TOWARDS A MODEL FOR PERSONALIZED

COMMUNICATION SERVICES BASED ON USER PERCEPTION

Cláudia M.F.A.Ribeiro

Departamento de Informática, UERN, Campus Central, Mossoró, Brazil

Nelson S.Rosa, Paulo R.F.Cunha

Centro de Informática, UFPE, Recife, Brazil

Keywords: User Perception, Quality of Service, Personalized Services, User Needs

Abstract: The emerging of the new generation applications like videoconference and the increasing in the demand of

QoS services have enforced the need of new service models. As in the most new applications the user can

perceives the level of quality a service is provided, data communication services are currently evolving

towards more personalized ones. A direct consequence of this trend is the necessity of explicit treatment of

the user perception. Challenges in this evolution include the better understanding of “how” users perceive

QoS and “how” the perception is actually realised by underlying QoS mechanisms. This paper addresses

these questions by presenting a formal architecture, namely ESCHER that implements a conceptual user

perception model for QoS services.

1 INTRODUCTION

The emerging of new generation applications like

videoconference and the increasing in the demand of

QoS services have motivated some key

transformations in the application development

process. Meanwhile, there is a clear necessity of new

service models (Pedersen, 2002).

Data communication services are currently

evolving towards more personalised ones, as the

users can perceives the level of quality a service is

provided (Ghinea and Thomas, 1998). In fact, those

services are becoming as personalised as health care

services, bank services and traditional voice

communication services. A direct consequence of

this trend is the necessity of explicit treatment of the

user perception.

However, typical users are not able to express

their QoS requirements in quantitative terms, as they

are not concerned with details of implementation of

QoS services. For instance, they know neither what

is the upper limit of tolerable packet delay nor jitter

in an IP telephony session. Moreover, they cannot

provide the traffic specification of their application

flow.

Actually, the user has a very subjective view of

QoS and he/she usually defines QoS constraints as a

set of non-functional requirements (NFRs) such as

performance and cost. In order to understand,

precisely specify and map user QoS specifications

into quantitative network parameters, new

capabilities must be incorporated by QoS

mechanisms.

Since Quality of Service is a key factor for

differentiating service offers in a competitive

market, there are many researches activities towards

the definition of service models which are easily

identified by users (EURESCOM, ETSI). Despite

these initiatives being a progress towards the

effective treatment of the individual necessities of

users, two important issues are still open: it is

necessary a better understanding of “how” the user

perceives quality; and “how” the perception is

actually realised into underlying quality of services

elements.

211

M.F.A.Ribeiro C., S.Rosa N. and R.F.Cunha P. (2004).

TOWARDS A MODEL FOR PERSONALIZED COMMUNICATION SERVICES BASED ON USER PERCEPTION.

In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 211-215

DOI: 10.5220/0002647602110215

 SciTePress

We addressed these mentioned issues by

conceiving a layered architecture namely ESCHER

that implements a conceptual perception model for

QoS services. This model focuses the precise

specification and mapping of user QoS requirements

into QoS parameters.

In ESCHER, the QoS specification and the QoS

mapping are based on the QoS abstractions of each

layer (user, application, middleware and QoS

mechanism). For instance, in the user layer the QoS

requirements are defined through non-functional

requirements, which express more properly the

constraints defined by the user.

Despite the high abstraction level in which NFRs

are commonly stated, there is a rationale to treat

with the NFRs defined by users. As resources are

traditionally scarce, the resource allocation based on

the quality perceived by the user yields a more

effective resource management. For example, a

video quality may be good for a particular user,

while its quality is not acceptable to others. The

optimisation of resource allocation embodies

benefits to communication service providers, whilst

the differentiation of services motivated by the user

perception leads to money saving on behalf of the

users.

This paper is organised as follows. Section 2

presents some related work. Section 3 presents the

conceptual user perception model principles. Section

4 is dedicated to description of ESCHER

architecture. In Section 5 is presented a modelling of

VoIP application using ESCHER. The Section 6

illustrates an Implementation scenario of ESCHER

operations. Finally, the last section presents the

conclusions and some directions for future work.

2 RELATED WORK

There are some research activities in capturing of

user QoS requirements (Bhatti et al, 2000) (Bouch et

al, 2000) (Widya, 2001). Many conceptual models

for the treatment of user perception were proposed.

The QoE Model (Quality of Experience) (Moorsel,

2001) proposes an extension of RM-OSI

(Zimmerman, 1980) layers by including an specific

layer to treat user perception. The QoBiz Model

(Quality of Business) (Moorsel, 2001) is an

extension of the QoE model for user perception

under business perspective. The QC Model (Quality

Class) (

Maurits Cornelis Escher (1898-1972) is a graphic artist.

He is most famous for his so-called “impossible”

structures in which allow multiple perceptions

Alfano, 1997) is a model which intends to

facilitate the identification of the level of quality the

users are interested.

Despite the advances the aforementioned

researches represent, there are open questions to be

addressed to explicit treatment of user perception for

effective offers of personalised QoS services. The

User Perception Model, described in next Section,

presents a service model to solve the questions

above mentioned, by considering the explicit

treatment of user quality of perception. It serves as a

basis for QoS management service implemented by

ESCHER architecture proposal.

3 THE CONCEPTUAL USER

PERCEPTION MODEL

The conceptual user perception model follows some

basic principles in order to explicitly treat the user

perception:

• The user may require different levels of quality

for multimedia services. In a personalized

service model, the user has the opportunity of

defining her/his desired quality level together

the cost limit;

• The user perceives quality by considering

personal characteristics. People react physically

in a different way to audio and visual

stimulations. Moreover, another factors, like the

interest on the particular activity in execution

and personal preferences also influence the way

people perceives quality;

• The user satisfaction level related to quality

perceived is more properly represented by the

trade-off between non-functional requirements

like performance, security and cost.

Traditionally, the quality of service has been

managed by considering only aspects related to

performance (Aurrecoechea, 1995);

• The effective treatment of the user perception by

low-level QoS mechanisms, suppose an

intermediary step to map subjective

specification defined by the user into objective

underlying parameters (Yamazaki and Matsuda,

1999). This important task can be used to guide

the process of resource allocation.

In addition to these principles, another key

feature of this model is the separation of concerns in

the requirement specification. By adopting this

principle, both the specification of non-functional

ICEIS 2004 - HUMAN-COMPUTER INTERACTION

212

requirements (including QoS requirements) and

functional requirements of a given application may

be treated separately. Additionally, it allows legacy

applications to use QoS services without any code

change.

4 THE ESCHER ARCHITECTURE

In this section, we show how the principles of the

model proposed, can be implemented through an

abstract architecture which is mainly focused on the

capture and the map of user perception into

underlying QoS parameters.

Figure 1 illustrates the ESCHER layers and the

relationships between QoS abstractions of each

layer.

There are two basic elements in ESCHER: layers

and relationships. The first one represents a

particular entity together its respective view of QoS.

The second one relates abstractions used to specify

quality requirements in each layer.

The ESCHER architecture (Ribeiro et al, 2003)

is structured into four independent but

interconnected layers: user, application, middleware

and QoS mechanism.

Figure 1: The ESCHER Architecture Overview

At the User Layer, the user requirements

specification is defined through non-functional

requirements (NFR) like performance and cost, their

associated constraint level, e.g. “high” or “low”, and

the relationship of priorities between them. At the

Application Layer, a set of QoS attributes (QoSattr)

realizes NFRs defined at User Layer. For instance,

the QoS attributes “frame rate” and “resolution” can

be used to realize a NFR “quality of video”.

The third layer represents the QoS middleware

view, a key architectural element for the

“transparent” support of QoS. At this layer, QoS

requirements are defined by QoS characteristics

(QoSchar) such as delay, jitter and loss. Finally, the

lower layer is the QoS Mechanism Layer. The

abstraction used in this layer is QoS parameters

(QoSparam) that represent low-level parameters

used by specific mechanisms such as RSVP protocol

(Braden et al, 1997).

In ESCHER, the mapping process is

bidirectional, from the highest layer to the lowest

one which is called downQoS mapping and the

opposite flow, called upQoS mapping. Each flow

consists of a three steps process involving the

mapping between two layers each. The detailing of

the form and use of each one of these flows will be

subject of section 6 that deals with implementation

scenarios. Despite the effective enforcement of

services with QoS guarantees is out of the scope of

this work, it was established a formal interface with

mechanisms which address this issues.

5 MODELLING USER

PERCEPTION

Down QoS mapping

NFR

Lev el

Level

NFR

Use r La y e r

ApplicationLayer

Va l u e

QoS attr

Va l u e

MiddlewareLayer

Va l u e

QoS char

Va l u e

QoS char

Va l u e

QoS MechanismLayer

Va l u e

QoS param

Va l u e

QoS param

Va l u e

Up QoS mapping

Figure 2 illustrates an IP Telephony application

modelled through ESCHER architecture

abstractions. It is possible to observe the QoS

mapping process along the layers. At ESCHER, the

user identifies the type of application and defines

QoS constraints and QoS priorities.

In this example, the user requires a “High”

“Quality of Speech” and a “Medium” level to the

“Cost” requirement. These two NFRs are

conflicting, since the user is interested in an

increasing of “Quality of Speech” (QofSpeech) and a

decreasing in the “Cost” (Cost). In this particular

case, we assume that the user considers “Cost” more

restrictive in the sense that it has priority over the

“Quality of Speech”.

In order to realize each NFRs defined, a set of

typical QoS attributes with respective values was

identified. It includes more specific parameters to

describe quality of speech ( R, LSR, e2e and Codec)

for Quality of Speech and one parameter more

general to describe Cost (perSec). These attributes

are mapped to a reduced set of QoS characteristics

which can be managed by the middleware.

The values of each one of these QoS

characteristics are well-known and respect the level

of quality defined for NFR quality of speech.

TOWARDS A MODEL FOR PERSONALIZED COMMUNICATION SERVICES BASED ON USER PERCEPTION

213

Finally, a set of underlying parameters is derived

from the last step of mapping process.

The relevance of this example is to show that it

is possible to integrate different requirements

specifications of quality, in an uniform way. From

the qualitative specification defined by the user to

quantitative parameters that adjust the network

functions. In addition, the precise definition and

formalization of mapping rules through ESCHER

elements, makes possible to create mechanisms to

adapt dynamicaly the service quality level.

Figure 2: An Illustrative Example of User Perception

Modelling

6 ESCHER IMPLEMENTATION

SCENARIO

Figure 3 depicts an abstract implementation scenario

of ESCHER architecture. In this abstract scenario,

three actors are directly involved: User/Customer -

Residential users and smaller enterprises connected

to the Internet by some type of connection; Service

Provider (SP) - delivering content, application

services or simply service management; and

Network Provider (NP) - delivering IP QoS

connectivity.

The activities played for ESCHER elements can

be grouped in two phases: configuration and

monitoring. The first one is made up of the

specification of user QoS requirements, the mapping

of QoS requirement into QoS parameters and the

negotiation for establishment of the QoS contract

(SLA – Service Level Agreement) (

Bouillet et al,

2002)

. The second one is composed by the activities

of monitoring contracts (SLA), the reverse mapping

to identify possible changes on agreed QoS level and

the adaptation process.

QoS desired

QoS agreed (SLA)

Service Provider

Us e r

Network Provider

QoS perceived

Down QoS map ping

NFR

Level Level

NFR

UserLay er

Ap plicat i onLayer

Value

QoS a t tr

QoS attr

Value

Middlewar eLay er

Value

QoS char

Val ue

QoS c har

Val ue

QoS Mechan ismLayer

Value

QoS p ar am QoS p aram

Val ue

QoS pa ra m

Val ue

Up QoS mapping

Down QoS map ping

NFR

Level Level

NFR

UserLay er

Ap plicat i onLayer

Value

QoS a t tr

QoS attr

Value

Middlewar eLay er

Value

QoS char

Val ue

QoS c har

Val ue

QoS Mechan ismLayer

Value

QoS p ar am QoS p aram

Val ue

QoS pa ra m

Val ue

Up QoS mapping

QoS desired

QoS agreed (SLA)

Service Provider

Us e r

Network Provider

QoS perceived

Down QoS map ping

NFR

Level Level

NFR

UserLay er

Ap plicat i onLayer

Value

QoS a t tr

QoS attr

Value

Middlewar eLay er

Value

QoS char

Val ue

QoS c har

Val ue

QoS Mechan ismLayer

Value

QoS p ar am QoS p aram

Val ue

QoS pa ra m

Val ue

Up QoS mapping

Down QoS map ping

NFR

Level Level

NFR

UserLay er

Ap plicat i onLayer

Value

QoS a t tr

QoS attr

Value

Middlewar eLay er

Value

QoS char

Val ue

QoS c har

Val ue

QoS Mechan ismLayer

Value

QoS p ar am QoS p aram

Val ue

QoS pa ra m

Val ue

Up QoS mapping

UserLayer

QofSpeech

High

Cost

Medium

ApplicationLayer

> 80

LSR

e2eD

Codec

< 250ms

G.711

perSec

unitValue

G.726

MiddlewareLayer

Delay

<35ms

Jitter

=0%

Loss

<150ms

QoS MechanismLayer

5280

Pe ak Ce l lRa t e

sustainable

Ce l l Ra t e

5280

QoS class

CBR

Figure 3: ESCHER Implementation Scenario

One important characteristic of ESCHER is the

facility in specifying the QoS requirements. The user

basically defines the QoS desired (QoS desired) by

the minimum set of requirements related to

constraints, priorities and the application type

(Figure 4). Since the ESCHER architecture focuses

the explicit treatment of user perception, the service

specification is the most important task to be

executed.

User QoS

Specification

down QoS

Mapping

QoS

Negotiation

QoS desired

(constraints, priorities

and application type)

QoS agreed (SLA)

NFRs

QoS parameters

Network Provider

Service Provider

Figure 4: ESCHER Service Configuration Phase

The mapping process flow is made from the top

to the bottom of ESCHER architecture layers (down

QoS Mapping) that is responsible for translating the

User

SLA QoS offered

User QoS

Specification

down QoS

Mapping

QoS

Negotiation

QoS desired

(constraints, priorities

and application type)

QoS agreed (SLA)

NFRs

QoS parameters

Network Provider

Service ProvidererUs

SLA QoS offered

ICEIS 2004 - HUMAN-COMPUTER INTERACTION

214

user QoS requirements, defined through NFRs, into

QoS parameters treated by the underlying QoS

mechanisms.

After the down QoS mapping, it is initiated the

negotiation to allocate resources to satisfy the

required QoS. The mapping process becomes

available a set of specific QoS parameters treated by

underlying mechanisms.

7 CONCLUSIONS AND FUTURE

WORK

We have presented an architecture, namely

ESCHER, which explicitly taken in account the user

perception, whilst it also proposes a systematic

mapping of QoS requirements (at user level) into

QoS parameters (treated by QoS mechanisms).

Besides the high level of abstraction the user

QoS requirements is specified, the benefits of our

proposal also include: the separation of concerns in

QoS specification and the treatment of user

satisfaction as a trade-off between NFRs. The first

one allows legacy applications to use QoS services

offered by the middleware. The second one serves as

a basis for more flexible QoS adaptation mechanism.

In terms of future work, we intend to concentrate

on the QoS mapping by identifying and formalizing

the rules to make this process automatic. We also

intend to verify some properties of our model. For

example, the capacity of our model to reflect

changes in the level of provided QoS and vice-versa.

REFERENCES

P.E. Pedersen, L.B. Methlie and H. Thorbjrnsen,

"Understanding mobile commerce end-user adoption:

a triangulation perspective and suggestions for an

exploratory service evaluation framework". HICSS-

35, Hawaii, US, Jan 7-10, 2002.

G. Ghinea and J.P. Thomas. “QoS Impact on User

Perception and Understanding of Multimedia Video

Clips”, Proc. of ACM Multimedia ’98, Bristol, United

Kingdom, 1998.

EURESCOM, “Offering Quality Classes to end users”,

Deliverable 1,Volume 1 of 2, Project P906-GI,

QUASIMODO,

http://www.eurescom.de/public/projectresults/results.asp

ETSI, TS 101329-2 v1.1.1, End to End Quality of Service

in TIPHON Systems; Part 2: Definition of Quality of

Service (QoS) Classes, 2000-07,

www.etsi.org

N. Bhatti, Anna Bouch, and Allan Kuchinsky –

“

Integrating user-perceived quality into web server

design”. In 9th International World Wide Web

Conference, Amsterdam, May 2000.

A. Bouch, A. Kuchinsky, N. Bhatti - “Quality is in the

Eye of the Beholder: Meeting Users' Requirements for

Internet Quality of Service” –

Proc. of the CHI 2000

conference on Human factors in computing systems,

p.297-304, April 2000.

I. Widya, R.E. Stap, L.J. Teunissen, B.F. Hopman. “On

the end-user QoS-awareness of a distributed service

environment”. PROMS'01). Enschede, Netherlands,

October 2001.

Aad P.A van Moorsel. Metrics for the Internet Age:

Quality of Experience and Quality of Business. HP

Labs Technical Report HPL-2001-179.

H. Zimmerman. OSI Reference Model – The ISO Model

of Architecture for Open Systems

Intercommunications. IEEE Transactions on

Communications, COM-28, 1980.

Marco Alfano. User requirements and resource control for

cooperative multimedia applications. Lecture Notes in

Computer Science: Multimedia Applications, Services

and Techniques - ECMAST '97, volume 1242,

Springer, 1997, pp. 537-552.

C. Aurrecoechea, A. Campbell and L. Hauw “A Survey of

Quality of Service Architecture”, Multimedia Systems

Journal, November, 1995.

T.Yamazaki and J. Matsuda, "On QoS Mapping in

Adaptive QoS Management for Distributed

Multimedia Applications", Proc.ITC-CSCC'99, vol.2,

pp. 1342-1345, July, 1999.

R. Braden, L. Zhang, S. Berson, S. Herzog, S. Jamin.

“Resource Reservation Protocol (RSVP) – version 1

Functional Specification”. RFC 2205, September

1997.

Ribeiro, Claudia M.F.A., Rosa, Nelson S. and Cunha,

Paulo R.F. “User Quality of Perception: Towards a

Model for Personalised Communication Services”.

Formal Methods Europe, Service-Based Software

Engineering Workshop. Pisa, Italy, September 2003.

Proceedings Series, Vol. 2805 2003 XVII, ISBN 3-

540-40828-2.

Eric Bouillet, Debasis Mitre, and ICG. Ramakrishnan,

"The Structure and Management of Service Level

Agreements in Networks", IEEE Journal on Selected

Areas in Communications, Vol. 20 No. 4, May 2002,

pp. 691-699.

TOWARDS A MODEL FOR PERSONALIZED COMMUNICATION SERVICES BASED ON USER PERCEPTION

215