Uncertainty Handling with Type-2 Interval-Valued Fuzzy Logic in IoT Resource Classification

Abstract

The growing supply of Internet-connected resources, often providing more than one service, add complexity to the procedures for discovering, classifying, and selecting the most appropriate resources to meet client demands. The specification of client preferences can lead to inaccuracies and uncertainties, as it depends on prior knowledge and experience for the correct details of parameters such as minimum, maximum, and measurement scales. This paper aims to address uncertainties in specifying and processing client preferences when classifying a set of discovered IoT (Internet of Things) resources. We propose a software architecture for resource discovery and classification in IoT called EXEHDA-Resource Ranking. The proposal stands out in IoT resource classification, exploring three approaches: (i) initial selection of resources with MCDA algorithm; (ii) pre-classification of newly discovered resources with machine learning; and (iii) treatment of uncertainty in preference processing using Type-2 Interval-valued Fuzzy Logic. In addition, one scenario containing resource request simulations applying different client preferences can be demonstrated in EXEHDA-RR features.