Time and Reliability Optimization Bat Algorithm for Scheduling Workflow in Cloud

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 06 | June-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 2048
TIME AND RELIABILITY OPTIMIZATION BAT
ALGORITHM FOR SCHEDULING WORKFLOW IN CLOUD
Prachi Chaturvedi1, Abhishek Satyarthi2, Sanjiv Sharma3
123
CSE/IT Dept., Madhav Institute of Technology & Science, RGPV, Gwalior, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Effective scheduling is one of the key
concerns while executing workflows in the cloud
environment. Workflow scheduling in clouds refers to the
mapping of workflow tasks to the cloud resources to
optimize some objective function. In this paper, we apply a
recently developed meta-heuristic method called the BAT
algorithm to solve the multi-objective problem of
workflow scheduling in clouds that minimizes the
execution time and maximizes the reliability by keeping
the budget within user specified limit. Comparison of the
results is made with basic, randomized, evolutionary
algorithm (BREA) that uses greedy approach to allocate
resources to the workflow tasks on the basis of low cost,
high reliability and improved execution time machines. It
is clear from the experimental results that the BAT
algorithm performs better than the basic randomized
evolutionary algorithm.
Keywords— Cloud Computing, BAT Algorithm, Workflow
Scheduling Algorithm.
I. INTRODUCTION
Cloud computing is emerging as a popular and widely used
platform in almost every field. Ambrust et al.[1] define
cloud computing as applications which are delivered over
the Internet in the form of services, as well as the
hardware and software systems in the data center that
actually provide these services. Buyya et al. [2] expresses
cloud computing on the basis of its utility-oriented aspect.
Cloud computing also offers several features like pay-per-
use, broad network access, virtualization, rapid elasticity
and on-demand access. These features make cloud
environment highly suitable for handling the scientific
workflow applications.
Workflow is a series of related tasks connected by control-
flow and data-flow dependencies. To run workflows on
cloud, high-level approach is to make a virtual cluster,
formed of cloud resources, and then schedule the tasks of
the workflow on that cluster. Scheduling in clouds is an
NP-Hard problem. Many scheduling methods which
include heuristic as well as meta-heuristic techniques have
been applied for workflow scheduling in clouds and more
are constantly being experimented and researched, still a
lot of work needs to be done for making scheduling in
clouds more effective. The quality of the output also
depends on whether the tasks are scheduled on reliable
virtual machines or not because virtual machines having
less failure rate are expected to successfully execute tasks
allocated to them. However, scheduling a task on more
reliable machine incurs more cost. Similarly, time and cost
are conflicting parameters that also needs to be balanced
and optimized. So, in order to get optimal solution, the
tradeoff among these objectives must be effectively
managed.
Unlike most of the scheduling algorithms that assign cloud
resources to the input tasks based on single objective
function, usually cost or execution time, in this paper, we
deal with multi-objective scheduling problem where we
link three objectives namely cost, execution time and
reliability into a single objective function. The aim of our
paper is to experiment a recently developed meta-
heuristic approach called the BAT algorithm for the
purpose of workflow scheduling in clouds with the
objective of maximizing reliability and minimizing
execution time ensuring that the cost does not exceed the
user specified budget.
II. RELATED WORK
There are several algorithms designed for scheduling
workflow in cloud. The algorithms are discussed below. A
market oriented hierarchical scheduling strategy in cloud
workflow system proposed by Wu, Zhangjun, et al. [3] is a
new approach which takes into account both the aspects
of workflow scheduling namely service level scheduling
and task level scheduling. Here a hierarchical methodology
is followed in which at the first level i.e. service level
scheduling is done by using a package based random
scheduling algorithm and the task level scheduling which
maps the tasks to VMs is done using three prominent meta
heuristic approaches namely Ant colony optimization
(ACO), Genetic algorithm (GA) and Particle Swarm
Optimization (PSO). All the three approaches are
compared and Ant Colony optimization is found to be best
according to the results. The comparison was done
considering three basic parameters at the task level which
eventually affects the Quality of Service (QoS) namely
Number of CPU cycles, Makespan and Cost. The algorithm
has been implemented using SwinDev-C [4] which is
specifically designed for cloud.

Suraj Pandey et al [5] have proposed a particle swarm
optimization based algorithm for scheduling workflow
applications. The particle swarm optimization [6]
technique was proposed by Eberhart and Kennedy in 1995
and is considered to be one of the best swarm intelligence
techniques. It gives a better convergence with a few runs.
This is because it adjusts individual value based on local
best and the global best which is the key for optimization.
The proposed algorithm mainly compares the costs
required for mapping tasks in workflow by the proposed
algorithm (PSO) and to total cost occurred when using
Best resource Selection. According to their results the PSO
based algorithm is three times more cost effective than
Best Resource Selection algorithm. The implementation is
done using Amazon EC2.
Similar to the above, Wu, Zhangjun, et al have used revised
discrete particle swarm optimization for scheduling
workflow in cloud [7]. This algorithm is similar to Particle
Swarm Optimization but with a set based concept
embedded in it. Here authors use GRASP (Greedy
randomized adaptive search procedure) to initialize the
swarm. The proposed algorithms results are compared
with Particle Swarm Optimization and Best Resource
Select algorithms results based on makespan and cost
optimization ratio. These results and statistics are
obtained by varying computational cost and
communication cost. According to the results RDPSO
proves to be better. These algorithms have been
implemented using Amazon elastic cloud compute.
The algorithms discussed till now consider only single
workflow execution but there is another algorithm which
considers scientific workflow ensemble [8]. Workflow
ensemble is a collection of several interrelated workflows
grouped together. Malawski, Maciej, et al. have proposed a
method for cost and deadline constrained provisioning for
scientific workflow ensemble in IaaS. Based on static and
dynamic strategies the authors have considered three
algorithms. The algorithms are namely Dynamic
Provisioning and Dynamic Scheduling (DPDS), Workflow
aware DPDS and Static Provisioning Static Scheduling. The
first two algorithms are dynamic in nature i.e. these
algorithms do resource provisioning and scheduling at run
time but SPSS creates the resource provisioning and
schedule plan before runtime. The implementation is done
by using CloudSim [9].
Similarly, Abrishami, Saeid, Mahmoud Naghibzadeh, and
Dick HJ Epema [10] have used same Partial critical path
for scheduling workflow in IaaS. The properties of SaaS
and IaaS are different and same approach cannot be
followed. So based on the properties of IaaS the authors
have proposed two approaches for solving the problem of
workflow scheduling in IaaS. Maintaining QoS is still the
main goal. This first approach named IC-PCP is one phase
algorithm which uses two methods namely parent
Schedule and path Schedule repetitively used to schedule
all the nodes. Here parent Schedule schedules all the
unscheduled predecessors given the scheduled node as
input and path Schedule schedules all the tasks into a
single instance of computational service with minimum
price and considering the latest finish time. Similarly
another approach called IaaS Cloud Partial Critical Paths
with Deadline Distribution (IC-PCPD2).It is two phase
approach with two phases namely Deadline Distribution
and planning phase. Both of these phases uses parent
Schedule and Path Schedule algorithm with some minor
changes in it. These two approaches are compared for
their efficiency and it is found that the IC-PCP performs
better in several cases [11].
There is another algorithm that is proposed by Lin et al
[12] which proposes solution for scheduling workflow in
elastic cloud. There are two phases involved in workflow
scheduling these are: task prioritizing phase and resource
allocation phase. In the task prioritizing phase tasks are
selected by using the proposed priority based algorithm.
In the second phase for resource allocation a new
algorithm by name Scalable Heterogeneous Earliest Finish
Time is proposed. This algorithm uses several parameters
such as Earliest Start Time, Earliest Finish Time, Earliest
Ready time and Task Finish time. This algorithm is
extension of Heterogeneous Earliest Finish time (HEFT)
and is specifically proposed for elastic cloud.
There is another algorithm proposed by Genez, Thiago AL,
Luiz F. Bittencourt, and Edmundo RM Madeira [13] for
Scheduling Workflow cloud specifically SaaS workflows
into Components of IaaS i.e. mapping the SaaS workflows
which are used by Customer and the IaaS components of
multiple IaaS which are present in the datacenter. This
method involves two levels of SLA's. Integer Linear
Program is formulated for solving the scheduling problem
and they have also proposed two heuristics to obtain
integer solution from the relaxed version of ILP. Both the
algorithms are implemented and it is found that ILP gives
a low cost solution in very less time and thereby achieving
the goal of proposing an optimized scheduling process.
Above all there is another paper by which checks the
efficiency with which scientific workflows runs in cloud.
This paper in short checks the suitability of cloud to run
scientific workflows. All the experiments are done based
on the astronomical tool named montage [14]. The
montage workflows are tested in three cloud
environments namely local environment, virtual
environment and wide area environment. According to
their results it is found that the local environment is better
for small workflows. But there is a scalability issue which
is solved by using virtual environment. But on the whole,
the wide area network (public) is found to be the best
option for scheduling scientific workflow in cloud.

In this paper the authors have proposed a new algorithm
by name iterative ordinal optimization. This method was
designed to reduce the scheduling overhead and to reduce
the time taken by the algorithm to schedule a complex
task. The authors have compared this methodology with
Monte Carlo method and blind pick methods. The
algorithm proves to be better than them. The authors have
done a detailed study and have demonstrated the working
of scientific workflows in different clouds. According to
the results the different types of clouds doesn't have any
specific significant differences. They also test scientific
workflow manager pegasus [15] over different cloud
infrastructure and come to the conclusion that sky
computing supports scientific workflow scheduling.
A lot of work has already been done in the field of
workflow scheduling in clouds using different meta-
heuristic methods. In our work, we chose the BAT
algorithm for scheduling tasks to cloud resources because
the BAT algorithm has shown good results when applied
to different optimization problems14-17 and exhibits
features like frequency tuning, parameter control and
automatic zooming into the region where the optimal
solution lie. So we are motivated to use this algorithm in
the field of workflow scheduling in clouds. Not only this,
most of the work done in this area generally take cost and
time as the objectives to be optimized. Reliability is also an
important factor along with cost and time since scheduling
tasks on low reliability machines can result in high failure
rate. The algorithm presented in the paper works to
optimize the execution time considering the user defined
budget constraint and also ensures that high reliability
virtual machines are assigned to the tasks.
III. PROPOSED WORK
3.1 The BAT Algorithm
In this paper we develop a workflow scheduling set of
rules based totally on a heuristic method ,For scheduling
workflow responsibilities on cloud assets, a new meta-
heuristic method called the BAT algorithm is carried out. It
is referred to as so due to the fact its working is primarily
based on the echolocation conduct of the digital bats17-19.
that it associates with the objective function which is to be
optimized. Bat algorithm is designed by Xin-She Yang [16].
This algorithm is a metaheuristic algorithm which is
designed for optimization purposes. This is one of the
recently developed technologies which was primarily
aimed at solving engineering problems. Metaheuristic
algorithms are known for optimization. There are several
other metaheuristic algorithms that are designed for
optimization. The most basic problem of optimization is
maximization or minimization of a function. Many
metaheuristic algorithms concentrate on this aspect of
optimization. Usually a good algorithm is characterized by
the capability of maximizing or minimizing more
accurately. Basically optimization is used for the problems
which do not have exhaustive solution. Thus the best
optimality involves finding out the best solution (Optimal
solution), usually by means of randomization.
Similarly bat algorithm uses randomization for calculating
the optimal value. To prove the superiority of an
algorithm, an algorithm is tested with several standard
functions available such as Rastrigin's function etc. If an
algorithm is able to minimize these standard functions
with utmost accuracy, then these algorithms are
considered to be good algorithms. In this context the Bat
algorithm, according to the authors is the best algorithm
as it is one of the few algorithms which gives accurate
results. This is one of the prime reasons for choosing this
algorithm for workflow scheduling. Mapping of tasks and
process units in a workflow is an important step. The
mapping process is an NP-hard problem. These NP-hard
problems cannot be solved in exhaustive manner i.e. it is
not possible by any algorithm to solve it exhaustively
within polynomial time. Hence a non-exhaustive approach
is suitable for these kinds of problem and Bat algorithm is
the obvious choice for this as it is the best among all the
metaheuristic approaches.
This behavior of echolocation can be formulated in a way
bat algorithms a kind of Swarm intelligence set of rules.
Bat set of rules turned into designed to contain residences
of several other metaheuristic algorithms so that it can
perform better than other algorithms with the aid of
thinking about the advantages and downsides of all of the
other algorithms. These algorithms include the simulated
annealing, Harmony seek and Particle swarm
optimization. The bat algorithm achieves this and is
considered to be a great set of rules which has the
positives of different metaheuristic algorithms. The
algorithm become advanced by means of Xin-She Zang
within the year 2010 and is based totally on the
subsequent regulations:
In order to sense the distance, bats use their echolocation
behavior. Bats fly with certain velocity vi at position xi.
They emit pulse whose frequency is adjusted
automatically and the rate of the emitted pulse r ‫א‬ [0, 1] is
adjusted depending on the distance of the bat from its
target. Loudness of bats varies from a large value L0 to a
minimum value Lmin.
The Bat Algorithm is iterative in nature, hence for each
iteration the parameters needs to be updated. After the
position and velocity are updated the loudness and pulse
emission rate needs to be updated. As the distance
between the bat and the prey changes the bat has to
automatically change the loudness and rate of pulse
emission. The algorithm is described in detail by Xin-She
Yang in his paper [20].

The algorithm above is described for continuous
optimization but discrete set of inputs are given here, so as
to suit the application, changes are made with reference to
the Binary Bat Algorithm (BBA) [22].
Binary Bat algorithm is based on bat algorithm with a
slight difference, bat algorithm is for continuous values
but Binary Bat Algorithm can be used to select from Binary
values. BBA algorithm is proposed for the feature
selection. For feature extraction continuous values are not
used for optimization.
Here, few modifications in binary bat algorithm are made
and with bat algorithm it is used here for selecting the
optimal cost.
Figure 1 Bat Algorithm adapted from [21]
The pseudo-code of the BAT algorithm is given in Fig. 1.
The BAT algorithm is a population based evolutionary
algorithm where each bat represents a solution. It is
designed according to the echolocation behavior of the
virtual bats. This property enables them to detect the
position of their prey. Frequency modulated signals are
used for echolocation. Bats listen to the sound pulse
emitted by them when it bounces back from the prey or
the surrounding objects. As the bat approaches near its
prey, it reduces the loudness of their echo and increases
the rate of the sound pulse. These characteristics are
formulated in a way described in the pseudo code given in
Fig. 1 such that they associate with the objective function
to be optimized. For the problem of scheduling workflows
in cloud environment, a bat individual represents the task
to virtual machine mapping for each task in the workflow.
Although the BAT algorithm has not yet been explored for
scheduling purposes in clouds but due to certain amazing
features that it exhibits like frequency tuning, capability of
automatically zooming into the regions where optimal
solution lies and parameter control, it can prove to be a
promising algorithm for scheduling cloud resources. Not
only this, algorithms like the Harmony search and PSO can
be considered as the special cases of the BAT algorithm.
Replacing variations of frequency in the BAT algorithm by
random parameters and setting loudness and pulse rate to
0 and 1 respectively, it behaves like the standard PSO. If
we fix loudness and pulse rate and do not use velocities in
the BAT algorithm, it simply reduces to the Harmony
search. So we can say that the BAT algorithm is very
powerful and have high chances to produce the good
quality results.
Objective Function
The core of any heuristic or meta-heuristic algorithm
(including the BAT algorithm) is the objective function. In
this paper, the goal is to schedule workflow tasks on cloud
resources in such a way that the execution time is
minimized and reliability is maximized while ensuring
that the cost is within the user specified budget. If S
denotes schedule, then the execution time, cost and
reliability of the schedule S are calculated as follows:
Execution time: The execution time of the schedule S is
denoted by ET(S) and is defined as the sum of the
execution time of all the tasks of the schedule S. If n is the
total number of tasks in a schedule, the execution time can
be calculated as:
where T(i) is the execution time of the task i in the
schedule and is calculated by the following formula:
Cost: The cost of the schedule S is denoted by C(S) and is
defined as the sum of execution cost of all the tasks of
the schedule S. If n is the total number of tasks, cost can be
calculated as:

where C(i) = (Instruction length of the task i) * (cost
scaling factor) and cost scaling factor is high for virtual
machines of higher reliability and low for less reliable
virtual machines.
Reliability: Reliability of a virtual machine depends on
certain characteristics of the virtual machine like the
processor speed (in MIPS), RAM size, transfer rate and
failure rate. Reliability of the schedule S is denoted by R(S)
and is decided by the following formula:
R(S) = min (r(1), r(2)...... r(n)) (4)
where n is the total number of tasks and r(1) denotes the
reliability of the VM on which first task is scheduled,
r(2) denotes the reliability of the VM on which second task
is scheduled and so on.
We denote the fitness function by F. If MET(S) is the
maximum execution time of the schedule among all the
Bat individuals and B is the user specified budget then the
fitness function is formulated as follows:
The aim is to minimize the value of the fitness function so
that the execution cost is minimized and reliability is
maximized so that the schedules whose cost exceeds the
budget constraint are ruled out.
The value of the fitness function will be less than 1 for the
schedules whose cost will be within the user specified
budget. For the schedules whose cost exceeds the user
specified budget, the value of the fitness function will
always be greater than 1.
It is to be noted that the fitness function is a minimization
function and it will assume lower values for fit schedules
and higher values for unfit schedules.
IV. RESULTS AND COMPARISONS
The WorkflowSim simulator developed by Chen et al.22 is
used for simulations. 200 virtual machines with MIPS rate
varying from 1000 to 4000 are considered for
experiments.
Population size is taken as 30 and the initial population is
generated randomly. The results of the BAT algorithm are
compared with the Basic Randomized Evolutionary
Algorithm (BREA).
BREA randomly generates a schedule and then randomly
selects a task and replaces the virtual machine assigned to
it with a machine having lower cost, then selects another
task and replaces the virtual machine assigned to it with a
machine having lower time and similarly randomly selects
third task and replaces the virtual machine assigned to it
with a machine having higher reliability.
Fig. 2 shows the comparison of the execution time of the
BAT and the BREA technique and Fig. 3 shows the
comparison of the cost value of the BAT and the BREA
technique when applied to five different workflows
namely Montage workflow with 25 nodes.
In spiral workflow with 50 nodes, Montage workflow with
100 nodes, Cyber Shake workflow with 1000 nodes and
Southeastern Coastal Ocean Observing and Prediction
Program (SCOOP) workflow.
Fig. 2. Execution time comparison of the BAT and the
BREA technique on 5 workflow applications
Fig. 3. Cost comparison of the BAT and the BREA technique
on 5 workflow applications

The horizontal axis in both the cases represents the
workflow application considered for the experiments.
The vertical axis of the graph in Fig. 2 gives the actual
execution time taken by the BAT and the BREA algorithm.
For all the five workflows, the BAT takes less time to
execute as compared to the BREA. The vertical axis in Fig.
3 represents the actual cost values. Comparison of the cost
value of the BAT algorithm and the BREA are made against
the user defined budget. The results show that for both the
algorithms, cost value does not exceed the user defined
limit and the BAT algorithm produces low cost schedule as
compared to the BREA for all the five workflows. The
results clearly show that the BAT algorithm outperforms
the BREA in all the scenarios.
The BAT algorithm schedules the cloud resources to the
tasks of the workflow in less time and higher reliability
within the user defined budget constraint as compared to
the BREA technique.
V. CONCLUSIONS
In this paper, a recently developed meta-heuristic
technique called BAT algorithm is used for workflow
scheduling in clouds. The algorithm not only optimizes the
execution time but also ensures that the tasks are assigned
to high reliability virtual machines while meeting the user
defined budget. Five different scientific workflows are
considered to experiment and evaluate the effectiveness of
the technique. The results obtained clearly show that the
BAT algorithm performs better than the Basic Randomized
Evolutionary Algorithm (BREA) in all the five scenarios.
References:
[1] Lin, Cui, and Shiyong Lu. "Scheduling scientific
workflows elastically for cloud computing." Cloud
Computing (CLOUD), 2011 IEEE International Conference
on. IEEE, 2011.
[2] Pandey, Suraj, et al. "A particle swarm optimization-
based totally heuristic for scheduling workflow
applications in cloud computing environments." Advanced
Information Networking and Applications (AINA), 2010
twenty fourth IEEE International Conference on. IEEE,
2010.
[3] Wu, Zhangjun, et al. "A revised discrete particle swarm
optimization for cloud workflow scheduling."
Computational Intelligence and Security (CIS), 2010
International Conference on. IEEE, 2010.
[4] Malawski, Maciej, et al. "Cost-and closing date-
restricted provisioning for medical workflow ensembles in
IaaS clouds." Proceedings of the International Conference
on High Performance Computing, Networking, Storage and
Analysis. IEEE Computer Society Press, 2012.
[5] Abrishami, Saeid, Mahmoud Naghibzadeh, and Dick HJ
Epema. "Deadline-limited workflow scheduling algorithms
for Infrastructure as a Service Clouds." Future Generation
Computer Systems 29.1 (2013)
[7] Wu, Zhangjun, et al. "A market-orientated hierarchical
scheduling method in cloud workflow structures." The
Journal of Supercomputing 63.1 (2013).
[8] Liu, Xiao, et al. "SwinDeW-C: a peer-to-peer based
totally cloud workflow machine." Handbook of Cloud
Computing. Springer US, 2010.
[9] Altintas, Ilkay, et al. "Kepler: an extensible machine for
layout and execution of medical workflows." Scientific and
Statistical Database Management, 2004. Proceedings. 16th
International Conference on. IEEE, 2004.
[10] Calheiros, Rodrigo N., et al. "CloudSim: a toolkit for
modeling and simulation of cloud computing
environments and evaluation of resource provisioning
algorithms." Software: Practice and Experience 41.1
(2011).
[11] Genez, Thiago AL, Luiz F. Bittencourt, and Edmundo
RM Madeira. "Workflow scheduling for SaaS/PaaS cloud
vendors considering two SLA ranges." Network Operations
and Management Symposium (NOMS), 2012 IEEE. IEEE,
[12] Kennedy, James, and Russell Eberhart. "Particle
swarm optimization." Proceedings of IEEE global
convention on neural networks. Vol. Four. No. 2. 1995.
[13] Montage : An Astronomical Image Mosaic Engine
[Online] Available:
http://montage.Ipac.Caltech.Edu/docs/gridtools.Html
[14] Tsai PW, Pan JS, Liao BY, Tsai MJ, Istanda V. Bat
algorithm stimulated set of rules for fixing numerical
optimization issues. In: Applied Mechanics and Materials
2012. Trans Tech Publications; 2012.
[15] Yang XS. Bat set of rules for multi-goal optimization.
International Journal of Bio-Inspired Computation.
[16] Ramesh B, Mohan VC, Reddy VV. Application of bat set
of rules for mixed financial load and emission dispatch. Int.
J. Of Electrical Engineering and Telecommunications.
2013.
[17] Musikapun P, Pongcharoen P. Solving multi-level
multi-system multi-product scheduling trouble using bat
algorithm. In: 2d worldwide conference on control and
artificial intelligence 2012. IACSIT Press Singapore.2012;
35:ninety eight-102.

[18] Altringham JD. Bats: biology and behaviour. Oxford,
UK: Oxford University Press; 1996.
[19] Colin T. The varienty of lifestyles. Oxford, UK: Oxford
University; 2000.
[20] Richardson P. Bats. London: Natural History Museum;
2008.
[21] Yang XS. A new metaheuristic bat-stimulated set of
rules. In: Nature stimulated cooperative strategies for
optimization (NICSO 2010). Springer Berlin Heidelberg;
2010.
[22] Chen W, Deelman E. Workflowsim: A toolkit for
simulating clinical workflows in allotted environments. In:
E-Science (e-Science), 2012 IEEE 8th International
Conference; 2012.
BIOGRAPHIES
Prachi Chaturvedi,pursued Bachlors of engineering
from Institute of Technology and Management(ITM),
(MP India in 2014. She is currentlyPursuing her Master
and Science,Gwalior (MP), India.
Sanjiv Sharma PhD, M.Tech(IT), B.E.(IT) is an
Assistant Professor in the Department of
Computer Science Engineering and Information
Technology at Madhav Institute of Technology
& Science Gwalior (MP), India. He received his
PhD degree (Computer Science & Engineering) from
Banasthali University, Jaipur (Raj.), India in 2014 and
M.Tech (Information Technology) with honors from
School Of Information Technology, Rajiv Gandhi
Proudyogiki Vishwavidyalaya, Bhopal(MP), India in 2007.
His current research interests include Social Network
Analysis, Data Mining, Network Security, Adhoc Network
and Mobile Computing and their interdependency.
of Engineering from Madhav Institute of Technology
Abhishek Satyarthi,pursued Bachlors of
Engineering From LNCT, Bhopal (MP),
India in 2013. He is Currently pursuing his
Master of Engineering from Madhav Institute
of Technology and Science,Gwalior (MP),
India

Time and Reliability Optimization Bat Algorithm for Scheduling Workflow in Cloud

Recommended

More Related Content

What's hot (16)

Similar to Time and Reliability Optimization Bat Algorithm for Scheduling Workflow in Cloud (20)

More from IRJET Journal (20)

Recently uploaded (20)

Time and Reliability Optimization Bat Algorithm for Scheduling Workflow in Cloud