Information Technology Reference
In-Depth Information
A Fuzzy Service Adaptation Based on QoS
Satisfaction
Barbara Pernici and Seyed Hossein Siadat
Politecnico di Milano
Dipartimento di Elettronica e Informazione
Piazza Leonardo da Vinci 32, 20133 Milano, Italy
{
pernici,siadat
}
@elet.polimi.it
Abstract.
Quality of Service (QoS) once defined in a contract between
two parties may change during the life-cycle of Service-Based Applica-
tions (SBAs). Changes could be due to system failures or evolution of
quality requirements from the involved parties. Therefore, Web Services
need to be able to adapt dynamically to respond to such changes. Adap-
tation and evolution of services are playing an important task in this
domain. An essential issue to be addressed is how to eciently select an
adaptation while, there exists different strategies. We propose a fuzzy
service adaptation approach that works based on the degree of QoS sat-
isfaction. In particular, we define fuzzy parameters for the QoS property
descriptions of Web Services. This way, partial satisfaction of param-
eters is allowed through measuring imprecise requirements. The QoS
satisfaction degree is measured using membership functions provided for
each parameter. Experimental results show the effectiveness of the fuzzy
approach using the satisfaction degree in selecting the best adaptation
strategy.
Keywords:
QoS, service adaptation and evolution, fuzzy logic.
1
Introduction
In Service-Based Applications (SBAs), Quality of Service (QoS) parameters may
change during the life cycle of the application. Web service adaptation is an im-
portant phase to deal with such changes. Handling changes in a demanding and
adaptive environment is a vital task. One main issue lies in QoS property descrip-
tions of Web Services. This involves specifying service requirements in a formal
way, monitoring and dynamically adapting and evolving the services with respect
to the QoS changes. Static adaptation is impractical due to the changing envi-
ronment and high cost of maintenance and development. Specifying all possible
alternative behaviour for adaptation at design time is impossible. Therefore, a
declarative approach is required at run-time to support adaptation decisions.
In order to perform run-time decisions for adaptation in a volatile environ-
ment, one issue is to consider the imprecise evaluation of QoS properties. Existing
approaches do not allow partial satisfaction of parameters. It is required that
