European Journal of Scientific Research ISSN 1450-216X Vol.66 No.3 (2011), pp. 428-440 © EuroJournals Publishing, Inc. 2011 http://www.europeanjournalofscientificresearch.com
QoS Based Efficient Web Service Selection S. Maheswari Assistant Professor, CSE Dept, PSG College of Technology Coimbatore-641004, India E-mail:
[email protected] G. R. Karpagam Professor, CSE Dept, PSG College of Technology Coimbatore-641004, India E-mail:
[email protected] Abstract QoS based Web Service Selection plays an essential role because consumers want to use the services that meets their requirements. Currently web service selection is based on reputation (Majithia A.et.al., Maximilien) which is calculated based on the ratings provided by the user for each web service. As the ratings alone cannot describe the quality of web service, other QoS parameters should also be considered. The seven QoS parameters considered in this paper are execution time, response time, throughput, scalability, reputation, accessibility and availability. The objective of the framework is to provide QoS based Semantic Web Service Selection. The framework consists of four major components namely OWL-S converter, Semantic Repository, QoS Broker and Matchmaker. OWL-S converter converts the syntactically described web service into a semantic web service. The Semantic repository contains the advertisements of the web services in OWL-S format. QoS broker assigns and stores the rank of all the accessed services based on the Web service Relevancy factor. Matchmaker identifies a set of services that satisfies the client’s functional and non-functional requirements.
Keywords: Semantic web services, OWL-S, Quality of service, WsRF.
1. Introduction Quality of service (Sodki Chaari) plays an important role in automatic web service selection. It is mainly used to establish valid and reliable web service and identity the best web service systematically from a set of functionally similar services. QoS parameter gives requestor assurance and confidence to use the services. Requestor aim to experience a good service performance e.g. low waiting time, high reliability and availability to successfully use services. The quality of a web service can be broadly categorized into performance, dependability and security (Yao Wang, et.al.). This paper considers parameters classified under first two categories. Performance based quality of services are Response time, Execution time, Throughput. Dependability based quality of services are Scalability, Availability, Accessibility. 1. Response time: The maximum time that elapses from the moment that a web service receives a SOAP request until it produces the corresponding SOAP response. It is calculated as RT = T1 − T2 . (1)
QoS Based Efficient Web Service Selection
429
Where T1=Time at which web service produces soap response T2=Time at which web service receives soap request. 2. Execution time: The execution time measures the expected delay between the moment when a request is sent and the moment when the result are received. It is denoted by qdu ( s, op ) = Tprocess ( s, op ) + Ttrans ( s, op ) . (2) Where Tprocess is the processing time (advertised by service provider) Ttrans (s,op) is the transmission time calculated based on past observations of transmission time n T ( s , op ) as Ttrans ( s, op ) = ∑ i =1 i n 3. Throughput: The number of Web service requests R for an operation o that can be processed by a service S within a given period of time. #R . tp ( s, o ) = (3) ( in sec ) . time period Where S service within a given period of time #R number of web service request 4. Scalability: A Web service that is scalable has the ability to not get overloaded by a massive number of parallel requests. It is calculated as trt . (4) SC = trt ( throughput ) Where trt(Throughput) is the round trip time which is evaluated during the throughput test. 5. Reputation: The reputation of a service is the measure of its trustworthiness. It mainly depends on the end users experience of the service. Different end users may have different opinions on the same service. The value of the reputation is defined as the average ranking given to the service by the end users. It is calculated as n R (5) qreputation = ∑ i . i =1 n Where Ri is the ranking of end user for the service n is the no of times the service has been ranked 6. Availability: Availability of the web service is the probability of the service is accessible. It is calculated using the following expression downtime AV = 1 − (6) uptime The downtime and uptime are measured in minutes 7. Accessibility: It is the capability of serving the Web Service request. The Web service might be available but not accessible because of a high volume of requests. Accessibility can be represented by the following formula: downtime Paccessibility = 1 − (7) . uptime time is measured in minutes. In section2, this paper discusses short overview of related work. Section 3 describes the proposed QoS based web service selection Framework. Section 4 presents the experimental results of QoS based web service selection. Finally, this paper is concluded in section 5.
430
S. Maheswari and G. R. Karpagam
2. Related Work Most of the researchers focused on either QoS-based service discovery or reputation management systems. This section presents a review of various methods that meet this challenge. Majithia Shaikhali, A., Rana, O. and Walker, D. (2004) have proposed a framework for reputation based semantic service discovery in which ratings of services in different contexts are collected from service consumers by a reputation management system. Serhani M.A.,Dssouli,R.,Hafid,A. and Sahraoui,H.(2005) have presented a web service selection architecture. The paper verifies QoS based on certification approach but lacks in specifying QoS parameters. Xu (2007) presented a reputationenhanced QoS-based Web Services discovery model. This model only considers the reputation management system. Rajendran and Balasubramanie, P.(2009) have proposed a framework for agent based web services discovery with QoS to select the suitable web service that satisfies the client’s requirements in syntactic model not in semantic model. Xia Wang proposed a QoS-aware selection model for semantic web service selection, QoS information selected in WSMO model. Stefania Galizia presented a prototype of a trust-based selection in IRS-III, based on WSTO model. In this paper used for OWL-S model. Maximilien and Singh have proposed computation details of the reputation systems. S.Maheswari and Dr.G.R.Karpagam (2010) proposed Reputation based Web Service selection. M.Aarthilakshmi, R.Suganyakala, Dr.G.R.Karpagam and S.Maheswari (2011) the authors proposed Ontology Based QoS Driven Web Service Discovery. They analyzed interdependencies among QoS parameters.Yutu Liu provides the QoS computation by using normalization techniques in syntactic model not in semantic model. In this paper same normalization techniques used for semantic model. This paper proposes a Framework to rank and select semantic web service based on seven QoS parameters.
3. Qos Based Web Service Selection Framework The objective of the Framework is to select the web service consider the seven QoS parameters namely Response time, Execution time, Throughput, Reputation, Scalability, Availability and Accessibility. The QoS based web service selection Framework is shown in Fig.1.
3.1. OWL-S Converter OWL-S converter converts the syntactically described web service into a semantic web service by converting the WSDL into OWL-S. The web service provider publishes web services as Web Service Description Language (WSDL) which syntactically describes a web service. It is converted in to OWLS description by using OWL-S plug-in available with Protégé. The semantic description of a web service consists of Service Profile, Process Model and Service Grounding. Service profile describes about what the service does, Process Model says about the how the service works and service grounding provides details about how to access the service.
QoS Based Efficient Web Service Selection
431
Figure 1: QoS Based service selection Framework
Semantic Repository
Access Get and Update the QoS information
OWLConverter (OWL-S)
Matchmaker
Requestor gets best service
Request the service
Web Service Requestor
QoS broker (WsRF Calculation)
Web service Provider
Fig. 2 depicts the interaction among the major components in the Framework by considering the following scenario. Step1: Initially Web service provider places the service in semantic registry along with the QoS values. Step2: QoS broker gets parameters from Repository. These parameters are used to calculate WsRF and calculated values are updated in semantic repository Step3: The web service Requestor sends a request to the Matchmaker. It finds a set of services that match the client’s functional and non-functional requirements. Step4: The web service Requestor gets the best web service from the Matchmaker. Step5: The user uses the required service and provides a feedback about that web service to the Repository. Step6: The QoS broker collects the feedback and updates the reputation parameter in Repository.
432
S. Maheswari and G. R. Karpagam Figure 2: Interaction among framework components
3.2. Semantic Repository The Semantic repository contains the advertisements of the services in OWL-S format. It maintains set of files in the form of folders. For each service a separate folder is maintained that consists of OWL-S and OWL-Q files. OWL-Q keeps track of the Quality of Service as specified by the vendor, and it is updated with the feedback provided by the users. The QOS parameters are retrieved from the OWL-Q files and are used for calculation of Web Service Relevancy Factor (WSRF). This factor is used to rank the web services and provide the results based on their relative rank. Whenever a user requests for a service the request will be matched against these advertisements.
3.3. QoS Broker The QoS broker is assumed as a trusted third party between the users and the service providers. It plays an important role in the QoS based selection of semantic web services. The user uses the required service and provides a feedback about that web service to the QoS broker. The QoS broker collects the feedback and updates the reputation parameter in QoS ontology. Generally, QoS broker assigns and stores the rank for all accessed services according to this reputation value. When the user provides weights for each QoS parameter in the service request, Web service relevancy factor is calculated according to those weights. Before calculating the web service relevancy factor all QoS parameters are normalized. The normalization provides a uniform distribution of QoS values that have different units. QoS values are normalized by following the flow chart given in Fig.3. First Matrix E is constructed with rows representing Web services and columns representing Quality of services. To normalize matrix E in WsRF, two arrays are defined. The first array is N = {n1, n2, ... nm} with 1 ≤ j ≤ m. The
QoS Based Efficient Web Service Selection
433
value of nj becomes 1 when the increase of qi,j benefits the service requester while the value becomes 0 when the decrease of qi,j benefits the service requester.. The second array is C = {c1, c2... cm} here cj is a constant which sets the maximum normalized value. The formulas used for normalization
(8)
• •
The result of the computation vi,j is stored in a matrix ‘Q’. The values of ‘Q’ are stored in E which is used for calculating WsRF.
3.3.1. Web Service Relevancy Factor The services looked up by the web service broker for a user’s request have to be ranked based on the QoS factors of the services. This section describes a way to rank the QoS factors of the web services retrieved. This uses a factor called Web service Relevancy Factor (WsRF) (Eyhab Al-masri). The service with highest WsRF value is the most relevant service chosen for the user.QoS based web service discovery is done through normalization algorithm (Yutu Liu ).The value of WsRF is calculated as follows. Assuming that there is a set of Web services that share the same functionality such that WS (WS = {ws1, ws2, ws3 …wsi}) and QoS attributes such that P (P = {p1, p2, p3…pj}), a QoS-based computational algorithm determines which wsi is relevant based on QoS constraints provided by the client. Using j criteria for evaluating a given Web service, we obtain the following WsRF matrix in which each row represents a single Web service wsi, while each column represents a single QoS parameter Pj. Q1,2 Q1, j (9) E= Qi ,1 Qi , j
434
S. Maheswari and G. R. Karpagam Figure 3: QoS normalization START
INPUT
n j==1
Y
N
qi,j ≠ 0
1/n
N (1/n
qi,j / (1/n ) ≥ cj
N
)/qi,j Y
Y
Y
Y
Vi,j =cj
Vi,j =cj
qi,j / (1/n )