Fuzzy Logic-Based Evaluation of IoT Training Simulators

Abstract

Developers offer various options for virtual simulators that allow training in diagnosing Internet of Things (IoT) systems. To select devices with the best characteristics, as well as to identify shortcomings and improve virtual simulators, tools for assessing their characteristics are in demand. An adequate selection of virtual simulators with the best characteristics, as well as the identification of their shortcomings for subsequent improvement, are urgent research problems. The presented research is aimed at solving these problems. A system has been developed for assessing the characteristics of virtual simulators intended for IoT systems training in diagnosis. It is based on the use of fuzzy logical inference. As indicators for assessing the characteristics of virtual simulators used for training IoT systems in diagnosis, it is proposed to use indicators of image realism, sound realism and diversity of training scenarios sets, as well as a general indicator of virtual simulator functioning. The parameters of the membership functions of fuzzy sets and individual outputs of fuzzy rules were automatically selected using a neural network setup. In order to study the proposed fuzzy logical inference system, a model of the process of assessing the virtual simulators characteristics intended for IoT systems training in diagnosis has been developed in the MATLAB software environment using the libraries of the Simulink package. Numerous computational experiments have been carried out using this software model. A comparison of their results with the results of real IoT systems diagnosing showed that the higher the values of the generalized indicator of the virtual simulator functioning, the greater the probability of correct diagnosis of IoT systems by the specialists who trained on it. Thus, the proposed fuzzy inference system can be recommended for evaluating the characteristics of virtual simulators designed to teach diagnostics of Internet of Things systems.