Evaluation of QoS based Web- Service Selection Techniques for Service Composition

M. Sathya, M. Swarnamugi, P. Dhavachelvan & G. Sureshkumar
International Journal of Software Engineering (IJSE), Volume (1): Issue (5) 73
Evaluation of QoS Based Web- Service Selection Techniques for
Service Composition
M. Sathya satsubithra@gmail.com
Department of Computer Science
School of Engineering Pondicherry University
Puducherry- 605014 India
M. Swarnamugi swathidevan@gmail.com
P. Dhavachelvan dhavachelvan@gmail.com
G. Sureshkumar mgsureshkumar@gmail.com
Pondicherry University Karaikal Centre, Karaikal
Abstract
In service oriented computing, services are the basic construct that aims to
facilitate building of business application in a more flexible and interoperable
manner for enterprise collaboration. To satisfy the needs of clients and to adapt
to changing needs, service composition is performed to compose the various
capabilities of available services. With the proliferation of services offering similar
functionalities around the web, the task of service selection for service
composition is complicated. It is vital to provide service consumers with facilities
for selecting required web services according to their non-functional
characteristics or quality of service (QoS). The objective of this paper presents
the exploration of various techniques of Quality of Service based Service
Selection (QSS) approach in the literature. To evaluate the service selection
process, a number of criteria for QSS approach have been identified and
presented in this paper.
Keywords: Web Service Selection, Service Composition, Web Semantics, Quality of Service.
1. INTRODUCTION
Service-Oriented Computing (SOC) is an upcoming organizational model that allows assembling
independent distributed services into complex ones. Services are autonomous, platform-
independent computational entities that can be used in a platform independent and programming
language independent way. The application functionality of SOC as services relies on its
dynamism. That is, it has the capability to dynamically assemble complex services for developing

massively distributed, interoperable, evolvable systems. Services are most often built in a way
that is independent of the context in which they are used. This means that the service provider
and the consumers are loosely coupled. Key to this concept is the service-oriented architecture
(SOA).
The Service Oriented Architecture (SOA) is a type of “software architecture that represents
software functionality as services over the network” [1]. Web Services are the predominant
implementation platform for SOA and it uses a set of standards, SOAP, UDDI, WSDL, which
enable a flexible way for applications to interact with each other over networks. Simple Object
Access Protocol (SOAP) is a standard protocol that allows network communication between
services. The easiest way to publish a web service is to use a SOAP container. When a software
component is published as a web service, any SOAP-enabled client that knows the network
address of the web service can send a SOAP request and get a SOAP response. To get the
message information, SOAP- enabled clients read a WSDL file that describes the web service.
Once the Web Services Description Languages (WSDL) file is read, the client can start sending
SOAP messages to the web service. WSDL describes what a web service can actually do, where
it resides, and how to invoke it. Universal Description Discovery and Integration (UDDI) is a
standard that allows information about businesses and services to be electronically published,
queried and stored. Published information is stored into one or more UDDI registries, which can
be accessed through SOAP.
All these standards are XML-based (Extensible Markup Language), which allows applications to
interact with each other over networks, no matter what languages and platforms they are using.
The two features, self-description and language-platform-independence, distinguish web services
from other distributed computing technologies, like CORBA (Common Object Request Broker
Architecture) and DCOM (Distributed Component Object Model).
Research in web services includes many challenging areas starting from service publication to
service mining. The most vital among them is web service composition. Web service composition
is needed when a client’s complex request cannot be answered by single service, but by
combining or composing various functionalities of available services or more than one services.
Composition involves three different issues [2]. The first, called selection of service is concerned
with selecting suitable services to composite that satisfy the user requirement. The second, called
composition synthesis is concerned with synthesizing a specification of how to coordinate the
component services to fulfill the client request. The third issue, called as orchestration is
concerned with achieving the coordination among services by executing the specification
produced by the composition synthesis.
This paper presents a study of one service selection approach called QoS based service
selection for service composition. The paper is organized as follows: Section II describes the
overview of service selection approaches. In Section III, the specifications of QoS based service
selection and the various techniques of QSS are presented. Section IV analysis the evaluation
criteria of QoS based service selection approach and compare the various techniques of QSS.
Finally section V concludes with discussion and highlights new challenges need to be addressed.
2. OVERVIEW OF SERVICE SELECTION APPROACHES
The current semantic web services architecture focus on solving the issues of service discovery,
service selection and service composition. Service discovery is the process of finding or locating
service implementations that meet a specified condition. In the same way, service selection is a
process that deals with choosing a service implementation from the located services. From this, it
is clearly seen that service discovery is a prerequisite requirement for selection process, but
selection is the main problem that needs to be addressed for retrieving Web services
successfully. For any service selection approaches the basic requirements include: Customer

service requirement, Service offerings by the service provider and aggregating the evaluation
results.
2.1 Customer-Service Requirement
The customer service requirement may be simple or complex. Simple requirement may not look
for composite services to satisfy the user query. Whereas, complex requirements may have both
functional and non-functional aspects which needs to be satisfied. For this kind of complex
requirement, the services need to be composite. The composite service is a service formed by a
composition of other available services. Google research application is accepted as a web
service and integrated with other services, such as Gmail, AdWords, Picasa, Orkut, You Tube
and Google Maps service, to provide an integrated environment for service consumers. The other
known example for service composition is a tour booking service that can be formed as a web
service and integrated with other services such as hotel booking, sight seeing, flight booking or
car-rental in order to provide a collaborated environment for user. However, there may exist huge
number of (tour booking) services which provide similarly functional characteristics. Service
consumers not only expect the service to meet functional aspects but they also require services
to meet non-functional aspects properties that is, quality of services (QoS) such as service
reliability, security, trust and execution cost, etc. Thus the selection of services based on non
functional qualities gain more advantages nowadays.
2.2 Provider-Service Offerings
The services offered by service provider are concerned about functional and non-functional
qualities of services. The functional properties make use of domain ontology. To provide
consumer the requested service with non-functional properties makes use of QOS ontology. The
problem that arises here is how to map the quality preferences offered by consumer with the
quality categorization in QOS ontology. This can be solved by labeling the qualities (eg.
performance, security) in QoS ontology with the service Identification.
2.3 Service Selection Process
This involves matching the customer required service with the offered service. The dynamic
selection of web services involves getting user requirements, the provider of service need to
publish or register their services using service description language, finally the matcher will match
the user requirements with the registered service description. The requirements specified by the
user or customer may vary from description of service and Quality of service (QoS). To overcome
this problem, domain ontology and QoS ontology may be used. The registered service
descriptions by the service provider contain the semantic profile and QoS parameters. The
provider of the service is also required to specify the location of a WSDL document describing a
web service. A query processor may be used to analyze the requirement specified by the user
with the domain ontology and QoS ontology. The semantic matcher will match the user request
with service description and locate available services matching with requirements. The
discovered services are then taken as input to the selection process to select the best service
that satisfies the user requirement. In basic form, service selection involves mapping a set of
services to a service—this can be thought of as the best service; in a more general form, service
selections maps a set of services to a ranking of the services in that set [6]. Multitude of service
selection techniques and algorithms are proposed in the literature such as Use of optimization
algorithm [3] for service selection, integer linear programming [4], broker-based architecture [5],
negotiation model for service selection etc [29] [31]. With the thorough study of service selection
process in the literature, the following approaches are identified.
1. Functional based service selection approach
2. Non-Functional based service selection approach
3. User based service selection approach
The Web Service Selection process is broadly classified as Functional based approach, Non-
functional based approach and User based approach. Functional based service selection
approach represents the Static and Dynamic semantics. Selecting an appropriate service is

concerned with retrieving functional descriptions from service repositories and then ensuring that
the described and required interfaces match with each other. Static semantics represents the
properties of messages and operation semantics. The properties of messages include parameter
passed (Data type, language, unit and business role) and message types (Serviceability, provider
type, purpose, consumer type). Dynamic semantics represents the properties of behavior and
operation logic. With dynamic semantics in the service selection process of Web service, the
resultant contains more than one service provider offering similar services.
With the rapidly growing number of available services, customers are presented with a choice of
functionally similar services. This choice allows customer to select services that match other
criteria, often referred to as non-functional attributes. Two fundamental questions arise because
of this: How can these extra attributes be described and how can one select the most appropriate
service. These questions should address both the selection of isolated services as well as the
selection of services within the context of other services. The non-functional based service
selection represents the QoS and Context in semantic web service selection. The properties of
QoS may be (security, reliability, response time, call cost etc.), the properties of Context may
include context of customer (location, intention, consumer’s name, application, e-mail, termination
of hardware and software) and context of service (provider’s details, service descriptions etc,).
User based approach represents the selection of best service among numerous discovered
services based on customers’ feedback, trust and reputation.
Approach I. Functional Based Service Selection
Today, the advancement in Web services requires growth in the areas of service interoperation,
discovery, selection, composition, choreography, orchestration and mining. A possible solution to
all these problems can be provided by converting Web services to Semantic Web [23]. Semantic
Web services (SWS) can provide a solution to the integration problem like composition. In
general, the semantics to be added to a Web service may be called as functional semantics. In
Web services, functional semantic is taken into consideration thereby avoiding unsatisfied results
which are not of customer interest. Functional property is the functional semantics of a service
that describes what a service actually does.
Web Service Selection is related to the process of evaluating and ranking the discovered web
services to identify the ones that fulfill a set of functional and non-functional properties requested
by the service customer. Most of the existing techniques rely on syntactic descriptions of service
interfaces to find web services with disregard to semantic service parameters. This generates
major problems in the service selection mechanism. To solve these problems, Web service
descriptions are enhanced with annotations of ontological concepts, semantic matching and by
considering non-functional properties.
Approach II. Non - Functional Based Service Selection
In a Web environment, multiple WSs may provide similar functionalities with different Non-
functional property values (e.g., different prices). Such Web services will typically be grouped
together in a single community. To differentiate the members of a community during service
selection, their non-functional properties need to be considered. These properties are
characterized as quality of service (QoS) and context based services. Both are highly important
and are to be taken into account during the WS selection.
The W3C working group (2003) defined various QoS attributes for web services (WS) in their
25th November 2003 publication. This include: performance, reliability, scalability, capacity,
robustness, exception handling, accuracy, integrity, accessibility, availability, interoperability,
security, network-related QoS requirements etc. Although regular QoS attributes are listed, it
remains some issues on selection of web services according to the user desired. First, there
exists some web services provided with similar functional requirements which, might lead to the
problem of differentiating the services with QoS. Second, the perception on QoS of web services

distinct between the customer and provider. There also exist a number of other issues which
need to be considered on QoS based service selection process.
Approach III. User Based Service Selection
A User based methodology is a mechanism using consumers’ feedbacks to identify good services
from bad ones. It has advantages in solving the selection problem for Web services. The service
consumer would like to choose a service that is trusted or a service with a high reputation. Trust
and reputation play an important role in a service selection process of user based service
selection. With this approach, web service selection may be customized according to users’
different constrains and preferences. Most approaches proposed in the literature about
personalized selection concentrate on how to rank web services according to users’ preferences
on various QoS metrics. A trust based methodology [7] for service selection is proposed. QoS-
based semantic web service selection solution with the application of a trust and reputation
management method is presented. This work is based on Virtual Internet Service Provider.
This paper focuses on one of the non – functional property known as QoS based Service
Selection approach, its specification [20], techniques and criteria for evaluating techniques of
QSS approach.
3. QoS BASED SERVICE SELECTION
A QoS property can be static or dynamic [24]. A static QoS property value is defined at the time it
is described whereas the dynamic QoS property value requires measuring and updating its value
periodically. The QoS value from the service consumer’s perspective can be positive, negative,
close, or exact. For example, consumers expect to buy a service with low price and expect to
retrieve the service in a low response time. Whereas performance, integrity etc., have positive
trend in which the consumer expects the positives values are better.
3.1 Specification of Service Selection Approaches
The specification or description for non functional based service selection approaches
concentrates on many factors. These factors are separately identified and presented by analyzing
various techniques of non functional based service selection approach. Table 1 depicts the QSS
Specification and Description.
Spec.
No.
Specification. Descriptions
S(1) QoS Modeling Specify the modeling language used.
Such as WSML and its variants WSML –
Core, WSML – Flight, WSML – Rule,
WSML – DL and WSML – Full
S (2) QoS
Categorization
Describe the Ontology of QoS
categorization with its identification value.
S (3) User
Preferences
Describe the varying preferences for the
non-functional criteria specified by the
service consumer
S (4) QoS Evaluation Specify the evaluation criteria used to
evaluate the non – functional properties.
S (5) Aggregating the
evaluation of
This deals with aggregating individual
scores to gain a final score for the service.

TABLE 1: QSS Specification and Description
(a) QoS Modeling
Service requestors need to distinguish services based on their non-functional criteria to make the
most appropriate choice amongst a number of services with equal or similar functionality.
Therefore, a QoS modeling is needed. That can be used in service descriptions as well as service
requests. Due to the adaptability of non-functional properties (the new ones might be required at
any time) it is unlikely that a complete standard set can be identified. The criteria differ depending
on the domain. For example the E-Learning domain service should consider the accuracy,
reputation, and cost. In contrast E-Publishing service should consider the security, price, quality
properties. Therefore it is desirable that the model for delivering non-functional properties is
designed in a simple way.
WSMO with its associated language, the WSML (Web Service Modeling Language) provides a
formal syntax and semantics to describe the QoS characteristics of services. The Web Service
Modeling Ontology (WSMO) defines four main elements as the main concepts of semantic Web
service. This includes Ontologies, Web Services, Goals and Mediators [25]. Ontology’s are formal
explicit specifications of a shared conceptualization [21]. They define a common agreed
terminology by providing concepts and relationships between the concepts. Goals are
descriptions of web services that satisfy the user desires when confer with a service in terms of
functional specification, behavior and quality of service. Web Services are description about
services. The description consist of functional, non-functional and the behavioral aspects of web
services. Mediators address the heterogeneity issues between different WSMO elements. The
Web Service Modeling Language is a formal language for describing ontologies, goals, web
services and mediators. It is based on logical formalisms of WSMO namely description logics,
first – order logic, and logic programming [22]. These formalisms are the basic point to describe
the variants of WSML. The variants includes, WSML – Core, WSML – Flight, WSML – Rule,
WSML – DL and WSML – Full.
(b) QoS Categorization
QoS
S (6) QoS Properties List the number of non –functional
properties considered.
S (7) Level of
Automation
States the level of automation
mechanisms. A – Fully automated, SA –
Semi automated, NA – Not applicable.
S (8) Coordination
Distribution
Describes how individual web service can
interact in order to accomplish an
application task. C – Centralized, CO –
Coordination, GCO – Global coordination.
S(9) Agent
Involvement
State whether agent participation is
involved in the process of service
selection mechanism.
S (10) Ranking
Algorithm
A service rank is a quantitative metric that
shows the “importance” of a service within
the process of service selection
mechanism to rank the services.

QoS properties are designed in hierarchical way. This involves grouping properties by domains
such as environment, performance or safety. Speed quality and response time on performance
aspects while security, privacy and authentication are safety aspects. If such hierarchical
structure exists then users should be able to express preferences at a higher level, while service
provides will express their offerings in a fine way. Using WSML [26], the simplest way of modeling
is done by assigning a simple value to non functional properties of WSMO elements. The data
value assigned to non functional properties is used as an identifier during service publication. To
specify QoS characteristics in particular it can be modeled separately with the use of building and
defining QoS Ontology. Figure 1 depicts the QoS ontology with the assumed identifier value.
W3C defines various QoS attributes such as performance, reliability, scalability, capacity, and so
on. Here the figure 1 covers ontology of characteristics such as interoperability, capacity,
integrity, environment, performance, reliability, security, business and availability. When a new
service is published, the value of QoS characteristics in service description is matched with the
value assigned in QoS ontology. By this way, the newly published services are aligned. Upon
receiving the request from the customer, the system extract the services require and QoS
characteristics specified and match with the QoS ontology to locate it.
1. QoS Characteristics
1.1 Interoperability
1.2 Capacity
1.3 Integrity
1.4 Scalability
1.5 Accuracy
1.6 Accessibility
1.7 Environment
1.7.1 Temporal
1.7.2 Location
1.8 Performance
1.8.1 Latency
1.8.2 Response Time
1.8.3 Throughput
1.8.4 Error Rate
1.9 Reliability
1.9.1 Recover
1.9.1.1 Failure
1.9.1.2 Disaster
1.9.2 Consistency
1.9.3 MTBF (Mean Time Between Failures)
1.10 Security
1.10.1 Encryption
1.10.1.1 Data
1.10.1.2 Messages
1.10.2 Authentication
1.10.3 Authorization
1.10.4 Auditability
1.10.5 Accountability
1.10.6 Non – Repudiation
1.10.7 Traceability
1.11 Business
1.11.1 Cost
1.11.2 Reputation
1.11.3 Monitoring
1.12 Availability
1.12.1 MTTR (Mean Time To Recovery)
1.12.2 Load Balancing

1.12.3 Up Time
1.13 Network related
1.13.1 Bandwidth
1.14 Stability
1.14.1 Method Stability
1.14.2 Interface Stability
FIGURE 1: QoS Categorization
(c) User Preferences
Depending on the situation service requestors may have varying preferences for the non-
functional criteria. In the same way, different requestors will have different preferences. A good
mechanism should not only allow expressing values for each property, but preferably also
represent the relations among the preferences. For example, a customer may consider the
security property as more important than privacy when requesting a financial service. Hence, the
selection approach needs to provide for mechanisms for users to specify their preferences, that is
which of the non-functional properties they feel more strongly about and also relations between
these properties.
(d) QoS Evaluation
It is difficult to predict how many non-functional properties will be available, and the type of these
properties for a customer requested service. For example, the evaluation function to compute the
speed criteria will be different from the function to calculate the location criteria. It is not easy to
define a Universal evaluation function for all kinds of non-functional properties. Hence, the
evaluation function for one property adapt to varying numbers of criteria, but should also
automatically identify the measurement methods to be used to evaluate each criteria.
(e) Aggregating the Evaluation of QoS
After evaluation the next step is to aggregate individual scores to gain a final score for the
service. In this step a suitable aggregation method needs to be selected. Global optimization or
local optimization may be used [27]. Using arithmetic or geometric means to aggregate QoS
properties results in complex situations.
(f) Level of Automation
Level of automation states the automation mechanisms like manual process of selection
mechanism, or semi-automatic service selection mechanism or fully automated service selection
mechanism involved in web service selection and composition. Most research contributions
handling the service selection for service composition focus on automatic process without human
intervention. For example human intervention may involve selecting QoS parameters used for
selection, and changing preferences etc. Semi – automatic process involves little human
intervention, the major task such as corrections and composing are done by the system [28].
Fully automated service selection approach may also use agents in the web service selection
process [32].
(f.1) Agent Involvement
State whether agent participation is involved in the process of service selection
mechanism. A software agent is a piece of software that acts for service consumer or provider in
semantic web service to make the process of service selection automatic. Agents work
cooperatively to evaluate either service providers or service consumers.
(g) Coordination – Service Composition
This describes how individual web service can interact in order to accomplish composite service
selection process. The WS-Coordination defines how the coordination among the services need
to take place, how the data items are to be exchanged in order to complete successful
composition as part of business process defined in a Business Process Execution Language
(BPEL) [30]. The composition algorithms may be centraily cooperated or globally cooperated.

(h) Ranking Algorithm
A service rank is a quantitative metric that shows the importance of a service within the process
of service selection mechanism. It is known that semantic based service discovery concerns on
the matchmaking process between customer’s requirement and service profile or description. Its
semantic matchmaking process plays a role as a ranking mechanism in service selection
process. However ranking based on semantic similarity does not suit for efficient service
selection. Because, from customers perspective, it is always not true that a web service with high
semantic similarity is suitable than a web service with lower similarity. The other difficulty with
semantic similarity is that the users find it hard to distinct which service is better suitable between
a pool of similar services [17]. To achieve better ranking performance many ranking algorithms
have been proposed in the literature. One such approach is to integrate more information besides
semantic information. The information may range from time, place, location [18], customer and
providers situation [19] etc. The limitation with this approach is that the system becomes more
complicated when new constraints are added. To overcome this, the authors [33] have proposed
a method a social collaborative filtering method for ranking. This method makes use of learning
other user’s previous experiences. This method is used most successfully in all kinds of
recommendation systems but the limitations with this method are information distortion and
independence of service selection.
3.2 QSS Service Selection Techniques
The various techniques of QoS based service selection identified from the literature are
discussed in this section. Figure 2 portraits the various QSS techniques identified.
FIGURE 2: QSS Service Selection Techniques
3.2.1 Service Adaptation Evaluation Based QSS Technique
Baopeng et al [8] proposed a QoS model and used hierarchy policy approach to capture goals of
users, applications, environment and resources to form rational service composition and
adaptation action. The authors have proposed a Service adaptation evaluation (SAE) algorithm to
handle service adaptation problem and service composition decision problem in pervasive
computing environment. The system model consists of property primitives for policy hierarchy
such as Control Construct, InterpretedAs, BelongTo, ExistIn, and Commit. The proposed
multidimensional QoS model is focused not only on the traditional QoS properties but also
Requirements of functional QoS parameters and Environmental QoS. The QoS model is
Service Adaptation
Evaluation
based
QoS Constraints
based
Quality Dependency
Graph based
Ordered Weighted
Averaging Operators
based
QoS Normalization
based
Entrophy Decision
based
Fuzzy Linear
Programming based
Non –Functional (QSS)
Service
Selection
Techniques

designed with four layers namely Resource layer, Environment layer, Application layer and User
layer. Each layer describes its own QoS properties. The use of policies at different layers triggers
the service adaptation and provides better service composition performance. A policy driven
service selection algorithm is proposed by the authors to make selection mechanism semantic-
aware and QoS-aware. It is said semantic-aware because, the algorithm performs well even if the
composite service semantic logic changes to form new semantic logic. QoS-aware represents the
input, output QoS parameter consistency, and end-to-end QoS properties such as delay etc. A
policy description language defined in this technique consists of three symbols namely primitive
symbols, action symbols and function symbols.
3.2.2 QoS Normalization Based QSS Technique
In order to enable quality-driven web service selection, Yutu Liu et al [9] proposed a dynamic and
secure framework to evaluate the QoS of a number of web services. The three key aspects that
are developed in this technique include Extensible QoS model, Preference-oriented service
ranking, fair and open QoS computation. The QoS model in this technique is designed to
evaluate the QoS of web services without changing the computational model. In service ranking,
this technique concentrates on representing QoS from the service requestor’s preference
perspective. The QoS computation aspects ensure that the information is collected in a fair
manner. For QoS based service selection modeling, three quality parameter or properties is
measured for generic quality services namely execution price, duration and reputation. It
considers transaction, compensation rate, penalty rate for business related quality criteria. In
order to rank the web services, this technique prefer normalization. The purposes of
normalization are: one to allow for a uniform measurement of service qualities independent of
units. Two, to provide a uniform index to represent service qualities for each provider. Three, to
allow setting a threshold regarding the qualities. The number of normalizations performed
depends on how the quality criteria are grouped. The authors have proposed a prototype model
to implement the QoS registry with hypothetical phone service. They have analyzed collecting
service quality information, collecting quality information from active execution monitoring and
collecting quality information from user feedback. In their proposed framework, the authors have
defined deterministic and non-deterministic criterion to indicate the value of QoS quality and when
a service is invoked. The non-deterministic indicate for QoS quality that is uncertain when web
service is invoked. The advantage of this technique is, it lessens the overhead of QoS registry,
and it dose not need expensive middleware to select the service provider.
3.2.3 Fuzzy Linear Programming Based QSS Technique
Ping et al [10] proposed a QoS-aware service selection model based on fuzzy linear
programming (FLP) technologies, to identify their dissimilarity on service alternatives and assist
service consumers in selecting most suitable services with needs and preferences of customers.
The proposed model has key aspects such as vague reference, weighting of QoS attributes, and
service ranking. In the process of selecting web services, the vague preference of QoS by service
consumer is handled by the proposed model. Weighting of QoS attributes is designed to explore
the optimal solution. Service ranking deals with ranking on web services. A fuzzy group
consensus aware service selection algorithm is proposed based on LINMAP (Linear
Programming techniques for Multidimensional Analysis of Preferences) model to find the optimal
QoS weighting attribute for web services. For the proposed service selection algorithm, the
authors have represented arithmetic operations on fuzzy numbers. This includes representation
for Triangular Fuzzy number, Fuzzy arithmetic operations for addition, subtraction, multiplication
and division. The normalized Euclidean distance between two triangular fuzzy numbers, and the
weighted square distance from positive ideal solution. Further, the authors have addressed the
consistence and inconsistence measurement of service customers by aggregating difference
between fuzzy performance rating and FIPS. The square distance defined is used for accessing
QoS attributes weights.
3.2.4 QoS Constraints Based QSS Technique
Tao yu et al [11] proposed the service selection problem in two models the combinatorial model
and the graph model. A QoS service broker acts as an external, independent broker entity that

can help users construct composite services. To conduct service selection for the general flow
structure the combinatorial model is used. The combinatorial model reflects the service selection
problem as multidimensional multichoice 0-1 knapsack problem. The graph model sees the
selection problem as a multiconstraint optimal path problem. To provide end – to – end QoS
constraint for distributed services, the authors have proposed a broker based architecture. This
architecture includes, service discovery, planning, selection and adaptation as its main function.
The service selection algorithm proposed in this technique is designed with different composition
structure. An efficient algorithm designed for quality driven web service composition ensures that
the services selected satisfies the QoS requirements of users. Four different algorithms have
been proposed by the authors and the algorithms does the task of service selection, algorithm for
designing QoS constraints, heuristic algorithm to find the near optimal solutions, and algorithm to
handle composition structures namely sequential, parallel, conditional, and loops. The QoS
service broker called QBroker help customers to select the best service for the process of
composite service before invocation. The authors have proposed different stages of process for
service composition namely: Process plan, Function graph, and Service candidate graph. This
technique supports constructs for composition model such as Sequential, AND split, XOR split,
Loop, AND join, and XOR join. The QoS service selection problem as MMKP is designed in such
a way that it ensures to select one service candidate from each service class to build composite
service that meets the QoS constraints. To find optimal solution, BBLP (branch and bound)
algorithm is used with MMKP. WS_HEU algorithm is used in this technique to find feasible
solution in polynomial time. It has three main important steps namely: To find an initial feasible
solution, Improve the solution by feasible upgrades, and to improve the solution by infeasible
upgrades.
3.2.5 Entrophy Decision Model Based QSS Technique
A fuzzy entropy decision model, called Linguistic Entropy Method is proposed [12] to assign
linguistic weights of QoS attributes and prioritizes the ranking order of service alternatives. To
overcome the issue of measuring the QoS criteria in web service selection process, the authors
have evaluated fuzzy weights of QoS attributes and rank the web services. The proposed
technique is composed of enhanced version of Linguistic Entropy Method (LEM) and Fuzzy
Synthetic Evaluation Method (FSEM). The Shannon entropy method uses probability function
estimate uncertainty of object based on information theory. The weights for linguistic terms are
evaluated with the use of triangular fuzzy number. The ratings to linguistic terms is provided by
decision maker and designed by triangular fuzzy number. The algorithm Linguistic Entropy
Method has accomplished a set of procedures to assign weights to QoS attributes in the web
service selection process. First step is to organize the evaluation framework. That is, the QoS
attributes are classified and taxonomy of QoS attributes is prepared. The next procedure is
weighting the QoS attributes. This is performed by the decision maker. The third procedure is to
select QoS attributes using fuzzy entropy weights assigned. Next procedure is to evaluate the
score for each QoS attributes. The next procedure deals with constructing the fuzzy decision
matrix by applying fuzzy weighting rules. Final procedure is about ranking the attributes and the
services are selected.
3.2.6 Quality Dependency Graph Based QSS Technique
Chao Lv et al [13] proposed a technique for service selection mechanism to utilize “serve, be
served” relationship and to evaluate the quality of services in business environment to select the
enterprise to collaborate with. Quality Dependency Graph (QDG) method is used to model the
relationship among enterprises. An Analytic Hierarchy Process (AHP) model is used to calculate
the evaluation result of each candidate organization. The authors have presented Quality
Dependency graph based on the characteristics of enterprise collaboration technique namely
Dependency and Diversity. This QDC is used to evaluate the candidate enterprises in the service
selection process. And an AHP model is used to weights the QoS attributes. The authors have
proposed two algorithms to do the service selection task. First algorithm to create QDC from
business specification. The second algorithm is used to get the service guideline for business
role.

3.2.7 Ordered Weighted Averaging Operator Based QSS Technique
Hong Qing et al [14] proposed a novel non functional property-based service selection method by
modifying the Logic Scoring Preference (LSP) method with Ordered Weighted Averaging (OWA)
Operators to automate the service selection process. The authors have focused on two main
issues of service selection process. They are Service automation and Dynamic aggregation
function. Service automation deals with the automated ranking of QoS attributes. In order to make
the selection process automatic, the ranking problem is transformed into OWA problem to
automatically calculate the LSP orness degree. To evaluate the aggregation function, a method
is used which combines LSP metrics with OWA operators. An algorithm is proposed to show the
modified LSP method. Two new operators called Conjunction and Disjunction is introduced by the
authors in the new LSP algorithm to represent relation between criteria such as replaceability,
simultaneity etc. This LSP algorithm evaluates quantitative features for the different entities. The
four main steps or procedure of this algorithm includes, specifying the evaluation variables,
defining the elementary criteria, analyzing the degree decision and analyzing the preference. To
overcome the change of criteria and preferences in the dynamic environment of service selection,
a type based evaluation matrix is proposed and defined three types of criteria. They are
Numerical type, Boolean type and Set overlap type. The advantage of this technique is that, this
addresses both the issues of service selection process by assigning a proper quantitative
aggregation metrics. And provided an automatic mechanism to facilitate the dynamic metric
invocation and aggregation.
3.2.8 Summary of QSS Based Service Selection Process
QoS based service selection plays an important role in the process of service composition. Table
2 shows the comparative study of QSS techniques with the specification discussed. QoS aware
service selection for compositing the services overcomes the problem faced in functional based
service selection in which they provide only similar functional semantic properties, which might
lead to the problem of differentiating available services. The techniques discussed above have
advanced the process of QoS-aware service selection. However, the issues that need to address
includes:
• Representation of QoS characteristics and QoS modeling.
• Assigning the QoS weightings.
• The fuzzy view on the QoS parameters between service consumers and service
providers.
• The universal metric for evaluating the QoS parameters.
T QSST
(1)
QSST
(2)
QSST
(3)
QSST
(4)
QSST
(5)
QSST
(6)
QSST
(7)
S (1)
Yes.
Multidime
nsional
QoS
model
Yes. Extensible
QoS model
Yes. QoS
model
based on
LINMAP
Yes Not
applicable
Yes No
S (2)
Average Yes Yes Yes Yes Yes Not
applicable
S (3)

Yes Yes Yes No Not
applicable
No Not
applicable
S (4)
No Yes Yes Yes Yes Not
applicable
Yes
S (5)
No Yes Yes Yes Yes Not
applicable
Not
applicable
S (6)
Availabilit
y of
resources
Execution price,
duration,
reputation
Not given
specifically
Not
given
specific
ally
Runtime,
Transacti
on, Cost,
Security,
Network
Not
applicable
Yes
S (7)
Semi -
automate
d
Automated Automated Automa
ted
Automate
d
Semi
automated
Semi
automated
S (8)
Centralize
d
Centralized Centralized Centrali
zed
Peer -
Peer
Not
applicable
Peer - peer
S (9)
No No No No No No No
S (10 Yes.
Policy
driven
ranking
algorithm
Yes. Using
normalization
matrix
Yes.
Consensual
ranking
BBLP
algorith
m and
WS_H
EU
algorith
m
No Yes( QDG,
AHP)
Yes(Logic
scoring
preference
s)
TABLE 2: Comparison of QSS Techniques.

4. ANALYSIS OF QSS APPROACH
The techniques of QSS approach have their advantages and disadvantages when compared with
each other. There are many issues related to QSS approach that need to be addressed.
Researchers all over the world are currently working on various aspects of QSS issues such as
achieving consensus achievement, QoS modeling, etc. The analysis of QSS approach evaluates
each technique based on evaluation criteria to identify which technique suits well for certain kind
of application development. This section describes the various possible evaluation criteria for
QSS approach.
To say whether a technique is good or bad for certain application development, they need to be
evaluated based on some parameters. This process is like testing a program or software. The
general parameters that are to be addressed for Non - functional based service selection
approach include, Accuracy of the technique, Performance of the technique, Service availability,
Complexity of Time, Complexity of cost, Scalability, Supportability, Failure rate, Threats to
validity, Selection rate, Effectiveness, Information Retrieval metrics like precision and recall,
Efficiency, F – measure, Mean average precision, Geometric mean average precision,
Interpolated precision, Interpolated recall. The following are the evaluation metrics used for
information retrieval system [15] [16]. The same set of metrics can be applied for evaluating QSS
techniques.
TABLE 3: Notations for True Positive and True Negatives
Precision (P): It is the fraction of retrieved documents that are relevant to the user’s need.
Where tp and fp are specified in Table 3.
Recall (R): It is the fraction of relevant documents that are retrieved to the user’s need.
Where tp and fn are specified in Table 3.
Accuracy (A): It specifies the fraction of classifications that are correct.
Where tp, fp, tn and fn are specified in Table 3.
F-measure: A measure that trades off precision versus recall is the F-measure. It is the harmonic
mean of precision and recall.
Notations Relevant Non relevant
Retrieved true positives (tp) false positives (fp)
Not retrieved false negatives (fn) true negatives (tn)

A new evaluation criteria is applied to [8] evaluate the adaptation of service selection
assessment. Considering the user and environment requirements the criteria is proposed. [9]
Conducted a series of experiment to investigate the relationship between QoS value and
business criteria, study of effectiveness of price and the sensitivity factors in QoS computation. In
[10], the approach not only deals with the decision maker’s imprecise perceptions under
incomplete information, but also objectively determines the importance weights of QoS criteria.
The computational time is evaluated for this. The performances of algorithms for sequential and
general flow structure are evaluated in [11]. This study includes two parts: the comparison of
optimal and heuristic algorithms where runtime, approximation ratio, memory usage as metrics
are used and the comparison of combinatorial and graph models where the provisioning success
rate as a metric is used. The performance rating of each service alternative and the score of each
alternative service is evaluated in [12]. The evaluation parameters for evaluating the QSS
techniques are depicted in Table 4.
TABLE 4: QSS Evaluation Metrics for Web Service Selection.
Spec. No. Specification. Descriptions
E(1)
Accuracy
Accurate gives many results in many
senses. In service selection, accuracy
defined as how relevant services are
acquired that satisfy the user requirement
E (2) Service
availability
Service availability defines the existence
of services in the registry.
E (3) Computational
Time
Time to retrieve the related or best
relevant services that satisfy the customer
need.
E (4) Computational
Cost
The total amount of cost required to get or
select the services from the register which
is been already registered by the service
provider.
E (5) Scalability The possibility to register or select more
services in the future.
E (6) Information
Retrieval
metric
The kind of metrics used to measure the
retrieved services.
E (7)q Supportability Support to modify or replace the services
in the registry by the service provider.
E (8) Security States the security measure defined in the
technique proposed.
E(9) Usability States how usable and efficient the
retrieved services are.

5. DISCUSSION AND CONCLUSION
With the increasing availability of Web services as a solution to enterprise application
integration, the QoS parameters offered by Web services are becoming the chief priority for
service providers and their service consumers. This paper have outlined the approach of non –
functional (QoS) Web service selection based on requirements and specification identified from
the thorough study from the literature. This paper reviewed a number of techniques in the
context of the QoS based approach and have presented a summary of QoS parameters involved
in the techniques identified and also the evaluation metrics that can be applied to obtain and
test how the techniques perform against the specification criteria.
Due to the agile and dynamic nature of the web, providing the suitable QoS for enterprise
business application is really a challenging task. In addition to this, modelling the QoS parameters
also relies on the consensus between service consumer and service provider. To achieve the
consensus among the service holders, their fuzzy view on QoS parameters have to be modelled
and weighted in universal manner. This may cause service providers and consumers to better
understand about QoS characteristics. The measurement process for each QoS parameters is
very complex since it should consider what and how to measure, who does the measuring and
where the measurements are taken. This raises the issue of conflicts on QoS characteristics
metrics between service consumer and provider.
It can be concluded that most approaches contribute specific aspects to the overall picture of
service selection, which requires methods for expressing user requirements, expressing service
offerings and also the actual service selection method. Approaches tend to concentrate on
specific of these areas and employ a variety of techniques to do that. It is more appropriate to
make some suggestions for future developments in the area of selection approaches.
Important aspects that need addressing are powerful mechanisms to capture user requirements
that are both user friendly and also expressive enough to capture large numbers of preferences
and the logical relations between preferences. One aspect that falls into this area is the
measuring of weights. Also, in the process of capturing the needs of users, their preference of
data, research has to show interest and capability to automatically capture this, to reduce the
burden on the user part, and to react to changes in circumstances automatically.
6. REFERENCES
[1] Matthias Klusch, Patrick Kapahnke. “Semantic Web Service Selection with SAWSDL-MX”.
German Research Center for Artificial Intelligence, 2008.
[2] Roy Grønmo, Michael C. Jaeger. “Model-Driven Methodology for Building QoS-Optimised
Web Service Compositions”. In Proceedings of the fifth IFIP International Conference on
Distributed Applications and Interoperable Systems (DAIS’05), pp. 68–82, Athens, Greece, May
2005. Springer Press.
[3] Matteo Baldon, Cristina Baroglio, Alberto Martelli, Viviana Patti. “Reasoning about interaction
protocols for customizing web service selection and composition”. The Journal of Logic and
Algebraic Programming, Elsevier, No. 70, pp. 53 – 73, 2007.
[4] Hassina Nacer Talantikite, Djamil Aissani, Nacer Boudjlida.“ Semantic annotations for web
services discovery and composition”. Journal of Computer Standards & Interfaces, Elsevier, pp. 1
– 10, 2008.
[5] Dong-Hoon Shin, Kyong-Ho Lee, Tatsuya Suda. “Automated generation of composite web
services based on functional semantics”. Journal of Web Semantics: Science, Services and
Agents on the World Wide Web, vol:7, pp. 332 – 343, Science Direct 2009.

[6] Matthias Klusch, Patrick Kapahnke. “Semantic Web Service Selection with SAWSDL-MX”.
German Research Center for Artificial Intelligence, 2008.
[7] Galizia, S. Gugliotta ,A. Domingue, J. Milton Keynes. “A Trust Based Methodology for Web
Service Selection”. In Proceedings of the IEEE International Conference on Semantic Computing,
pp. 193 – 200, CA, 2007.
[8] Baopeng Zhang, Yuanchun Shi and Xin Xiao. “A Policy-Driven Service Composition Method
for Adaptation in Pervasive Computing Environment”. First International Symposium on Pervasive
Computing and Applications, pp. 619 – 624, 2006.
[9] Yutu Liu, Anne H. Ngu, Liang Z. Zeng. “QoS computation and policing in dynamic web service
selection”. In Proceedings of the thirteenth international World Wide Web conference on
Alternate track papers & posters , pp. 66 - 73 , 2004.
[10] Ping Wang, Kuo-Ming Chao, Chi-Chun Lo, Chun-Lung Huang, Yinsheng Li. "A Fuzzy Model
for Selection of QoS-Aware Web Services”. In Proceedings of the IEEE International Conference
on e-Business Engineering (ICEBE'06), ICEBE, pp.585-593, 2006.
[11] Tao Yu and Kwei-Jay Lin. “Service Selection Algorithms forWeb Services with End - to- End
QoS Constraints”. In Proceedings of the 2005 IEEE International Conference on e-Technology,
e-Commerce and e-Service , pp. 129–136, Hong Kong, China, March 2005.
[12] Ping Wang. “An Entropy Decision Model for Selection of QoS-Aware Services Provisioning”.
Department of MIS, Kun Shan University, Taiwan, 2006.
[13] Chao Lv, Wanchun Dou, Jinjun Chen. “QoS-Aware Service Selection Using QDG for B2B
Collaboration”. In Proceedings of the fourteenth IEEE International Conference on Parallel and
Distributed Systems, pp. 336 – 343, 2008.
[14] Hong Qing Yu, Stephan Reiff-Marganiec. “A Method for Automated Web Service Selection”.
Interaction and Context Based Technologies for Collaborative Teams, project. IST-2006-034718,
2007.
[15] Christopher D. Manning, Prabhakar Raghavan, and Hinrich Schütze. “An Introduction to
Information Retrieval”. Cambridge University Press Cambridge, England, 2009.
[16] “Common Evaluation Measures”. In Proceedings of the Text Retrieval Conference (TREC)
Appendices, 2006.
[17] Wenge Rong, Kecheng Liu and Lin Liang. “Personalized Web Service Ranking via User
Group combining Association Rule”. In Proceedings of the IEEE International Conference on
Web services, pp. 445 – 452, 2009.
[18] J. Kuck, and M. Gnasa. “Context-Sensitive Service Discovery meets Information Retrieval”.
In Proceedings of fifth IEEE International Conference on Pervasive Computing and
Communications Workshops, pp. 601-605, 2007.
[19] Z. Maamar, S. Mostéfaoui, and Q. Mahmoud. “Context for Personalized Web Services”. In
Proceedings of the 38th Annual Hawaii International Conference on System Sciences, Hawaii,
USA, 2005.
[20] Swarnamugi .M, Sathya .M. “Specification Criteria for Web Service Selection Approaches”.
International Journal on Computer Engineering and Information Technology, pp. 29 – 38, vol( 23),
Issue No: 01, ISSN 0974-2034, May 2010.

[21] T. R. Gruber. “A translation approach to portable ontology specifications”. International
Journal on Knowledge Acquisition, vol: 5(2), pp. 199–220, 1993.
[22] Ioan Toma, Douglas Foxvog, Michael C. Jaeger. “Modeling QoS characteristics in WSMO”.
In Proceedings of the first workshop on Middleware for Service Oriented Computing, pp. 42 – 47,
2006.
[23] Gennady Agre, Tatiana Atanasova, Joachim Nern. “Migrating from Web Services To
Semantic Web Services: InfraWebs Approach”. EU Project INFRAWEBS - IST
FP62003/IST/2.3.2.3 Research Project No. 511723.
[24] Vuong Xuan Tran, Hidekazu Tsuji ,Ryosuke Masuda. “A new QoS ontology and its QoS-
based ranking algorithm for Web services”. Journal on Simulation Modelling Practice and Theory,
ScienceDirect, vol(17), Issue No: 8, pp. 1378-1398. September 2009.
[25] Web Service Modeling Ontology (WSMO) - http://www.wsmo.org/TR/d2/v1.3/20061021/
[26] The Web Service Modeling Language WSML -
http://www.wsmo.org/TR/d16/d16.1/v0.21/20051005/
[27] Mohammad Alrifai, Thomas Risse. “Combining global optimization with local selection for
efficient QoS-aware service composition”. In Proceedings of the 18th international conference on
World Wide Web, ACM, ISBN: 978-1-60558-487-4, pp. 881- 890, 2009.
[28] Miguel Angel Corella, Pablo Castells. “Semi-automatic Semantic-Based Web Service
Classification”. In Proceedings of Business Process Management Workshops, LNCS, pp. 459 –
470, 2006.
[29] Sathya M, Dhavachelvan P, Swarnamugi M, Sureshkumar G. “A Negotiation Model for
Context-Aware Web Service Selection”. In Proceedings of the International Conference on
Advanced Computing and Communication, pp. 38 – 44, Kerala, India, May 2010.
[30] IBM Business Process Execution Language for Web Services.
http://www.ibm.com/developerworks/library/specification/ws-bpel/
[31] Swarnamugi M, Sathya M, Dhavachelvan P. “A Negotiation Model for Web Service
Selection”. In Proceedings of International Conference on Recent Trends in Soft Computing and
Information Technology, pp. 251 – 256, Bhopal, India, Jan 2010.
[32] Yijun Chen, Abdolreza Abhari. “An Agent-based Framework for Dynamic Web Service
Selection”. In Proceedings of the 2008 Spring simulation multiconference, ACM, pp. 13 – 16,
2008.
[33] U.S. Manikrao, and T.V. Prabhakar. “Dynamic Selection of Web Services with
Recommendation System”. In Proceedings of 2005 International Conference on Next Generation
Web Services Practices, pp. 117-121, 2005.

Evaluation of QoS based Web- Service Selection Techniques for Service Composition

More Related Content

What's hot (15)

Similar to Evaluation of QoS based Web- Service Selection Techniques for Service Composition (20)

More from Waqas Tariq (20)

Recently uploaded (20)

Evaluation of QoS based Web- Service Selection Techniques for Service Composition