Um arcabouço para a coleta de métricas de desempenho para a avaliação da qualidade de experiência em ambientes imersivos virtuais e conectados.

Abstract

This study seeks to contribute to the growing demand for generalizable and effective methods for quantifying quality of experience (QoE). The unique feature of this proposal lies in integrating technical network and application parameters with user experience, encompas sing both objective and subjective factors. The main objective is to investigate, design, and develop a framework capable of correlating these two dimensions of experience, uniting the specific requirements of applications in virtual and connected immersive environments, and conducting experiments using the Random Forest (RF) machine learning algorithm. Quan titative and exploratory, the study uses a methodological approach that involves collecting technical parameters through controlled experiments with users, as well as using statistical analysis and machine learning techniques to develop a quality of experience inference mo del. The research procedures include data collection and preprocessing, as well as training, validation, and testing of the developed model, with a final evaluation of its performance. The results obtained by the indicators indicate high efficiency in the model’s performance in predicting quality of experience, achieving an accuracy rate of 74.07% (representing a hit rate of approximately 3 out of every 4 predictions) and a precision rate of 76.12% (indicating that the values predicted by the model are correct approximately 75% of the time). The study highlights the importance of specific technical parameters in determining QoE, contributing to advances in the area of assessing quality of experience in immersive and connected virtual environments. In conclusion, the framework demonstrates that machine learning models for collecting technical parameters are powerful allies in analyzing quality of experience, ena bling the implementation of real-time monitoring and providing relevant insights for future research.