International Journal of Computer Science Engineering

and Information Technology Research (IJCSEITR)

ISSN(P): 2249-6831; ISSN(E): 2249-7943
Vol. 4, Issue 6, Dec 2014, 35-40
TJPRC Pvt. Ltd.

AN EFFECTIVE GUIDANCE METHOD FOR WEB SERVICE SELECTION USING PSO

AND MA
VIJAYALAXMI S JEURE & Y. C. KULKARNI
Bharati Vidyapeeth Deemed Universiy, College of Engineering, Pune, Maharashtra, India

ABSTRACT
In service oriented architecture there may be many web services providing similar providing similar properties,
effects, capabilities, and interfaces. Hence selecting a service which is as per the user requirement is a difficult task.
Differentiations among the competing services are obtained through quality of service attributes. This allows the
user/consumer to select the best service as per his/her requirement. This paper has considered this issue and discusses
about the use of these quality of service attributes which are nothing but the non functional parameters for guiding the user
in selecting the service according to his/her requirement. The non functional parameters which are considered for this are
nothing but response time, reliability, and execution time of the competing web services. For this in the paper proposed
two algorithms particle swarm optimization and memetic algorithm. These two algorithms are used for above non
functional parameters. Using graphical representation the comparison of co`11`mpeting web services for their respective
non functional parameters is shown. Both approaches are compared for these three non functional parameters.

KEYWORDS: Particle Swarm Optimization, Memetic Algorithm, Web Service, Qos

INTRODUCTION
Software systems which are designed to support interoperable machine to machine interaction over the network
are called as web services. Consumers and business processes use these web services. These web services are the
implementation of well defined business processes and functionalities. The creation and deployment of these web services
is defined through some architecture. This architecture has interaction between three roles. These are the service providers,
service registry and service consumer.
Service provider is the owner of the service, who hosts an access to the service. Service registry is the searchable
registry, which contains the description of the service as specified by the service provider. Service requestor is the one who
accesses the service for his own purpose. The interaction of all these three entities in web service involves three functions
as: publish, find and bind. Publish operation is used by the service provider to publish the service description.
Find operation is used by the service requestor in order to find the service. The bind operation is used by the service
requestor to invoke or initiate an interaction with the service at runtime.
There are many web services providing similar providing similar properties like interfaces, even they are having
similar effects and capabilities. Selecting a best service as per the clients requirement is the difficult task. Current service
oriented architecture considers the functional properties of the web service. What is service does is specified by the
functional properties, but how the service does it is specified by non-functional properties. The non functional properties
are used to measure the quality of service.

www.tjprc.org

editor@tjprc.org

36

Vijayalaxmi S Jeure & Y. C. Kulkarni

Quality of Service can be defined as: the capability to respond to the requirements a client and to fulfil these
needs with the best condition specified from client. The non-functional properties describe the service and including its
performance. The performance specifies how and in which condition the service is executed. Few non-functional
properties can be defined as:

Availability: It can be calculated as (1 Down Time)) (Measurement Time). The Measurement Time can be
adapted to the application.

Accuracy: Error rate that produces a service. It is calculated as the number of errors divided by the total number
of executions.

Execution Time: Measures precisely the time the provider takes to execute a request.

Reliability: Probability with which the provider correctly answers a request within a maximum expected time. It
is measured as the number of success request divided by the number of request.

Robustness: Establishes the capacity of the system to give an accurate answer in presence of invalid, incomplete
or conflicting inputs. These can be estimated at the provider side or tested and making an average from the last
experiences.

Security: Comprises other non-functional properties (i.e. data encryption, authentification). It evaluates the
security in the transaction who assures the ACID properties (Atomicity, Consistency, Isolation and Durability).
This value can be presented by the provider separately or as an aggregation of the different qualities.

Throughput: Measures the maximum network capacity the client and the provider can get to communicate. It is
time dependent, so if a quality reservation is not possible on the network, the throughput would be either the
maximum throughput registered or the average.

WEB SERVICE SELECTION ALGORITHMS

The problem of service selection on the web consists of having an efficient algorithm that can match multiple
service consumers and service providers efficiently, while optimizing multiple objectives (QoS parameters). There are
many clients requesting similar services and assignment process of the service consumer and the service provider should
be satisfied. Please note that one consumer can only be matched with one provider [1]. Since we have several service
consumers and equally numbered service providers. Using Memetic Algorithm (MA) and Particle Swarm Optimization
(PSO) approach our aim is to match the consumer-provider pairs as closely as possible. In the further discussion we will
talk about these two algorithms for web service selection.

MEMETIC ALGORITHM
Evolutionary algorithm with local search technique forms the memetic algorithm. For finding high quality
solutions to optimization problems memetic algorithms are very much effective and efficient. Memes in. memetic are very
much similar to genes in genetic algorithm. Genes are predefined. They already contain the set of values for optimization
problem. Memes are not pre defined, rather they are derived by analyzing the current information. Memes represent current
position and compares it with objective function accordingly they keep on updating to reach to the goal position. Memetic
Algorithm is expansion of traditional Genetic Algorithm. It reduces the premature convergence in local search itself.
Impact Factor (JCC): 6.8785

Index Copernicus Value (ICV): 3.0

37

An Effective Guidance Method for Web Service Selection Using PSO and MA

The Algorithm Can Be Given As:

Step 1: generates an initial population of random solutions.
Step 2: improve these solutions through local search method.
Step 3: formulate it in chromosome representation.
Step 4: derive fitness value for each chromosome.
Step 5: apply crossover and mutation operators exchanging the information in the former solution and mutated.
Step 6: apply local search procedure to improve the obtained solution that is child solution.
Step 7: worst solution is improved by child solution then this solution is replaced.
Step 8: Choose highest fitness value solution

PARTICAL SWARM OPTIMIZATION

Particle Swarm Intelligent System (PSIS) developed by Dr. Everhart and Dr. Kennedy in 1995. It is an
optimization technique known as Particle Swarm Optimization (PSO) which is derived by the flocking of birds. The main
idea is to simulate the unpredictable choreography of a bird flock. Based on the observation of the evolution of the
algorithm, it has been realized that the conceptual model is in fact an optimizer. It was developed for real-valued problems.
It can also be extended to cover binary and discrete problems.
The Algorithm for Particle Swarm Optimization Given As:
Step 1: initialize an array of population of particles with random positions and velocities in D dimensions in the
problem space.
Step 2: evaluate fitness function in D variables for each particle.
Step 3: compare each particles fitness with its pbest.
If current value better than pbest
Then update current value to pbest and let its location correspond to the current location in D-dimensional space
Step 4: evaluate this fitness with the populations overall previous best.
If the current value better than gbest
Then update current value to gbest and save it as gbest to the current particles array index and value.
Step 5: modify velocity and position according to equations as given

(1)

(2)
Step 6: it the required criteria not achieved go to step2 else stop process.

www.tjprc.org

editor@tjprc.org

38

Vijayalaxmi S Jeure & Y. C. Kulkarni

EXPERIMENTS AND RESULTS

The two approaches described above are implemented. For the implementation a banking system is considered.
Two web services having similar functionalities and interfaces are created i.e. webservice1 and webservice2. These two
web services are having the functionalities for different banking transactions as: creating and validating an account, credit
fund, debit fund, getting a view of all the previous and current transactions.
Webservice1 is of connected architecture i.e. connection to the database must be open to access data retrieved
from the database. Webservice2 is of disconnected architecture i.e. data retrieved from database can be accessed even
when connection to database was closed is called as disconnected architecture. Disconnected architecture of ADO.net was
built on classes. These classes are like data adapter, connection, dataset, data view and common builder.
The non-functional parameters considered for this are response time, execution time and reliability of web
services.

Response Time: when the user is giving input to the software system whatever time the system takes to respond
to the user for the given input.

Execution Time: Measures precisely the time the provider takes to execute a request.

Reliability: Probability of answering correctly the given request by the provider within maximum expected time.
It is measured as the number of success request divided by the number of request.
Database design contains three tables. First table is there to maintain the information of customers for different

transactions. Second table maintaining all the transaction details of the customers like transaction type, transaction amount,
transaction date etc. Third table is an important table in which all the webservic usage information is stored which contains
the execution time, response time and error information of every transaction of the web service.
The above two algorithms are executed on15 transactions. Firstly the service consumer is supposed to ask about
the respective response time accordingly the graph is displayed to the user showing the comparison of different
transactions(as stated above) of a web services according to memetic and particle swarm optimization algorithm.

Figure 1: Response Time According to PSO and MA for Different Web Service Transactions
The blue bar is for PSO algorithm and brown bar is for memetic algorithm. One can determine that response time
of a transaction 1 according to PSO is greater than that of memetic algorithm and so on.

Impact Factor (JCC): 6.8785

Index Copernicus Value (ICV): 3.0

39

An Effective Guidance Method for Web Service Selection Using PSO and MA

The second graph the execution time of each web service is shown. Here response time is nothing but the
execution time of the web services each transaction.

Figure 2: Execution Time According to PSO and MA for Different Web Service Transactions
The blue bar is for PSO algorithm and brown bar is for memetic algorithm. One can determine that response time
of a transaction 1 according to PSO is greater than that of memetic algorithm and so on.
The third graph is about the reliability of web services.

Figure 3: Error Report of Web Service 1 and Web Service 2

CONCLUSIONS
This paper describes about how non-functional properties are important for web service selection.
Two approaches are used for web service selection one is particle swarm optimization and other is memetic algorithm.
These algorithms are implemented to match the requirement of service consumer with the service provider as per his/her
own requirement of functionality. Depending on how many transactions are executed of the web service it is going to
affect the response time and execution time of the web service. We can find PSO performance is better than memetic
because in memetic extra time is required to perform local search.

www.tjprc.org

editor@tjprc.org

40

Vijayalaxmi S Jeure & Y. C. Kulkarni

REFERENCES
1.

Simone A. Ludwig. (2011). Memetic Algorithm for Web Service Selection. Proceedings of the 3rd workshop on
Biologically inspired algorithms for distributed systems international conference on automatic communication and
computing.

2.

Shang-Chia Liu & Sung-Shun Weng. (2012). Applying Genetic Algorithm to Select Web services Based on
Workflow Quality of Service. Journal of Electronic Commerce Research VOL 13 NO 2.

3.

Taher L. El Khatib H. (2005). A framework and QoS matchmaking algorithm for dynamic web services selection.
Proceedings of the 2nd International Conference on Innovations in Information Technology (IIT05).

4.

Holland J. H. (1975). Adaptation in Natural and Artificial Systems. University of Michigan Press, Ann Arbor.

5.

Wolpert D, Macready. (1997). No free lunch theorems for optimization. IEEE Trans. Evol. Comput, vol. 1, no. 1,
pp.67-82.

6.

Dawkins R. (1976). The Selfish Gene. New York: Oxford Univ. Press.

7.

Krasnogor N. & Smith, J. (2005). A Tutorial for Competent Memetic Algorithms: Model, Taxonomy, and Design
Issues. IEEE Trans. On Evolutionary Computation, vol. 9, no. 5.8.

8.

Vazquez M. & Whitley L.(2000). A hybrid genetic algorithm for the quadratic assignment problem. Proceeding
on Genetic Evolution Computing Conference. pp. 135-142.

9.

Ku K. Mak. (1998). Empirical analysis of the factors that affect the Baldwin effect. Lecture Notes in Computer
Science. Parallel Problem Solving From Nature, pp. 481-490.

10. M. C. Frdrique & Claudia Catalina, Evaluation. Framework for Quality of Service in Web Services:
implementation in a pervasive environment. Master Thesis Master Research in Informatics Specialized in
Technologies of Information & Web.
11. J. O'sullivan. D. Edmond and A. T. Hofstede. (2002). What's in a service? Towards accurate description of
non-functional service properties. Distributed and Parallel Databases Journal.
12. N. P. Padhy. (2013). Artificial Intelligence and Intelligent Systems. Oxford University Press, 12th impression.
13. WSMO Deliverable D28.4 V0.1 (2006). Non Functional Properties of Web Services WSMO Working Draft.
14. Mona niyakan Lahiji. (2012). Particle Swarm Optimization (PSO) Algorithm.CNB scholar journals. Journal of
Biology and today's world volume 1, issue 1.
15. C. Rajeswary*. A survey on Efficient Evolutionary algorithms for Web Service Selection. IJMIE Volume 2,
Issue 9

Impact Factor (JCC): 6.8785

Index Copernicus Value (ICV): 3.0