FUZZY LOGIC BASED QUALITY OF SERVICE MODELS
Jo
˜
ao Antunes, Andr
´
e Vasconcelos and Jos
´
e Tribolet
INESC Inovac¸
˜
ao, Instituto de Engenharia de Sistemas e Computadores and
Instituto Superior T
´
ecnico, Technical University of Lisbon, Lisbon, Portugal
Keywords:
Fuzzy logic, Quality of service models, Complex information systems, CEO framework.
Abstract:
The continuous monitoring of information systems’ quality of service increases importance as business be-
comes more and more dependent of those systems. In order to obtain that view, quality models need to be
defined for those systems. Because of its complexity and today modelling frameworks, quality models tend
to result in a poor representation of reality, mainly because of their lack of ability to represent uncertainty. In
this work, we investigate the use of fuzzy logic’s properties to create a new kind of quality of service models,
which handles uncertainty and imprecision naturally. The objective is to obtain models that are a better repre-
sentation of reality and easier to create and understand. This article presents the investigation on related topics
to support the identified problem and motivations, followed by a solution proposal and a validation scenario.
1 INTRODUCTION
Along the years, several quality models have already
been proposed, both for the implementation of the in-
formation systems as for the evaluation of its general
performance (S.O. et al., 2010). Their main goal is
to find the answer to the question ”What is quality?”.
This answer is commonly built by defining what char-
acteristics are considered important to exist in a ser-
vice (ISO, 2001).
In the case of evaluation of information systems’
performance, there are multiple dimensions that can
be quantified, and from those the definition for qual-
ity’s characteristics can be created. By performance
we mean how well a system, already assumed correct,
works (Khaddaj et al., 2004).
As systems increase in complexity, so do the rules
that reason about its state, or in other words, the
quality of service model becomes more complex and
harder to understand. In addition, the quantitative val-
ues, obtain from monitoring specific system’s dimen-
sions, are usually modeled using boolean like logics
where, for instance, abrupt variations in quality level
can occur, consequence of strict thresholds have to be
defined (Zadeh, 1973).
In such complex information systems, concluding
on the quality level carries much uncertainty and by
forcing fixed thresholds to separate different levels,
the model will no longer represent the reality, where
for some range of values the conclusion about the cor-
respondent level of quality can be fuzzy (Campbell
et al., 1996).
So, the challenge presented in this scenario is to
create models that give a more accurate view of reality
and at the same time require less effort to create and
understand.
1.1 Motivation
Human reasoning is known for its ability to process
incomplete and vague information in order to infer re-
sults or make decisions. As said in the previous sec-
tion, that property presents serious challenges when
creating a model to represent it. That is especially
true if logics that work with scales measured by dis-
crete values are used. A alternative to such logics is
Fuzzy logic (Zadeh, 1973).
By combining the paradigm of the if-then rules
with the descriptive capacity of the linguistic vari-
ables, more comprehensive models can arise, while
at the same time reducing the effort of their develop-
ment and improving its final quality.
1.2 Objectives
The main goal of this investigation is then to evalu-
ate how fuzzy logic’s concepts can improve the con-
struction of an answer to the question ”what is qual-
ity?”. Our objective is also to improve that process, so
a more natural transition from human reasoning real-
516
Antunes J., Vasconcelos A. and Tribolet J..
FUZZY LOGIC BASED QUALITY OF SERVICE MODELS.
DOI: 10.5220/0003694305160519
In Proceedings of the International Conference on Evolutionary Computation Theory and Applications (FCTA-2011), pages 516-519
ISBN: 978-989-8425-83-6
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
ity to a quality of service model, can be achieved.
In order to build such models we’ll need some
framework to provide us with a common set of con-
cepts to describe all the parts and their relations. Be-
ing this investigation under the scope of the Center
for Organizational Design and Engineering (CODE)
the preferred framework for this task will be the CEO
(Vasconcelos et al., 2007). Because CEO framework
does not consider fuzzy logic and some quality of ser-
vice related concepts, an extension proposal will be
presented in order to enable the creation of the mod-
els required to this investigation.
2 RELATED WORK
To have models, we perform modeling. According
to (Silva and Videira, 2001) modeling is both an art
and a science, and a model is the interpretation of a
subset from the real world. Through the simplifica-
tion of the reality down to a set of concepts and rela-
tions, different languages can be used to describe the
model according to the audience expectations, knowl-
edge and/or objectives.
2.1 Quality of Service
Quality is generally defined as a multidimensional
concept (Khaddaj et al., 2004). These dimensions are
used to construct a quality model, as also described
in the International Standard ISO 9126. As for ser-
vice, in the domain of enterprise architectures, (Open,
2006) defines it as a mechanism that provides access
to capabilities. In the field of networking, the notion
of quality of service is associated with the guaran-
tees of performance transporting a flow of informa-
tion, measured mostly through specific metrics.
A broad review of quality of service investigation
is presented in (Campbell et al., 1996), which de-
fends that investigation is mostly focused on individ-
ual layers, instead of addressing the overall QoS Ar-
chitecture. Also, a generalized QoS framework is de-
scribed, to include principles about its construction,
specification and mechanism to handle the system’s
behaviour. QoS specification is described as the cap-
turing of quality level requirements and management
policies, which will be different for each architectural
layer. As for QoS Mechanisms, it will vary accord-
ing to the specification and can be separated in three
groups: provision, control and management mecha-
nisms.
On the topic of QoS specification, (S.O. et al.,
2010) presents a survey of models, according to a
classification set. The authors conclude that perfor-
mance is itself subjective and open to different in-
terpretations. Also, most models were found to be
limited by their inability of handle uncertainty and
imprecision. Some approaches, like the probabilistic
models, which can handle vagueness, is however not
adequate to handle information expressed in natural
language.
2.2 Enterprise Architectures
Based on (Lankhorst, 2005) to manage the complex-
ity of large systems it’s necessary to have an archi-
tecture, which captures all the components, their re-
lationships with each other and their surrounding en-
vironment. An architecture also provides a common
language to describe the system, it’s components and
their relations, improving overall communication be-
tween the stakeholders.
This notion was extended to the field of enterprise
engineering and from that originated the term of En-
terprise Architecture (EA). An enterprise is seen as
a complex system where through a ”whole of princi-
ples, methods and models” it can be decomposed in
individual functional parts with respective relations.
2.2.1 Framework CEO
CEO framework (Vasconcelos et al., 2007) originated
in CODE investigation group, and its goal was set to
describe organizational knowledge as its various lev-
els and the dependency between them. CEO decom-
poses an organization on three separate levels: organi-
zational goals, business processes and resources, each
with adequate forms of representation according to
the concepts in question.
CEO uses the Unified Modeling Language (UML)
for implementation, with the help of developed
stereotypes. In order to represent information sys-
tems’ concepts, the framework extended its meta
model with the necessary concepts, which will be the
starting point for our own extension proposal.
3 SOLUTION PROPOSAL
Our solution is targeted at reducing some of the com-
plexity in models by creating models that don’t force
the removal of uncertainty. That will be achieved
through the use of fuzzy logic’s concepts, using the
method described in the following section.
FUZZY LOGIC BASED QUALITY OF SERVICE MODELS
517
3.1 Fuzzy Models
The fuzzy controller, as the most successful applica-
tion of fuzzy logic, will be used to guide the develop-
ment of the new quality of service models. So, the
main focus for the new models will be in the cre-
ation of the knowledge database. In order to do so,
the necessary variables need to be defined, that in this
case correspond to all system’s measurable dimen-
sions and quality factors.
Regarding the if-then rules, we’ll consider as input
variables those corresponding to measurable dimen-
sions, and as output variables the quality factors. In
order to obtain the final and crisp results for the eval-
uated rules, its also important that the chosen method
to the defuzzification is not too complex to avoid mak-
ing the models hard to understand and/or create (e.g.
Centre of Gravity).
As for the all necessary fuzzy reasoning knowl-
edge, we’ll follow the original work done in (Zadeh,
1973).
3.2 Metamodel Extension
Regarding the extension to CEO framework, after de-
scribing the main characteristics of the new QoS mod-
els, the concepts to be proposed are the following:
Measurable Dimension. Any system dimension
that can be observed and quantified. This concept
is connected with ”IT Block”, from CEO Frame-
work.
Quality Factor. Relevant characteristic for the
definition of quality in a system. This concept is
connected with ”IT Service”.
Linguistic Value. Represents a value of a lin-
guistic variable. Each measurable dimension and
quality factor will be composed by these values.
By connecting two different values, from a mea-
surable dimension and a quality factor, we’ll be
defining a if-then rule.
Fuzzy Operator. This represents an operator to
compose more complex rules.
Using the concepts present previously, the new
models can be represented in order to document or
share between stakeholders. A visual representation
of the extension proposed to the meta model can be
seen in figure 1. As different stakeholders may have
different perspectives over a model, we also propose
a set of views, each with unique objectives:
System Quality View. The first view is an high
level view of the model, where for a specific sys-
tem and services, only the measured dimensions
and considered quality factors are present.
Figure 1: Proposed extension to CEO meta model.
Quality Factor View. This view focus on a spe-
cific service’s quality factor, relating it to all di-
mensions that cause some impact, and also the
values involved.
Measured Dimension View. In this view the fo-
cus is on a system’s dimension, showing all qual-
ity factors that are influenced by it, as for respec-
tive values.
4 SOLUTION VALIDATION
In order to validate the proposed solution, we’ll use
it to develop a quality of service model for an ac-
tual information system, inside a real company, Por-
tugal Telecom Comunica
˜
oes (PT-C). This model will
be used in the continuous and real-time monitoring of
such system, which is performed by the Pulso Plat-
form.
PT-C created the Pulso platform (Alegria et al.,
2005) because of a growing concern on its informa-
tion systems’ conditions. One of its goals was to mon-
itor performance, availability and errors of their IT
assets with near real-time precision, to determine the
quality of service being provided to users. To achieve
that a network of agents, attached to relevant system’s
components, provides basic metrics that correlate and
aggregate into indicators.
The solution presented previously will be applied
to obtain the necessary quality of services indicators.
The current version of Pulso platform, specifically the
models used today, will serve as basis for comparison
with the developed fuzzy models, in order to deter-
mine the gains and/or limitations imposed by this so-
lution.
4.1 Preliminary Test
As a preliminary test for the creation of the new qual-
ity of service model, this extension of CEO’s meta
model was applied to a fictional example. In this ex-
ample we considered an ”IT Block” representing a
FCTA 2011 - International Conference on Fuzzy Computation Theory and Applications
518
server, named ”Server X”. As for the service it pro-
vides we consider it as ”Infrastructure Support”. As
for measurable dimensions from the server, ”CPU Us-
age” and ”Memory Usage” are available, and for the
service’s quality factors, ”Availability” is to be evalu-
ated. All those variables are composed by two values,
”Low” and ”High”. So, the knowledge base for this
example will be the following:
IF CPU Usage IS Low AND Memory Usage IS Low
THEN Availability IS High
IF CPU Usage IS High OR Memory Usage IS High
THEN Availability IS Low
This model can be viewed in figure 2 which uses
the new concepts proposed in the meta model exten-
sion.
Figure 2: Preliminary test with meta model extension.
5 CONCLUSIONS
From research done so far, we are confident to say
that including the tools to handle imprecision and un-
certainty is a vital step to achieve better quality of ser-
vice models. By having models that better represent
the reality, we also expect to see great improvement in
at least two properties: the effort required to develop
models and there complexity should decrease by con-
sequence. In companies where many complex infor-
mation systems exist, these improvements can have
significant impact. In this article we proposed a solu-
tion based on fuzzy concepts, which provide us with
a formal method to handle uncertainty. From a set
of quantified dimensions, quality factors can now be
defined by correlation rules, even enabling the mod-
eling of very specific cases. Also it enables an easier
extension of the knowledge that defines the quality of
service. An extension for the CEO Framework’s meta
model also is part of the proposed solution. This ex-
tension permits a visual representation of the defined
rules, and also to establish the link between the pre-
vious concepts in the framework. Three new views,
related to quality of service, are also proposed to ad-
dress different perspectives. Finally, the application
of the solution to a fictional system, gives an idea of
future application to validate this investigation.
REFERENCES
Alegria, J., Carvalho, T., and Ramalho, R. (2005). Uma
experi
ˆ
encia open source para “tomar o pulso” e “ter
pulso” sobre a func¸
˜
ao sistemas e tecnologias de
informac¸
˜
ao. Technical report, Portugal Telecom.
Campbell, A., Campbell, A., Aurrecoechea, C., and Hauw,
L. (1996). A review of qos architectures. multimedia
systems, 6:138–151.
ISO (2001). ISO/IEC 9126-1:2001, Software engineering –
Product quality – Part 1: Quality model.
Khaddaj, S., Khaddaj, S., and Horgan, G. (2004). The eval-
uation of software quality factors in very large infor-
mation systems. Electronic Journal of Information
Systems Evaluation, pages 43–48.
Lankhorst, M. (2005). Enterprise Architecture at Work:
Modelling, Communication and Analysis. Springer.
Open, O. (2006). Reference model for service oriented ar-
chitecture 1.0. Technical report, OASIS Open.
Silva, A. and Videira, C. (2001). UML - Metodologias e
Ferramentas Case. Centro Atl
ˆ
antico.
S. O., O., E. O, O., F. M. E, U., Mbarika, V., and B. A,
A. (2010). A survey of performance evaluation mod-
els for distributed software system architecture. Pro-
ceedings of the World Congress on Engineering and
Computer Science 2010, 1:35–43.
Vasconcelos, A., Sousa, P., and Tribolet, J. (2007). Infor-
mation system architecture metrics: An enterprise en-
gineering evaluation approach. Electronic Journal of
Information Systems Evaluation, 10(1):91–122.
Zadeh, L. (1973). Outline of a new approach to the analy-
sis of complex systems and decision processes. IEEE
Trans. on Systems, Man, and Cybernetics, SMC-3:28–
44.
FUZZY LOGIC BASED QUALITY OF SERVICE MODELS
519