Arquitetura multissensorial em fog computing para dispositivos IoT com foco em agricultura de precisão
Morijo, João Paulo dos Santos
MetadataMostrar registro completo
The Internet of Things (IoT) is characterized by a computational environment that can be divided into three main layers: sensor network, communication and intelligence, the latter located in a fog or cloud environment forming an IoT environment. The first layer can be composed of a wide variety of objects, usually called terminals, with various computational capacities, architectural features, a variety of sensors, actuators and different communication interfaces and standards for interconnection. The second layer can make use of many different wireless communications technologies, including Wi-Fi, ZigBee, Bluetooth or emerging 6LoWPAN communication technologies such as Lora, and transport protocols that incorporate strategies like publish / subscribe to send messages containing sensor data / endpoints for the intelligence layer. The results can be used for monitoring, problem inference, business-level decision making, as well as sensor-level action by sending an actuation message to a terminal. As the IoT sensor network grows, a huge amount of data from various sources flows from the sensor layer to the intelligence layer in the IoT environment. The problem is that to make informed decisions about these data, it needs to be concrete and accurate. Data fusion is an effective way to improve data quality. However, IoT environments are still evolving and the best way or place where data fusion should happen is an open problem. This project presents an architecture for merging IoT sensor data by implementing data fusion with microservices using a container platform embedded in an IoT opensource middleware based on a fog infrastructure that is capable of automatically scaling according to the layer data flow sensor grows. Several data fusion performance tests were performed for different amounts of sensor points and readings on ZigBee and LoRa technologies using the Chauvenet data fusion algorithm in an agricultural environment which can be applied to the concepts of precision agriculture.