14
Anveshana Search for the Right Service Dr Mydhili K. Nair Associate Professor, Dept. of ISE, MSRIT Varun M Deshpande [email protected] PhD Student, Dept. of CSE, Jain University & Software QA Engineer, McAfee Software (India) Pvt. Ltd. 1 I2CT 2014

QoS based service ranking and selection - Anveshana

Embed Size (px)

Citation preview

Page 1: QoS based service ranking and selection - Anveshana

Anveshana

Search for the Right Service

Dr Mydhili K. Nair

Associate Professor,

Dept. of ISE, MSRIT

Varun M Deshpande

[email protected]

PhD Student,

Dept. of CSE, Jain University

&

Software QA Engineer,

McAfee Software (India) Pvt. Ltd.

1

I2CT 2014

Page 2: QoS based service ranking and selection - Anveshana

Agenda Background

Problem Domain

Literature Survey

Design and Analysis

Results

Conclusion and Future Scope

Bibliography

2

Page 3: QoS based service ranking and selection - Anveshana

Background

Service Delivery

Requirements

CustomerService Provider

Service Model – Customer requests for service. Service provider provides the same

But each customer is different. They have different needs.

3

Page 4: QoS based service ranking and selection - Anveshana

Problem Domain 4

“Does Quality of a Product/Service

have an impact on its cost (Do you

think that better quality product

costs more)?”

“Do you feel each end user has

different quality requirements?”

“Do you always choose higher

quality product than what you

need/afford (Do you prefer ‘Best’

product in the market even if it is

out of budget)?”

Page 5: QoS based service ranking and selection - Anveshana

Literature Survey

Mohammad Alrifai [1] et al., in their work define QoS-based service selection

problem as “finding the best component services that satisfy end-to-end quality

requirements.” They model this problem as a multidimensional multi choice 0-1

knapsack problem. They discuss shortcomings of solutions based on linear

programming techniques for large data sets.

Guofeng Chang [2], in his work proposed a genetic algorithm based approach to

solve web service selection problem. His idea was based on traditional evolutionary

ideas of natural selection and genetics. The core philosophy that is focused on is

“Survival of the Fittest.” This means that only the best services are preferred to

during service selection while others are phased out of the competition.

Yilei Zhang et al., [3] recognized finding the desired web service among the

available repository an emergent and challenging research problem. Their search

framework considers both functional and non-functional attributes of publically

available web services which have similarities to user’s request. They further

propose 3 searching strategies which can autonomously fetch the right web service

suitable to the end user.

5

Page 6: QoS based service ranking and selection - Anveshana

Literature Survey

Maheshwari et al. [4] discussed the importance of QoS parameters on top of ranking

provided by users of web services. Their architecture included OWL-S convertor

which converts syntactically described web service into a semantic web service.

Semantic repository which contains the list of advertised web services in OWL-S

format, a QoS broker equipped with match making algorithm.

Tajudeen Adeyemi Ajao et al. [5] have very recently published work related to

current research domain. They are of opinion that identifying the optimal service is

still an active research area. They propose usage of QoS-based Filtering, Ranking

and Selection Algorithm for this purpose. They adopt a similar approach as in [3] and

filter out the services which fall short of the requirements of end user. Later they rank

each service and identify the service with highest score as the optimal web service.

Zibin Zheng et al. [6] in their work lay a firm mathematical model for addressing the

problem of web service recommendation when there are abundant web services

available. They discuss system architecture to solve the problem. They have provided

mathematical representation for key aspects like similarity computation and web

service recommendation.

6

Page 7: QoS based service ranking and selection - Anveshana

Design & Analysis 7

Customer QoS Request ObjectAdvertized parameters of Service

Workflow of Service Selection

Page 8: QoS based service ranking and selection - Anveshana

Design & Analysis Cont. 8

Steps during execution of Anveshana

Page 9: QoS based service ranking and selection - Anveshana

Algorithm Step 1: Calculation of Individual Variance

Let Linear Relevance be denoted as “L”. LetRequested value be denoted as

“R”. Let Advertised value be denoted as “A”. Normalization criterian is

denoted as Range. In our case, it is taken as “10”. This may be altered

based on requirements of a particular domain.

For each QoS parameter “i” , Li = |Ri – Ai| / Range

Step 2: Priority Consideration

Suppose there are “n” QoS parameters, priority for any particular QoS

parameter would be unique integer between 1-n. Weight of each QoS

parameter is calculated using below formula.

Let priority values be denoted by “P”. Let weight of a QoS parameter be

denoted by “W”.

For each QoS parameter “i”, Wi = [ n - Pi + 1 ] / ∑n

9

Page 10: QoS based service ranking and selection - Anveshana

Algorithm contd. Step 3: Calculation of Individual Relevance

Individual relevance is calculated for each parameter seperately. This is

done using below formula. This indicates that lesser the variance between

advertised value and requested value, more is the relevance score.

For each QoS parameter “i”, Ii = 1 – [Li * Wi]

Step 4: Calculation of Total Relevance

This is technically, the final step of Anveshana. In this step, each of the

Individual Relevances are integrated to arrive at the final score. Formula for

calculation Total Relevance is given below. Let Total Relevance be denoted

as “T”.

T = ( ∑ Ii ) / n

10

Page 11: QoS based service ranking and selection - Anveshana

Results 11

User requestRanked services for user request

Feedback on usefulness of

Anveshana

• Core aim of this project is to move

towards searching for the “Right”

service than “Best” service.

• This approach is beneficial for end

users who are looking for services

which match their custom

requirements rather that best ( and

hence more costly ) service.

Page 12: QoS based service ranking and selection - Anveshana

Conclusions and Future Scope

We conducted an online survey to understand end user

perspective of QoS based service selection

Our approach with Anveshana was to help end users get the

“Right” services based on their custom requirements rather than

“Best in class” services.

Need to use real world data and compare with other algorithms for

performance and correctness

This approach can be implemented for customer driven SLA

management to find solutions to some of security and privacy

issues in cloud computing

Other real world applications of Anveshana are search engines for

e-commerce website, web service discovery and composition,

game consoles etc.

12

Page 13: QoS based service ranking and selection - Anveshana

Bibiliography Mohammad Alrifai, Thomas Risse, Perter Dolog and Wolfgang Nejdl, “A Scalable Approach for

QoS-based Web Service Selection,” Service-Oriented Computing – ICSOC 2008 International Workshops, p 190-199

Guofeng Chang, “QoS-Based Web Service Selection Approach”, Software Engineering and Knowledge Engineering: Theory and Practice, 2012, Volume 2, p 887-892

Yilei Zhang, Zibin Zheng, Lyu, M.R., “WSExpress: A QoS-aware Search Engine for Web Services”, Web Services (ICWS), 2010 IEEE International Conference

Maheswari, S. and G.R. Karpagam, “QoS Based Efficient Web Service Selection,” European Journal of Scientific Research, 2011. 66(3): p. 428-440.

Tajudeen Adeyemi Ajao, Safaai Deris, Isiaka Adekunle Obasa, “QoS-based Web Service Selection Using Filtering, Ranking and Selection Algorithm,” International Journal of Scientific & Engineering Research, Volume 4, Issue 7, 2013

Zibin Zheng, Hao Ma, Michael R Lyu and Irwin King, “WSRec: A Collaborative Filtering Based Web Service Recommender System,” 2009 IEEE International Conference on Web Services

Varun M Deshpande, Dr. Mydhili K. Nair, Balaji Sowndararajan, Customer Driven SLA in Cloud Based Systems, In Proceedings published by Elsevier of International Conference of Emerging Computations and Information Technologies, SIT, Tumkur, Karnataka (India), 22-23 November, 2013, pp 508-518

Online Survey on QoS based Service Selection Published Results: https://docs.google.com/forms/d/1VFZ8m7pPsaCSKQplOLz-auOTSd2AwYHV-tsSpMkfNU8/viewanalytics#start=publishanalytics

13

Page 14: QoS based service ranking and selection - Anveshana

Thank You

For Your Time

14

I2CT 2014

[email protected]