Please use this identifier to cite or link to this item: https://app.uff.br/riuff/handle/1/17111
Title: Suportando a adaptação de aplicações pervasivas pelo uso de funções utilidade
Keywords: Processamento distribuído;  Computação pervasiva;  Computação móvel;  Arquitetura de computadores;  Auto-adaptação;  Dimensões de QoS;  Adaptação em tempo de execução;  CR-RIO;  Monitoração;  Requisitos não-funcionais;  Seleção de recursos;  Função utilidade;  contexto;  Auto-configuration;  QoS dimensions;  Run-time adaptation;  CR-RIO;  Monitoring;  Non-functional requirements;  Resource selection;  Utility function;  Context
Issue Date: 9-Jul-2007
Abstract: Applications operate in dynamic environments, due to the variation of the availability of the resources or due to the variation of the user's preferences. To keep user's satisfaction during the operation of such applications, administration costs have grown up. Is this scenario, pervasive applications are aware of such variations in the operation context and adapt themselves without the action of users or administrators. This work aims to list the questions that have to be solved to make adaptation possible, among them: application's architecture, resource discovery, monitoring, user's preferences elicitation and, mainly, best available resource selection. Trying to solve this last question, we propose - as our main contribution - to use a utility function, as a technique which takes into account user's preferences (to weigh conflicting QoS dimensions) and the availability of the resources needed to provide the proper quality. An extra feature is the ease of implementation and the low processing costs, what makes it possible to be used in small devices, like PDAs. In order to validate our proposal, we present some examples where we used a utility function to select the best resource. In the example where a client selects the best AP, we analyzed real traffic traces of a corporate wireless network and traces generated by a simulator parameterized by realistic probability distributions. We verified the quality and robustness of a solution based in a utility function and concluded that the main QoS dimension commercially available to select the best AP, the signal strength, should not guide this selection exclusively.
URI: https://app.uff.br/riuff/handle/1/17111
Appears in Collections:TEDE com arquivos

Files in This Item:
File SizeFormat 
Diego Schenck-dissert.pdf1.81 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.