IoT Resources Ranking: Decision Making Under Uncertainty Combining Machine Learning and Fuzzy Logic

The Internet of Things (IoT) is characterized by a broad range of resources connected to the Internet, requesting and providing services simultaneously. Given this scenario, suitably selecting the resources that best meet users’ demands has been a relevant and current research challenge. Based on the non-functional parameters of Quality of Service (QoS), IoT plays an important role in the ranking of these resources according to the offered services. This paper presents a proposal to classify and select the most appropriate resource for the client’s request, applying fuzzy logic to address uncertainties in the definition of ideal weights for QoS attributes, and aggregating machine learning to the pre-classification of EXEHDA middleware resources, in order to reduce the computational cost generated by MCDA algorithms. As the main contribution, the pre-classification of new resources of the EXEHDA-RR is presented. The experimental results show the efficiency of the proposed model.