Submitted by Kamila Costa (kamilavasconceloscosta@gmail.com) on 2015-06-11T19:32:18Z No. of bitstreams: 1 Dissertação-Alex F D Monteiro.pdf: 2575711 bytes, checksum: b0ebf6137549e07b5c23b0bb5d4a8632 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-15T18:00:46Z (GMT) No. of bitstreams: 1 Dissertação-Alex F D Monteiro.pdf: 2575711 bytes, checksum: b0ebf6137549e07b5c23b0bb5d4a8632 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-15T18:04:56Z (GMT) No. of bitstreams: 1 Dissertação-Alex F D Monteiro.pdf: 2575711 bytes, checksum: b0ebf6137549e07b5c23b0bb5d4a8632 (MD5) / Made available in DSpace on 2015-06-15T18:04:57Z (GMT). No. of bitstreams: 1 Dissertação-Alex F D Monteiro.pdf: 2575711 bytes, checksum: b0ebf6137549e07b5c23b0bb5d4a8632 (MD5) Previous issue date: 2015-02-20 / Não Informada / The demand for different applications and services on the Internet has grown dramatically in recent years. To meet this demand, various wireless communication technologies with different characteristics have been used to provide access these services. However, not always technology wireless access used is the most appropriate to the requirements of the applications and services or even the user's interest in issues such as access speed, financial cost and power consumption. The main issue is how to identify the best network and access interface that offers the best value for money, meeting user expectations and operating autonomously. In order to propose a solution to address this issue, this paper describes a new interface selection mechanism based on analysis of contextual information obtained from the user and the device. This mechanism, called CANS (Context-Aware Network Selection), is composed of an acquisition module of context information and selection strategies and management interfaces specific to each type of access technology. The CANS is designed to operate on computers that support the Linux operating system. The evaluation of the proposed mechanism is held in the real environment. The results show that the adoption of the proposed selection strategies enables the user to better exploit the advantages of access technologies in an environment, achieving a reduction in device power consumption and financial cost. / A demanda por diferentes aplicações e serviços na Internet tem crescido drasticamente nos últimos anos. Para atender esta demanda, diversas tecnologias de comunicação sem fio com diferentes características têm sido utilizadas para fornecer acesso estes serviços. Contudo, nem sempre a tecnologia de acesso sem fio utilizada é a mais adequada aos requisitos exigidos pelas aplicações e serviços ou mesmo a questões de interesse do usuário como velocidade de acesso, custo financeiro e consumo de energia. A questão principal é como identificar a melhor rede e interface de acesso que ofereça a melhor relação custo benefício, atendendo as expectativas do usuário e que opere de forma autônoma. Com o intuito de propor uma solução para tratar essa questão, este trabalho descreve um novo mecanismo de seleção de interfaces baseado em análise de informações contextuais obtidas a partir do usuário e do dispositivo. Este mecanismo, denominado CANS (Context-Aware Network Selection), é composto por um módulo de aquisição de informações de contexto e por estratégias de seleção e gerenciamento de interfaces específicas a cada tipo de tecnologia de acesso. O CANS é projetado para operar em computadores que suportem o sistema operacional Linux. A avaliação do mecanismo proposto é realizada em ambiente real. Os resultados mostram que a adoção das estratégias de seleção propostas possibilita ao usuário explorar melhor as vantagens das tecnologias de acesso em um ambiente, obtendo uma redução no consumo de energia do dispositivo e do custo financeiro.

Les travaux développés dans cette thèse ont pour cadre général la mise en œuvre d’approches adaptatives permettant de faire évoluer la gestion du réseau en migrant d’une vue "centrée réseau" où l’on se contentait uniquement des paramètres issus du réseau lui-même, vers une vue "centrée utilisateur". Plus particulièrement, ces travaux se sont focalisés sur un des composants principaux de l’ensemble de la chaîne de traitement et de transport, celui de la sélection de la meilleure interface réseau embarquée dans le terminal mobile, l’objectif étant de répondre au mieux aux contraintes imposées par l’environnement. Ces travaux reposent sur une approche décisionnelle dynamique tenant compte de changements dans les paramètres réseaux et des besoins et préférences des utilisateurs au regard des services qui leur sont proposés. En effet, dans l’environnement actuel, se caractérisant par une multiplicité de technologies, d’applications et d’utilisateurs, les terminaux mobiles sont équipés de plusieurs interfaces réseaux. Ces derniers donnent ainsi la possibilité aux utilisateurs de pouvoir basculer dynamiquement d’une interface à une autre dans l’objectif d’assurer une connexion satisfaisant le mieux possible leurs besoins en termes de services en tout lieu, à tout moment et de la meilleure manière possible (ABC, Always Best Connected). Les approches mises en œuvre dans le cadre de la thèse ont permis d’associer simultanément chaque flux d’application à l’interface la plus appropriée de manière à optimiser les performances globales du système. Ainsi, ces travaux ont mené à la proposition d’approches hybrides ayant pour cadre de départ la technique TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) et en y intégrant des modèles issus de la théorie des fonctions de croyance. Pour l’association flux/interface, une proposition basée sur des algorithmes bio-inspirés a été faite dans le cadre de ces travaux. Les résultats obtenus, à la fois par simulation et sur un cas d’usage réel en lien avec le domaine de la santé connectée, ont montré l’efficacité des approches proposées / The work developed in this dissertation has as a general framework the implementation of adaptive user-centric approaches for network interface selection (NIS) and flow/interface association (FIA) in Heterogeneous Wireless Networks (HWNs). The NIS mechanism relies on a dynamic decision approach that considers the change in networks’ parameters and users’ needs and preferences for their current applications. In a heterogeneous environment, the mobile terminals that are equipped with multiple network interfaces provide the possibility for mobile end-users to switch among available network interfaces and select the one that best fulfills their needs anywhere at any time which is known as Always Best Connected (ABC). The approaches proposed in this thesis allowed to associate simultaneously each application flow to its suitable interface in a way that best maximizes the global system performance. This work has led to the proposal of hybrid approaches originating from TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) and integrating models from the belief functions theory for the NIS. Concerning the FIA, a proposal based on bio-inspired algorithms was made as part of this work. Experimental results based on synthetic datasets and an experimental test bed for health monitoring showed the effectiveness of the proposed approaches