Avaliação de Sensores Virtuais em Aplicações IoT Tolerantes a Falhas

Abstract

In recent years, the use of IoT devices has grown exponentially in several ar eas, such as industry, health, agriculture, and smart cities. With the growing demand for more efficient and integrated technological solutions, the use of virtual sensors in IoT devices is becoming increasingly relevant. Virtual sensors are a technique used to estimate data from sensors in IoT devices. This approach reduces costs and complex ity, allowing flexibility and adaptability in different usage scenarios. In addition, the use of virtual sensors can also contribute to fault tolerance in IoT devices. By simulating different usage scenarios, virtual sensors can be used to test the effectiveness of systems, identify failures and correct them. Given this scenario, this work aims to study models of virtual sensors aimed at fault tolerance in IoT systems. During the work, a case study was implemented focused on scenarios in the context of COVID-19. This disease shook the world and forced everyone to implement prevention and control methods. Allied with this context, we found that replicating sensors is not necessarily the best fault tolerance method and that each context must be analyzed individually. The results show that the errors introduced by the different models of virtual sensors investigated indicate that moving averages are more accurate than sensor replication. The work contributes to virtual sensors and fault-tolerant IoT applications, providing valuable directions for researchers and professionals working with these technologies.