Service Request Scheduling in Cloud Computing using Meta-Heuristic Technique: Teaching Learning Based Optimization (TLBO)

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 2198
SERVICE REQUEST SCHEDULING IN CLOUD COMPUTING USING META-
HEURISTIC TECHNIQUE: TEACHING LEARNING BASED OPTIMIZATION
(TLBO)
Kritika Shrivastava1, Dr.Ramesh Kumar2
1PG Scholar, Department of Computer Science & Engineering, Bhilai Institute of Technology (BIT), Durg, Chhattisgarh,
India
2Professor, Department of Computer Science & Engineering, Bhilai Institute of Technology (BIT), Durg, Chhattisgarh,
India
-----------------------------------------------------------------------------***-------------------------------------------------------------------------
Abstract - Cloud computing is the field in the IT world, which comprises of all kind of services provided over the internet. But,
for its flawless performance and smooth services delivery to the users it needs to be effectively managed, that can be by the mode
of “Scheduling”. There are different types of scheduling done in cloud environment. Many optimization algorithms have also been
applied in scheduling which gives sub-optimal solution for the problem domain i.e. Heuristic optimization technique. But, in
Heuristics optimization time taken is too large for wider set of problems. So, for solving the large problem set we use Meta-
Heuristics Optimization technique which gives near optimal solution for the problem and solves it in a particular limited time for
the given set of search spaces. Heuristic optimization is the subset of the Meta-Heuristic optimization, which is problem
independent which means it can be applied to any set of problems whereas heuristic approach is problem specified which means it
is designed for solving a particular problem only. In this paper, we are using nature inspired meta-heuristics method “Teaching
Learning Based Optimization” (TLBO) for scheduling in cloud computing among the user and the cloud service provider: Service
Request Scheduling using TLBO. We will generate the effective scheduling result in form of comparative analysis with the other
meta-heuristic algorithm. comparative analysis with the other meta-heuristic algorithm.
Keywords: Cloud Computing, Service Request Scheduling, Meta-Heuristics, Heuristic, TLBO, GA, PSO, Fitness.
1. INTRODUCTION
Cloud computing-an environment which deals with 3-tiers architecture: ‘Consumer/User’, ‘Service provider’ and ‘Resource
provider’. We know, very well now that cloud computing environment provides shared pool of resource on-demand request
from the user via internet. The cloud computing fulfills different features like storage management, computation management,
web based resources for the users as requested by them.
There are different types of clouds based on their deployments they are: Public, Private, Hybrid & community cloud. [3]
Flexibility, scalability, reliability, multi-tasking, availability, virtualization & easy computation are the various characteristics of
the cloud which makes it so popular among the people. [1]
The Cloud Environment Services is categorized in 3 parts as: Infrastructure as a Service (IaaS), Platform as a Service (PaaS) &
Software as a Service (SaaS). IaaS provides the resources, virtualized resources to the user as a stored data etc that too on
demand. PaaS makes it easy for user to program in cloud or to develop some application.[2] Software, Application for use is
provided by SaaS. Various Examples of cloud are Amazon’s EC2, S3 and salesforce.com. [9].But to get the proper blend of each
of its characteristics, features to the users there is need to have a staunch functioning among each of the level of the 3 tier
architecture of Cloud, for which “Scheduling” is a must, so that effective services is being reached to the user in less time and
with ease of use.
There are many scheduling done in cloud computing, but mostly two types of scheduling is being focused in 3 - tier
architecture of cloud “Resource Scheduling” . Resource scheduling is Scheduling between the service provider and
resource provider [11] whereas service request scheduling is between the user and the service provider.Many researchers
have focused on the scheduling issue of allocating resources and tasks in the cloud computing system i.e. on resource
scheduling part. But, scheduling between the user and the service provider is more complicated [4] as service provider is

being flooded with many of the user request at the same time and not many researchers have been carried out in this
scheduling as compared to the resource scheduling.
2. LITERATURE REVIEW
Scheduling whether in the field of cloud computing or in the other field, it is a must so as to carry out efficient performance in
particular area of field. Scheduling is done, so as to schedule different hardware, task, resources and users to one another
according to their assigned task/purpose. The main reason behind scheduling is that the number of one of the main module is
less or there is scarcity, so to avoid scarcity problem and to use the available no. of hardware (Virtual machine, servers, PCs
etc) we do scheduling, so that the task can be effectively done in available sources only.
In cloud computing the problem of scheduling falls under the NP-hard problem, which is related to the time complexity to
solve any problem. NP-hard means that the problem cannot be sorted out in a polynomial time complexity optimally by
algorithm. [16]Therefore, many algorithms have been used in the cloud computing environment for the scheduling of its
various functionalities in different forms. The use of algorithm for scheduling in cloud environment is categorized into 3
categories: “General traditional scheduling Algorithm”, “Heuristic optimization algorithm” & “Meta-Heuristics Optimization”.
[13]
Where general algorithm includes various traditional scheduling algorithms like First come first serve (FCFS), Shortest Job
First (SJF), Round Robin and much more.[10] Heuristics include min-min, enhanced min-min algorithm etc.[17] And Meta-
Heuristics include Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO) and Teaching
Learning Based Optimization technique (TLBO), which we are using in this paper. We will discuss some of the methods
proposed using this algorithm in cloud environment:
a) First Come First Serve (FCFS) is the most common scheduling method used. [4] This is the simplest of all method in which
the task which arrives first n the scheduler will be scheduled first rather less any priority preferences it is having. It first
initializes the task and then assign queued task to the n numbers and hence process it. For which the next task have to be
wait long in a queue which concludes in large time consumption.
b) Shortest Job Scheduling Algorithm (SJF) [2] is another general scheduling algorithm used. SJF works on the basis of
execution time which can be taken by the job to be executed. The job which will have lowest time to be taken in execution
is being executed first and is queued first for the execution process, because of which the job with highest execution time
has to wait long in the queue for its execution.[12]
c) Round-Robin Scheduling which assigns particular time to each job to be processed. It is fairer in nature as it allots the time
slices for that particular time only it will process and if it processed after the allotted time then the process is being added
to the tail of the queue. Queue in Round robin works in a circular form.[18]
d) Efficient Task Scheduling [19] is the blend of Longest cloudlet fastest Processing (LCFP) and Shortest cloudlet fastest
Processing (SCFP) algorithm which helps in calculating the completion time of cloudlets and to minimize the time taken
for the completion of the cloudlets.
e) Authors proposed a Service request scheduling algorithm on “Dynamic Priority Scheduling Algorithm (DPSA)”
algorithm on service request scheduling in cloud computing, scheduling is done with help of DPSA and is compared to
SPSA which concluded that DPSA gives better performance than the existing technologies. [4].
f) In the paper [14],Author proposes a scheduling method in cloud environment using Meta heuristic approach i.e. Particle
swarm Optimization (PSO) and compared the enhanced method to other method like Genetic Algorithm (GA),Brute
Force(BF) and FIFO but found that the PSO scheduling was much better than the rest.
g) In this paper [5], the TLBO Comparison with different Meta heuristic approach is provided, and concluded that TLBO is
better than GA, ACO and improvised TLBO method is proposed.

h) This paper , uses Genetic Algorithm(GA) for the dynamic scheduling of cloud data by using the memory usage and the
computation as a factors , the scheduling is performed using GA.[15]
i) The paper [13], gives the detailed comparative analysis of various Meta heuristics algorithm used in the cloud computing
scheduling.
j) The paper describes the use of Min – Min Heuristic to generate the initial population. [20]
k) Various problem of size optimization is solved using TLBO and the TLBO algorithm is explained very efficiently in this
paper. [21]
l) In the paper “Efficient Resource Utilization Algorithm for Service Request Scheduling in Cloud”, EURA algorithm helps to
improve the QoS between the user and the service provider by the help of the utilization ratio which increases the
resource utilization and hence resulting in efficiency rate enhancement [7].
m) In the paper, “Profit –Driven Service Request Scheduling in Clouds”, where sharing through maximum utility and
maximum profit based model is being made. [8].
3. TEACHING LEARNING BASED OPTIMIZATION (TLBO): OVERVIEW
In this paper, we are proposing a scheduling method in Cloud Computing between the user and cloud service provider (Service
Request Scheduling) using a natural meta-heuristic optimization technique: Teaching Learning Based Optimization (TLBO)”.
TLBO is the Meta-heuristics algorithms, which is nature inspired. It works on the teachers-learners based behavior in
classroom. In this technique there is basically two phase:-
1) Teacher Phase
2) Learner Phase
It is the optimization technique based on population, design variable and its possible solution. It is a non-traditional
optimization technique. It takes out best near to optimal solution for a problem.
1) First Phase, Teacher Phase: The rest of the learner gets to learn from the teacher and improve their knowledge. As we
know that the teachers taught the students to increase and explore their knowledge, similarly here in algorithm, teacher
phase, the teacher will try to enhance the performance of learner.
In Teacher’s phase among whole population the ‘best solution’ is calculated and is termed as ‘Teacher’ (Best Solution)(Xteacher)
among all, this is the procedure how we select the teacher among all the population .The Best solution (Xteacher) is being
compared to all the learner’s mean result(Xmean),so that the teacher can enhance the mean performance of all learners
accordingly like Best Solution (Xteacher).Parameters ‘r’ & ‘Tf’ are applied in the TLBO method so as to maintain the staunch
features of TLBO search as it is. Where, ’r’ is the random number & ‘Tf’ is factor, teaching factor (which ranges between 1 - 2
only). [21]
TLBO can be formulated as:
Xnew = Xi + r (Xteacher – Tf. Xmean)
Where, Xi is the existing learner and Xnew is the new updated Xi.
2) Second Phase: Learners Phase: Learners phase, the learner gets to another learner to learn more and more and increase
its knowledge. If our learner is not better than other from whom it learns it will move towards that learner or get itself
updated, and it can be formulated as:
Xnew = Xi + r. (Xi- Xj)
Where f (Xi ) < f (Xj) i.e. Xi is better than Xj (another learner from whom learner learns). [6]
Similarly, If, f (Xj ) < f (Xi ) then:
Xnew = Xi + r. (Xj – Xi)

Finally, if Xnew gives the best among all then it is finally selected. The pseudo code or the algorithm procedure of TLBO (Fig.1):
Fig - 1: TLBO Pseudo Code
4. GENETIC ALGORITHM: BREIF OVERVIEW
Genetic Algorithm (GA) is the biological inspired population generation technique for calculating the optimal solution for the
particular problem. [22] When we use GA, initial population called as chromosome is generated randomly. ‘Fitness Value’ is
also present in this GA as TLBO; so as to select the effective chromosome among the population its fitness value is measured
and then the chromosome is selected through the selection process. After this, on the selected chromosome genetic factors like
crossover and mutation is performed so as to create offspring for the new chromosomes and the process will terminate
depending on the size of the chromosome having the best fit value among rest of the population.
The Pseudo Code of GA is:
Genetic Algorithm
Step 1:- Chromosome generation or Initial Population
generation
Step 2:- Fitness value is calculated
Step 3:- Selection process is performed
Step 4:- Selected Chromosome go through the
Crossover.
Step 5:- Operation of Mutation is done.
Step 6:- If the best fitness value chromosome is
generated terminate else repeat from step 3 again
Fig - 2 : Genetic Algorithm Pseudo Code
TLBO Process
Step 1:- Population Initialization
Step 2:- Fitness Evaluation among the
population
Step 3:- Best Solution or Teacher Selection
Step 4:- If Termination Criteria met, then
Select the Optimal Solution required and Stop,
else Update the population via Learner phase
then again repeat the Step 2 again until the
criteria is meet.
Step 5:- Best Solution as Output

5. RESULT OF TLBO SCHEDULING IN CLOUD ENVIRONMENT
In our proposed method we are taking two features for service request scheduling (user and the cloud service provider) in
cloud environment i.e. firstly, No. of Users (Un) & secondly, the No. of virtual machines (Vn).The scheduling takes place
between these two aspects. As, in TLBO we are having 3 main characters i.e. population, Design variable and its Possible
solution in relation to our proposed method the ‘population’ will be the ‘No. of Users’, ‘Design variable’ will be the ‘Virtual
Machines’.
As per the TLBO formula, we do the ‘Fitness evaluation’, so as to calculate the fitness value among all the population and the
design variable (here, formulated in a matrices form as a random no.). After the fitness evaluation we do the ‘Trainer Selection’
according to which the no. of users can be updated. The ‘Termination Criteria’ will be the maximum no. of iterations to search
for best solution.
If the Criteria are not meet, then again it will update the no. of users (Un) the learner phase followed by the fitness evaluation
again and then the trainer selection and when the termination criteria is meet then select the best optimal combination of Un
& Vn and then the process will be terminated as the best cost or best solution is being calculated.
Implementation: We used MATLAB for this TLBO scheduling. Flowchart of our TLBO Service Request Scheduling is as
follows:
Fig- 3: Flowchart of TLBO Scheduling

When we execute the TLBO scheduling the results occurs as such given in Fig. 4 & Fig.5:
Fig- 4: Initial Iterations
Fig- 5: Final Iteration with Best Fit

After the execution of the TLBO scheduling, the graph of Fitness Vs Iteration is given in Fig-6:
Fig-6: TLBO Scheduling: Fitness Vs Iteration
Now to see our TLBO scheduling is effective or not we compare our TLBO method with that of Genetic Algorithm (GA).The
graph is being plotted with two factors like fitness value against
the iterations. Where X-plane represents the iteration and Y-plane represents the fitness. Blue line represents the Genetic
Algorithm and Red line represents the TLBO. [Fig – 7 & 8]
Fig - 7 : TLBO Vs GA Graph Comparison
In the above Fig.7 graph showed the comparison between the TLBO & GA where for TLBO when initial population is 100, then
in less no. of iterations only we got our best solution where as not in the GA case.
.

Fig-8: TLBO Vs GA Graph Comparison 2
Similarly in the graph Fig - 8 when the initial population is 50, then also we get our best solution much more early then GA as
we can see that TLBO gets saturated before GA that means in less time and is less complex then GA and the best fit in case of
TLBO is much feasible.
6. CONCLUSION AND FUTURE SCOPE
In this paper, we had used TLBO meta-heuristic method so as to improve the service request scheduling part of the cloud
computing, where we had done the scheduling among the no. of users and the virtual machines. By the help of the TLBO
method we calculated the best solution for the selection of the optimal combination of the user and the virtual machine pair
which resulted in reducing the delay, increasing the performance and hence therefore helping to use the optimization
technique rather than the exhaustive algorithm who were complex and as well as had large time complexity to do scheduling
in cloud environment. When compared to GA the TLBO resulted in faster processing as for calculating the best fit or best
solution the GA has to go through several operations whereas in TLBO it’s only about the best solution through two phases
itself in less time as compared to GA.
The future work which could be carried out can be that we can consider much more factors another than no. of users and no.
of virtual machine for service request scheduling using enhanced TLBO method thereby improving the QoS of the cloud
Computing.
REFERENCES
[1] Raja Manish Singh et al, “Task Scheduling in Cloud Computing:Review”,International Journal of Computer Science and
Information Technologies,Vol.5(6),2014.
[2] Kaur Rajeev, Kinger Supriya, “Analysis of Job Scheduling Algorithm in Cloud Computing”,International Journal of
Computer Trends and Technology(IJCTT),March 2014.
[3] Shrivastava Kritika et al, “A Study on Different Services of Cloud Computing Environment”,International Journal for
Research in Applied Sciences & Engineering Technology (IJRASET),December 2016.
[4] Zhongyuan Lee, Ying Wang and Wen Zhou, “A dynamic priority scheduling algorithm on service request scheduling in

Cloud Computing”, International Conference on Electronic & Mechanical Engineering and Information Technology, IEEE,
2011.
[5] Kunjie Yu · XinWang · Zhenlei Wang, “An improved teaching-learning-based optimization algorithm for numerical and
engineering optimization problems”, Springer Science, April 2014.
[6] Bikash Das, T. K. Sengupta, “Economic Load Dispatch Using PSO and TLBO”, Michael Faraday IET International Summit:
MFIIS-2015, September 12 – 13, 2015.
[7] Ramkumar N, Nivethitha S, “Efficient Resource Utilization Algorithm (EURA) for Service Request Scheduling in Cloud”,
International Journal of Engineering and Technology (IJET), 2013.
[8] Young Choon Lee, Chen Wang, Albert Y.Zomaya and Bing Bing Zhou, “Profit –Driven Service Request Scheduling in
Clouds”, IEEE, 2010.
[9] Lokesh kumar et al, “Workflow Scheduling Algorithms in Cloud Environment – A Survey”, RAECS UIET Panjab University,
Chandigarh, 2014.
[10] Vinita Tiwari,Shikha Agrawal,“A Survey on Service Request Scheduling in Cloud Based Architecture”, International Journal
for Scientific Research and Development,2015
[11] Nimisha Singla, Seema Bawa, “Review of Efficient Resources Scheduling Algorithms in Cloud Computing”, International
Journal of Advanced Research in Computer Science and Software Engineering, 2013.
[12] Vinita Tiwari,Shikha Agrawal,”Analysis on Resource Management in Cloud Base Architecture”,International Journal on
Advanced Technology in Engineering & Science,2016.
[13] Kalra Mala,Singh Sarbjeet,“A review of meta-heuristic scheduling Techniques in cloud computing”,Egyptian Informatic
Journal,2015,pg.275-295.
[14] Syed Hasan Adil et al, “Cloud Task Scheduling Using Nature Inspired Meta-Heuristic Algorithm ”, International Conference
on Open Source Systems and Technologies (ICOSST),2015.
[15] A.Kaleeswaran et al,“Dynamic Scheduling Of Data Using Genetic Algorithm In Cloud Computing”,International Journal of
Advances in Engineering & Technology, Jan. 2013.
[16] M.R.Garey et al,Computer and interactability:a guide to the theory of NP-Completeness,1979.
[17] Vignesh V et al, “Resource Management and scheduling in Cloud Environment”,International Journal of Scientific and
Reasearch Publication,Volume 3 , Issue 6,June 2013.
[18] Jangra Dr.Ajay et al, “Scheduling Optimization in Cloud Computing”,International journal of Advanced Research in
Computer Science and Software Engineering,April 2013
[19] Sharma Neha et al, “A Survey on Heuristic Approach for Task Scheduling in Cloud Computing”, International Journal of
Advanced Research in Computer Science, IJARCS , Volume 8, No. 3, March – April 2017
[20] Kaur K,Chhabra A, “Heuristics Based Genetic Algorithm for scheduling static tasks in homogenous parallel
system”,International Journal of Computer Science Security,183-98.
[21] Baghlani.A,Makiabadi M.H,“Teaching & Learning Based Optimization Algorithm for shape and size Optimization of Truss
Structures with Dynamic Frequency Constraints”,IJST,Vol.37,2013.
[22] Agarwal Mohit ,Dr. Gur Mauj Saran Srivastava, “ A Genetic Algorithm inspired task scheduling in Cloud Computing” ,
International Conference on Computing, Communication and Automation (ICCCA),IEEE,2016

Service Request Scheduling in Cloud Computing using Meta-Heuristic Technique: Teaching Learning Based Optimization (TLBO)

Recommended

More Related Content

What's hot (16)

Similar to Service Request Scheduling in Cloud Computing using Meta-Heuristic Technique: Teaching Learning Based Optimization (TLBO) (20)

More from IRJET Journal (20)

Recently uploaded (20)

Service Request Scheduling in Cloud Computing using Meta-Heuristic Technique: Teaching Learning Based Optimization (TLBO)