebrahim Shiri1
, Mohammad Reza Moazami Goudarzi2
,
Hamid Haj Seyyed Javadi3
1
Department of Computer Engineering, Borujerd Branch, Islamic Azad University, Borujerd, Iran
2
Department of Mathematics, Borujerd Branch, Islamic Azad University, Borujerd, Iran
3
Department of Mathematics and Computer Science, Shahed University, Tehran, Iran
Article Info ABSTRACT
Article history:
Received Feb 20, 2021
Revised Mar 10, 2022
Accepted Apr 8, 2022
Due to the development of communication device technology and the need
to use up-to-date infrastructure ready to respond quickly and in a timely
manner to computational needs, the competition for the use of processing
resources is increasing nowadays. The scheduling tasks in the cloud
computing environment have been remained a challenge to access a quick
and efficient solution. In this paper, the aim is to present a new tactic for
allocating the available processing resources based on the artificial bee
colony (ABC) algorithm and cellular automata for solving the task
scheduling problem in the cloud computing network. The results show the
performance of the proposed method is better than its counterparts.
Keywords:
Artificial bee colony algorithm
Cellular automata
Cloud computing
Resource allocation
This is an open access article under the CC BY-SA license.
Corresponding Author:
Mohammad Ebrahim Shiri
Department of Computer Engineering, Borujerd Branch, Islamic Azad University
Borujerd, Iran
Email: shiri@aut.ac.ir
1. INTRODUCTION
With the dramatic increase in the variety of information technology. equipment and services, the
management of services offered in this area has also faced many challenges. Managing problems and
requests, managing equipment and resources related to technical support services, and allocating them to
users, as well as monitoring, controlling, and scheduling are among the causes that force information
technology (IT) [1]. Managers to provide useful and efficient tools themselves. There is also a need for
people to do their heavy computing work without having expensive hardware and software through services
[2]. Cloud computing has been the latest technology response to these needs. The National Institute of
Technology and Standards defines cloud computing as: “cloud computing is a model for providing easy
access based on the demand of user by the network to a set of changeable and configurable computing
resources (e.g., networks, servers, storage, applications, and services) that this access can be provided or
release with minimal need for resources management or the need for direct intervention by the service
provider”. Sharing “intangible and consumable” computing power among several tenants can improve
productivity rates because other servers are not idle for any reason with this effective way and computers are
being used more because cloud computing clients do not need to calculate and determine their maximum
load [3].
A cellular automaton is a mathematical model for representing systems in which objects called cells
together model system behavior, which can be defined in one or more dimensional formats [4]. The

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 12, No. 5, October 2022: 5657-5666
5658
homogenous and parallel structure of cellular automata makes it suitable for modeling different types of
physical systems. To optimally model a physical system, the simple structure of cellular automata is used as
finite and local interactions between cells. The simple structure is defined as a one-dimensional cellular
automaton and each cell with two states (1.0) with a uniform ternary neighbor (itself and its right and left
neighbors). The most important feature of a cellular automata structure is its modularity. According to
interactions [4], it is observed that provide one-dimensional three-neighbor cellular automata provides the
best efficiency for modeling physical systems.
In this paper, a new task scheduling problem in a cloud environment, a highly distributed computing
platform, is concentrated. A time-aware scheduling algorithm based on artificial bee colony (ABC) is
proposed to solve this problem. The primary objective of the proposed algorithm is to achieve a good
trade-off between response time, schedule length ratio, and efficiency to complete the tasks in the Cloud
system. Our approach was experimentally evaluated and on many datasets of various sizes. The results prove
that the proposed algorithm achieved better results and was more optimal.
In [5], a new different optimization algorithm, evolution by combining elite to schedule tasks in the
cloud computing environment is voluntary, which is called multi-objective ant lion optimizer (MOALO).
MOALO algorithm has performed better than other algorithms. In [6], a new load balancing was also, which
proposed by combining a cuckoo search algorithm with a bee colony called a self-adaptive artificial bee
colony (SABC). It increases the utilization of resources, and execution time is reduced. Best task-to-virtual
machine mapping is computed by the proposed algorithm. Speed of virtual machines (VM) processing and
submitted workload length has an effect on it.
In [7], proposes an in-depth reinforcement learning model based on learning the quality of service
(QoS) feature to optimize data center resource planning. In the in-depth learning phase, we propose a QoS
feature learning method based on the enhancement of automatic voice removal encoders to extract stronger
QoS feature information. The study [8] provides a technical analysis of cloud service deployment approaches
in internet of things (IoT) systems. The key point of this technical analysis is to identify the basic studies in
service placement approaches that need more attention to develop more efficient and effective strategies in
IoT locations.
In [9], a new cloud computing task-scheduling algorithm that introduces min-min and max-min
algorithms to generate initialization population selects task completion time and load balancing as double
fitness functions and improves the quality of initialization population, algorithm searchability, and
convergence speed was proposed. In [10], for task scheduling in the cloud, we use particle swarm
optimization (PSO), firefly algorithm (FA), bat algorithm (BA), and grasshopper optimize cloud algorithm
(GOA), which are swarm-based algorithms. The experimental results indicate that the improved GOA can
optimize task scheduling problems by effective utilization of available resources.
In [11] provides a new approach for improving the task scheduling problem in a cloud-fog
environment in terms of execution time (makespan) and operating costs for bag-of-tasks applications. A task
scheduling evolutionary algorithm has been proposed. A single custom representation of the problem and a
uniform intersection is built for the proposed algorithm. In [12], a task scheduler based on a genetic
algorithm for the cloud computing system was invented that used the genetic algorithm [13] to minimize the
time to complete tasks in the cloud computing environment.
In [14], an ant algorithm for symmetric scheduling of tasks on cloud computing was presented. The
goal is to allocate the optimal resource to each task according to the source and task characteristics. Each task
is regarded as an ant, and the weight of the resources is regarded as a pheromone. In other words, the higher
the source weight, the higher the pheromone. The results show the optimum completion time and balance. In
[15] to enhance the performance of cloud computing and reduce delay time in the queue waiting for jobs. The
proposed algorithm tries to avoid some significant challenges that throttle from developing applications of
clouding computing. Our experimental result of the proposed job scheduling algorithm shows that the proposed
schemes possess outstanding enhancing rates with a reduction in waiting time for jobs in the queue list.
Paper [16], apply the latest whale optimization metamorphosis (WOA) algorithm to schedule cloud
work with a multi-objective optimization model, with the aim of improving the performance of a cloud
system with given computing resources. Accordingly, we propose an advanced approach called WOA cloud
scheduling (IWC) Improvement to further improve the ability to search for the optimal WOA-based
approach. In [17] paper, a modified henry gas solubility optimization (HGSO) is presented, which is based on
the WOA and comprehensive opposition-based learning (COBL) for optimum task scheduling. The proposed
method is named henry gas solubility whale cloud (HGSWC). HGSWC is validated on a set of thirty-six
optimization benchmark functions, and it is contrasted with conventional HGSO and WOA.
In [18] proposes three main contributions to solve this load balancing problem. First, it proposes a
heterogeneous initialized load balancing (HILB) algorithm to perform a good task scheduling process that
improves the makespan in the case of homogeneous or heterogeneous resources and provides a direction to

Int J Elec & Comp Eng ISSN: 2088-8708 
Multi-objective tasks scheduling using bee colony algorithm in cloud computing … (Mehdi Salehi Babadi)
5659
reach optimal load deviation. Second, it proposes a hybrid load balance based on a genetic algorithm
(HLBGA) as a combination of HILB and genetic algorithm (GA). Third, a newly formulated fitness function
that minimizes the load deviation is used for GA [19] proposes an efficient algorithm traffic-aware adaptive
server load balancing (TAASLB) to balance the flows to the servers in a data center network. It works based
on two parameters, residual bandwidth and server capacity. It detects the elephant flows and forwards them
towards the optimal server where they can be processed quickly.
In [20], proposed hybrid electro search with a genetic algorithm (HESGA) to improve the behavior
of task scheduling by considering parameters such as makespan, load balancing, utilization of resources, and
cost of the multi-cloud. The proposed method combined the advantage of a genetic algorithm and an electro
search algorithm. The genetic algorithm provides the best local optimal solutions, whereas the electro search
algorithm provides the best global optima solutions. The proposed algorithm outperforms existing scheduling
algorithms such as hybrid particle swarm optimization genetic algorithm (HPSOGA), GA, evolution strategy
(ES), and ant colony optimization algorithm (ACO). In [21], an efficient hybridized scheduling algorithm
that replicates the parasitic behavior of the cuckoo and food gathering habit of the crow bird, named the
cuckoo crow search algorithm (CCSA), had been presented for improvising the task scheduling process. The
crow bird always stares at its neighbors, looking for a better food source than the one it currently possesses.
In [22], a resource scheduling optimization model based on service level agreement based on a
random planning approach in cloud computing is presented. In [23], the flexible task scheduling problem in a
cloud computing system is studied and solved by a hybrid discrete artificial bee colony (ABC) algorithm,
where the considered problem is first modeled as a hybrid flow shop scheduling (HFS) problem. Both a
single objective and multiple objectives are considered. In [24], a cloud computing multi-objective task
scheduling optimization based on a fuzzy self-defense algorithm is proposed. Select the shortest time, the
degree of resource load balance, and the cost of multi-objective task completion as the goal of cloud
computing multi-objective task scheduling, establish a mathematical model to measure the effect of
multi-objective task scheduling and construct the objective function of cloud computing multi-objective task
scheduling.
2. METHOD
2.1. Problem statement
The cloud management program manages all cloud resources using various cloud modules such as
network module, operating system image module, cost module, and endorsement module. The tasks are
distributed in different data centers (DC’s) available through the cloud infrastructure. Each data center
divides the user tasks into several “sub-task” and makes them available to processing elements (PE) [25].
2.2. Proposed schema
The proposed scheduling module will have the task of assigning the right job to the right source at
the right time in the framework of cloud. In Figure 1, DC represents a data center and PE represents the
processing elements. In this model, a new cloud is regarded as a set of user tasks that its complex computing
is performed using cloud resources. Suppose UserJob=(𝑈1, 𝑈2, … , 𝑈𝑁) is a set of user programs that enters
(execution-request) at a given time. Each UserJob (Ui) is represented by a double <ai, di>, where ai
represents the time of entry, and di represents the time limit of the UserJob. If a task fails within its time
limit, it is designated as a failed task and must be re-entered into a new scheduling queue. In the scheduling
process, user tasks are assigned to available data centers (𝐷1, 𝐷2, … , 𝐷𝑀), where M<=N, means that the
number of data centers may be less than the number of tasks requested. Each data center (Di) is represented
by a double <Ci, mi>. In this dual, Ci is the cost of executing tasks in the data center per unit time, and mi is
the number of PEs available for executing user tasks. Each data center has a number of processing elements
{𝑃𝐸1, 𝑃𝐸2, … , 𝑃𝑒𝑘} to execute users' work. Each processing element with a PE’s characteristic also means
processing speed[26].
2.3. Problem parameters
2.3.1. Response-time problem analysis
These tasks are embedded in one of the existing datacenters called D, which has a certain number of
PE processing units, and these tasks must be distributed among these processing units. Each U has a required
processing volume, i.e., P and an allowed time i.e., T. Each PE also has a PEj processing speed and a
processing cost. If the execution time of Tk in 𝑃𝑒𝑗 is shown by τk, then the termination time can be stated:
this processing time is proportional to the difference in processing volume required for 𝑃𝑘 in 𝑃𝑒𝑗.
𝑃𝑘𝑟𝑒𝑚𝑎𝑖𝑛𝑖𝑛𝑔
= 𝑃𝐾𝑖𝑛𝑖𝑡𝑖𝑎𝑙 − 𝜏𝑘 ∗ 𝑃𝐸𝑗 <= 𝜏𝑘 =
𝑃𝑘
𝑃𝐸𝑗
(1)

 ISSN: 2088-8708
5660
If 𝑃𝑘> 𝑃𝑒𝑗 is, then the Tk spent for 𝑈𝑘 task processing is decreased from the amount of processing
required, and some processing remains. Naturally, we want this processing time to be less than the time
allowed to perform the full task of the user task, i.e., Tim. In general, assuming the parallel distribution of 𝑈𝑘
tasks among 𝑃𝑒𝑗 processors, the time of completing task U is equal to the time of completing the longest tasks.
𝑀𝑎𝑘𝑒𝑠𝑝𝑎𝑛 = 𝑚𝑎𝑥{𝐹𝑖𝑛𝑖𝑠ℎ (𝜏𝑘)} (2)
Figure 1. Management structure of a cloud processor [26]
2.3.2. Energy consumption problem analysis
If the unit processing cost per 𝑃𝑒𝑗 is equal to 𝑃𝐶𝑗, then the energy consumed to perform the whole
task U is equal to (3).
EU = ∑ (τk ∗ PCj)
n
k=1 (3)
This is equal to the cost spent for the UserJob in the D datacenter.
2.3.3. The efficiency formulation
Here, a new relationship must be established to determine the efficiency of the task distribution
algorithm. This productivity can be obtained by comparing the amount of processing that could do with the
whole time of D (data) with U (UserJob) with the amount of processing that actually performed. That’s mean:
eff =
∑ (τk∗PEjk)
n
k=1
Makespan∗∑ PEj
J
j=1
−1
(4)
This equation is equal to the inverse of the sum of the time that each processor unit was occupied by
a particular task divided by the total time U task multiplied by the total available processing power.

5661
Therefore, the lower the total execution time of the task and the more optimized processing power is
distributed among the tasks, the higher the efficiency of the algorithm. The schedule length ratio (SLR)
parameter can also be obtained from this (5).
𝑆𝐿𝑅 =
𝑚𝑎𝑥{𝐹𝑖𝑛𝑖𝑠ℎ (𝜏𝑘)}
𝑚𝑖𝑛{𝐹𝑖𝑛𝑖𝑠ℎ (𝜏𝑘)}
(5)
Now suppose each task U has linear cell automata that correspond to the order of T subtasks, and
each cell automata have a bee with greedy selection. At each m period of the processing resource allocation
algorithm operation, we have a number of 𝑃𝑘 residual from different 𝑈𝑘, each of which has a bee with
greedy selection. To continue task for each 𝑃𝑘, one power of choice is definedwhich is the ability of the bee
assigned to it, which is equal to the amount of processing operations residual from the 𝑈𝑘 task. In short, any
remaining or new task in any given period will choose the most powerful of the available processing
resources, depending on the selection power it has, and this process will continue until all tasks are
completed. If we show the periods of allocating tasks by m, each bee's selection power is (6).
𝑃𝑂𝑊𝑘 = 𝑃𝑘𝑖𝑛𝑖𝑡𝑖𝑎𝑙 − ∑ 𝜏𝑘(𝑚)
𝑀
𝑚=1 ∗ 𝑃𝐸𝑆𝑗(𝑚) (6)
That is, the selection power of each bee for its own task is equal to the initial amount of processing
required for that task minus the processing power allocated to that task at different times. However, with each
iteration of the resource allocation algorithm of task 𝑈𝑘, that has the most selection power to obtain the
highest 𝑃𝑒𝑗. However, the energy consumption equation is also rewritten as (7).
EU = ∑ ∑ (τkm ∗ PCjm)
n
k=1
M
m=1 (7)
And the efficiency of the algorithm is also equal to (8).
eff =
∑ ∑ (τk∗PEjm)
n
k=1
M
m=1
Makespan∗∑ PEj
J
j=1
−1
(8)
The unknown inputs of the problem are the number of tasks available and the processing volume of
each one, the number of processors and processing power, and the processing cost of each, and the outputs of
the problem, as well as the total processing time, total energy consumed, and algorithm efficiency that are
compared in two states of distribution with greedy selection and constant random distribution. In the constant
random distribution, each task is assigned as a processing resource at the beginning of the task, and this
allocation does not change anymore. One remaining point is that given the extension of servers and
processing resources in cloud systems, the probability that the number of tasks assigned is greater than the
number of processors is very low, usually 𝐽 ≥ 𝐾.
3. RESULTS AND DISCUSSION
3.1. Test data of the problem
After introducing the overview, examining the technical terms, and reviewing similar work, we were
able to present a new theory for the distribution of cloud computing resources among existing tasks. It is time
to look at the results and judge the validity and effect clouds of the proposed technique. First, we implement
the simulation with a 12-sub task process on a cloud processor with 20 hardware sources. Naturally, a virtual
machine is defined simultaneously for each task, but how each virtual machine accesses the hardware is set
by the proposed algorithm.
The first step is to determine the number of different hardware resources with different processing
power. The reason for the random selection is that according to the use of cloud servers from discrete
processing resources in a very large area of the Internet network and connection and disconnection of many
of these resources, it cannot be assured from existing processing power. The processing values required in
each sub-task are also unknown and randomly determined in a reasonable range.
Also, the processing cost depends not only on the processing power but also on other parameters
such as the distance, and position of the processor. This issue should also be determined randomly. After
specifying the tasks and hardware resources available for each task, automata and a bee is assigned, and their
performance is such that the location of each subtask may vary depending on the selection power of the bee during
the execution of the program, as shown in the Table 1. Whole task execution using cellular automata to divide
training resources and tasks in this example took 48 time periods, i.e., 12 parallel automata changed 48 times.

 ISSN: 2088-8708
5662
Table 1. Change of the 12-cell automata in the first 15 iterations of the algorithm
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 7 8 9 10 11 12
7 14 11 6 2 4 12 8 16 10 20 12
7 14 11 6 2 4 12 8 16 10 20 12
7 14 11 6 2 4 12 8 16 10 20 12
7 14 11 6 2 4 12 20 8 10 16 12
7 14 11 6 2 4 12 16 20 10 8 12
7 14 11 6 2 4 12 16 20 10 8 12
7 14 11 12 2 4 6 16 20 10 8 12
7 14 11 6 2 12 4 16 20 10 8 12
7 14 11 4 2 12 6 16 20 10 8 12
7 14 12 4 2 6 11 16 20 10 8 12
7 14 6 4 2 11 12 16 20 10 8 12
7 14 11 12 2 4 6 16 20 10 8 12
7 14 4 12 2 6 11 16 20 10 8 12
7 14 12 6 2 11 4 16 20 10 8 12
3.1.1. The response times
As we said, the whole task is processed by the available cloud resources once using the variable
cellular automata and once by the allocation of random processing resources. Now we compare the output
parameters between these two modes. As shown in Figure 2, processing time of whole tasks with variable
automata in a left column equals 48 seconds, and random allocation of resources in a right column equals
220 seconds the difference is very significant. As it can be seen, the proposed algorithm in this study reduces
the processing time by more than four times.
Figure 2. Processing time of whole tasks with variable automata
3.1.2. Energy consumption criterion
In Figure 3 shows energy consumption criterion of fault data in favorable case. The energy
consumption in the two methods used in the cellular automata method was about 450,000 joules and in the
random allocation method was about 3.6 million joules. It can be seen that the proposed method reduces
energy consumption by eight times compared to the random allocation method 1. In Table 1 change of the
12-cell automata in the first 15 iterations of the algorithm. After these 15 iterations, due to zero of most
processors remained, other automata elements were not changed until the last remaining processing.
3.1.3. The efficiency criterion
Figure 4 shows the efficiency criterion. Comparison of processing efficiency in two random
distribution modes and cellular automata with astronomical differences is shown in Figure 4. It is observed
that the improvement of all system performance parameters in the use of cellular automata with greedy bee
selector is more significant than the random allocation of resources.
3.1.4. The SLR criterion
Figure 5 show comparison of SLR parameter in mentioned methods. It can be seen that the ratio of
the longest process to the shortest one in cellular automata has been 45 times and in random distribution
method 220 times. It can be seen that the improvement of all system performance parameters in the use of

5663
cellular automata with a greedy bee selector compared to random allocation of resources is very significant.
It seems that the proposed method can have a significant impact on improving cloud computing services.
In order to investigate the performance of the algorithm under different conditions, we are cloud the
simulation in 45 sub-tasks with a maximum load of 4,000 processing units and distributed among
65 hardware sources with a processing capability of 1,000 units and a processing cost of 10,000 units. The
results are Figure 6, show the time taken to perform all tasks in the presented method was only eight periods,
and in the random allocation method, the tasks were over 1,400 periods, which the difference is very
significant. The cost of processing in the proposed method was nearly two thousand times less than the
random allocation method as shown in Figure 7.
Figure 8 show the processing efficiency seems to have decreased here, unlike the previous
simulation, and it is less than the random distribution of resources. In the case of random distribution, some
processing resources seem to have spent much more time than other processing sources, and this has been
influenced in the following graph. Figure 9 shows the SLR parameter, which shows the time difference of
working among the cloud processor, is much less than the random distribution method in the presented
method, which indicates the balanced distribution of tasks in the current algorithm.
Figure 3. Energy consumption criterion
Figure 4. The efficiency criterion
Figure 5. Comparison of SLR parameter

 ISSN: 2088-8708
5664
Figure 6. Total processing time with random processor dist.
Figure 7. The cost of processing in the proposed method was nearly two thousand times less than
the random allocation method
Figure 8. Processing efficiency with greedy Bees alg.
Figure 9. SLR with greedy Bees alg.

5665
4. CONCLUSION AND FUTURE WORKS
In the two simulations above, it is assumed that the power of the preceding available hardware
resources was less than the amount of processing volume required for each task. To finish the discussion, we
run the simulation in a different way. This time, 65 tasks with a maximum processing volume of 5,000 units
are distributed among a hundred cloud processors with a maximum processing power of 7,000 units. The
processing cost is 10,000 units. The results showed that the current method still has had superiority to the
random allocation of resources, but this level of superiority was not as significant as in the previous cases.
The proposed algorithm for actively distributing processing resources between the tasks presented seems to
be more effective when the available hardware processing power is less than the processing power required
by the tasks. The remaining point is the time allowed for processing, which according to the speed limit and
the number of processors, it cannot consider this time constraint. While from the users’ point of view, the
length of time is undesirable, but at the same time, the user cannot expect any task that expects the cloud
system to be done by the user. however, the allowed processing time of a parameter is independent, and any
value can be assumed, and it cannot expect the process distribution system to do a very large task using a
small processor small in a short time.
There are several suggestions for future research: i) applying input tasks in specific cases, infinitely
different states can be considered for sequencing and processing tasks, and system performance is examined
in this case; ii) considering two parallel user tasks for engaging processors that become idle before
completing all tasks; iii) comparison of techniques other than a random distribution of tasks for comparison
with the proposed method; iv) considering unforeseen factors such as the non-availability of some processing
resources. The current method has many advantages compared to the random distribution method, but to
complement other discussions, researchers can provide other tactics of distributing processing resources at
the same time as the method proposed in this study and compare their performance with the implementation
of both simulations.
REFERENCES
[1] R. Ding and Z. X. Guo, “Microstructural modelling of dynamic recrystallisation using an extended cellular automaton approach,”
Computational Materials Science, vol. 23, no. 1–4, pp. 209–218, Apr. 2002, doi: 10.1016/S0927-0256(01)00211-7.
[2] H. Tabrizchi and M. Kuchaki Rafsanjani, “A survey on security challenges in cloud computing: issues, threats, and solutions,”
The Journal of Supercomputing, vol. 76, no. 12, pp. 9493–9532, Dec. 2020, doi: 10.1007/s11227-020-03213-1.
[3] F. Mahan, S. M. Rozehkhani, and W. Pedrycz, “A novel resource productivity based on granular neural network in cloud
computing,” Complexity, vol. 2021, pp. 1–15, May 2021, doi: 10.1155/2021/5556378.
[4] P. Christen and O. Del Fabbro, “Automatic programming of cellular automata and artificial neural networks guided by
philosophy,” in New Trends in Business Information Systems and Technology, Springer, 2021, pp. 131–146. doi: 10.1007/978-3-
030-48332-6_9.
[5] L. Abualigah and A. Diabat, “A novel hybrid antlion optimization algorithm for multi-objective task scheduling problems in
cloud computing environments,” Cluster Computing, vol. 24, no. 1, pp. 205–223, Mar. 2021, doi: 10.1007/s10586-020-03075-5.
[6] R. Kumar and A. Chaturvedi, “Improved cuckoo search with artificial bee colony for efficient load balancing in cloud computing
environment,” in Smart Innovations in Communication and Computational Sciences, 2021, pp. 123–131. doi: 10.1007/978-981-
15-5345-5_11.
[7] B. Wang, F. Liu, and W. Lin, “Energy-efficient VM scheduling based on deep reinforcement learning,” Future Generation
Computer Systems, vol. 125, pp. 616–628, Dec. 2021, doi: 10.1016/j.future.2021.07.023.
[8] L. Heng, G. Yin, and X. Zhao, “Energy aware cloud‐edge service placement approaches in the Internet of Things
communications,” International Journal of Communication Systems, vol. 35, no. 1, Jan. 2022, doi: 10.1002/dac.4899.
[9] G. E. Weiqing and C. Yanru, “Task-scheduling algorithm based on improved genetic algorithm in cloud computing environment,”
Recent Advances in Electrical and Electronic Engineering (Formerly Recent Patents on Electrical and Electronic Engineering),
vol. 14, no. 1, pp. 13–19, Jan. 2021, doi: 10.2174/2352096513999200424075719.
[10] A. Zandvakili, N. Mansouri, and M. M. Javidi, “Swarm-based algorithms using chaos for task scheduling in cloud,” in 2021 7th
International Conference on Web Research (ICWR), May 2021, pp. 211–215. doi: 10.1109/ICWR51868.2021.9443157.
[11] M. N. Abdulredha, B. A. Attea, and A. J. Jabir, “An evolutionary algorithm for task scheduling problem in the cloud-fog
environment,” Journal of Physics: Conference Series, vol. 1963, no. 1, Jul. 2021, doi: 10.1088/1742-6596/1963/1/012044.
[12] K. Karmakar, R. K. Das, and S. Khatua, “Resource scheduling for tasks of a workflow in cloud environment,” in International
Conference on Distributed Computing and Internet Technology, 2020, pp. 214–226. doi: 10.1007/978-3-030-36987-3_13.
[13] Vijindra and S. Shenai, “Survey on scheduling issues in cloud computing,” Procedia Engineering, vol. 38, pp. 2881–2888, 2012,
doi: 10.1016/j.proeng.2012.06.337.
[14] A. Mubeen, M. Ibrahim, N. Bibi, M. Baz, H. Hamam, and O. Cheikhrouhou, “Alts: an adaptive load balanced task scheduling
approach for cloud computing,” Processes, vol. 9, no. 9, Aug. 2021, doi: 10.3390/pr9091514.
[15] A. S. Abdalkafor and K. M. Ali Alheeti, “A hybrid approach for scheduling applications in cloud computing environment,”
International Journal of Electrical and Computer Engineering (IJECE), vol. 10, no. 2, pp. 1387–1397, Apr. 2020, doi:
10.11591/ijece.v10i2.pp1387-1397.
[16] X. Chen et al., “A WOA-based optimization approach for task scheduling in cloud computing systems,” IEEE Systems Journal,
vol. 14, no. 3, pp. 3117–3128, Sep. 2020, doi: 10.1109/JSYST.2019.2960088.
[17] M. Abd Elaziz and I. Attiya, “An improved Henry gas solubility optimization algorithm for task scheduling in cloud computing,”
Artificial Intelligence Review, vol. 54, no. 5, pp. 3599–3637, Jun. 2021, doi: 10.1007/s10462-020-09933-3.
[18] W. Saber, W. Moussa, A. M. Ghuniem, and R. Rizk, “Hybrid load balance based on genetic algorithm in cloud environment,”
International Journal of Electrical and Computer Engineering (IJECE), vol. 11, no. 3, pp. 2477–2489, Jun. 2021, doi:
10.11591/ijece.v11i3.pp2477-2489.

 ISSN: 2088-8708
5666
[19] C. Fancy and M. Pushpalatha, “Traffic-aware adaptive server load balancing for software defined networks,” International
Journal of Electrical and Computer Engineering (IJECE), vol. 11, no. 3, pp. 2211–2218, Jun. 2021, doi:
10.11591/ijece.v11i3.pp2211-2218.
[20] S. Velliangiri, P. Karthikeyan, V. M. Arul Xavier, and D. Baswaraj, “Hybrid electro search with genetic algorithm for task
scheduling in cloud computing,” Ain Shams Engineering Journal, vol. 12, no. 1, pp. 631–639, Mar. 2021, doi:
10.1016/j.asej.2020.07.003.
[21] P. Krishnadoss, “CCSA: hybrid cuckoo crow search algorithm for task scheduling in cloud computing,” International Journal of
Intelligent Engineering and Systems, vol. 14, no. 4, pp. 241–250, Aug. 2021, doi: 10.22266/ijies2021.0831.22.
[22] Q. Li and Yike Guo, “Optimization of resource scheduling in cloud computing,” in 2010 12th International Symposium on
Symbolic and Numeric Algorithms for Scientific Computing, Sep. 2010, pp. 315–320. doi: 10.1109/SYNASC.2010.8.
[23] J. Li and Y. Han, “A hybrid multi-objective artificial bee colony algorithm for flexible task scheduling problems in cloud
computing system,” Cluster Computing, vol. 23, no. 4, pp. 2483–2499, Dec. 2020, doi: 10.1007/s10586-019-03022-z.
[24] X. Guo, “Multi-objective task scheduling optimization in cloud computing based on fuzzy self-defense algorithm,” Alexandria
Engineering Journal, vol. 60, no. 6, pp. 5603–5609, Dec. 2021, doi: 10.1016/j.aej.2021.04.051.
[25] X. Zhang et al., “Runtime model-based management of diverse cloud resources,” in Proc. of the 16th International Conference on
Model Driven Engineering Languages and Systems, 2013, pp. 572–588. doi: 10.1007/978-3-642-41533-3_35.
[26] C. A. Lee, “A perspective on scientific cloud computing,” Oct. 2010. doi: 10.1145/1851476.1851542.
BIOGRAPHIES OF AUTHORS
Mehdi Salehi Babadi received his M.Sc. degree in software Engineering from
Department of Computer Engineering, Borujerd Branch, Islamic Azad University, Borujerd,
Iran, in 2012. Currently, He is a Ph.D. student in the Department of Computer Engineering at
Borujerd Branch, Islamic Azad University, Borujerd, Iran since 2015. His Ph.D. advisor is
Professor Mohammad Ebrahim Shiri. His research interests include social network, cellular
automata, IoT, cloud computing, cryptography and security. He can be contacted at email:
msalehib@hotmail.com.
Mohammad Ebrahim Shiri received his Ph.D. degree in Computer Science from
the Department of Computer Science, university of Montreal, Montreal, Canada in 2000.
Currently, he is an Assistant Professor in the Department of Computer Science at Amirkabir
University of Technology in Tehran, Iran. His research interests include machine learning,
e-learning, database, network security and cloud computing. He can be contacted email:
shiri@aut.ac.ir.
Mohammad Reza Moazami Goudarzi received his Ph.D. degree in the
Department of Mathematics, Islamic Azad University, Tehran, Iran in 2011. Currently, he is an
assistant professor in the Department of Mathematics at Borujerd Branch, Islamic Azad
University, Borujerd, Iran. His research areas include operation research and data envelopment
analysis. He can be contacted at email: mrmoazamig@gmail.com.
Hamid Haj Seyyed Javadi received the M.Sc. and Ph.D. degrees in Amirkabir
University of Technology, Tehran, Iran in 1996 and 2003 respectively. He has been working as
a fulltime faculty member and Professor in the Department of Mathematics and Computer
Science at Shahed University, Tehran, Iran. His research interests are IoT, computer algebra,
cryptography and security. He can be contacted at email: h.s.javadi@shahed.ac.ir.

Multi-objective tasks scheduling using bee colony algorithm in cloud computing

Recommended

More Related Content

Similar to Multi-objective tasks scheduling using bee colony algorithm in cloud computing (20)

More from IJECEIAES (20)

Recently uploaded (20)

Multi-objective tasks scheduling using bee colony algorithm in cloud computing