Publication: Simulación en la nube de nodos medidores de radiación
solar desplegados en un contexto IoT
Loading...
Official URL
Full text at PDC
Publication Date
2021-09-07
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
El despliegue de sensores siguiendo una arquitectura IoT (del inglés, Internet of Things) permite procesar gran cantidad de datos de forma ordenada. Estas mediciones son generalmente analizadas en las capas edge y fog y almacenadas en la capa cloud para contar con un histórico que permita ver cómo han fluctuado en los últimos años los valores para estudios posteriores. Un ejemplo es el estudio de la radiación solar, donde se despliegan multitud de sensores para analizar la evolución de la radiación en las distintas estaciones del año. No obstante, no existen herramientas para analizar mediante métodos de simulación el despliegue de estos sensores en un contexto IoT. En este TFM se propone realizar un sistema simulado dentro de un contexto IoT en el que existen elementos que generan mediciones de radiación, que son transmitidas a través de eventos a la capa fog de procesado. Esta se encarga de detectar valores atípicos, que son reemplazados por valores correctos utilizando distintos métodos de regresión. A su vez se integran técnicas de kriging para calcular la radiación en otros puntos donde no existen mediciones. Para que la simulación sea escalable, se propone un despliegue de la simulación en la nube pública (Cloud). Toda la configuración necesaria para ejecutarla en Cloud se ha planteado utilizando la filosofía DevOps de infraestructura como código. Se introduce con esto la automatización del despliegue y puesta a punto de todos los pasos que conlleva la simulación. Se finaliza con un ejemplo de cómo el framework desarrollado permite realizar simulaciones de manera automática.
The deployment of sensors following an IoT (Internet of Things) architecture allows us to process large amounts of data in an orderly manner. These measurements are generally analyzed in the edge and fog layers and stored in the cloud layer to have a history of how the values have fluctuated in the past few years for subsequent studies. An example is the study of solar radiation, where a multitude of sensors are deployed to analyze the evolution of radiation in the different seasons of the year. However, there is no analysis of the deployment of these sensors in an IoT context. It is proposed to implement a simulated system within an IoT context where there are elements that generate radiation measurements, which are transmitted through events to the processing layer. This is responsible for detecting outliers, which are replaced by correct values using different regression methods. At the same time, kriging techniques are integrated to calculate radiation from other points where no measurements exist. To ensure that the simulation is distributed and scalable in terms of dimensions, it is taken to a public cloud provider. All the configuration required to run it in the Cloud has been approached using the DevOps philosophy of infrastructure as code. It is introduced with this the automation of the deployment and configuration of all the steps involved in the simulation.
The deployment of sensors following an IoT (Internet of Things) architecture allows us to process large amounts of data in an orderly manner. These measurements are generally analyzed in the edge and fog layers and stored in the cloud layer to have a history of how the values have fluctuated in the past few years for subsequent studies. An example is the study of solar radiation, where a multitude of sensors are deployed to analyze the evolution of radiation in the different seasons of the year. However, there is no analysis of the deployment of these sensors in an IoT context. It is proposed to implement a simulated system within an IoT context where there are elements that generate radiation measurements, which are transmitted through events to the processing layer. This is responsible for detecting outliers, which are replaced by correct values using different regression methods. At the same time, kriging techniques are integrated to calculate radiation from other points where no measurements exist. To ensure that the simulation is distributed and scalable in terms of dimensions, it is taken to a public cloud provider. All the configuration required to run it in the Cloud has been approached using the DevOps philosophy of infrastructure as code. It is introduced with this the automation of the deployment and configuration of all the steps involved in the simulation.
Description
Trabajo de Fin de Máster en Ingeniería Informática, Facultad de Informática UCM, Departamento de Arquitectura de Computadores y Automática, Curso 2020/2021.