Algoritmo da Confiança Inversa para Mineração de Dados Baseado em Técnicas de Regras de Associação e Lógica Nebulosa

A Logics-based Scheduling Method for Hard Real-Time Embedded Systems with Multiple Processors

Anderson Araújo CasanovaSofiane Lobidi

Marcelo CustódioRaimundo BarretoRuiter CaldasJosé Magalhães Netto

This article proposes an algorithm for data mining that presents a new measure for assistance in the extraction of knowledge. The algorithm uses association rules to extract rules of the databases and fuzzy logic for the classification and comparison of the rules. The cyclic scheduling problem is known to be NP-Complete and thus coding a procedure to solve it is difficult and error-prone. This is particularly dangerous for hard real-time systems since they have critical timing requirements. It is necessary a formal strategy to abstract unnecessary details and assure correctness. Furthermore, time-to-market for this kind of systems is continuously decreasing, demanding development processes even more agile. This work proposes a declarative strategy where the user keeps the focus on what is to be computed and not how to implement it. This strategy consists in formally declaring the properties of the system with multiple processors using first-order logics. All the job of finding a feasible schedule is done by Prolog inference engine which carries the scheduling process out. Thus the lack of formality and agility is overcome. In spite of simplicity, this solution is very powerful and expressive.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: