Abstract: In recent years, virtualizing network services and functions has enabled optimizing hardware resources on resource-constrained devices, such as CPU, memory, and storage. Traditional virtualization is achieved through virtual machines using a layer known as a hypervisor. While this form of virtualization offers advantages such as scalability and portability, it has disadvantages in terms of performance compared to nonvirtualized deployments. In this context, alternative virtualization technologies (like containers) allow virtualization on the same physical infrastructure, improving overall performance, portability, and service scalability. This paper implements the deployment of network services on the Raspberry Pi development platform, which has limited resources. This is achieved through a multi-container virtualization solution using the Docker Compose tool, based on Docker containerization technology. Finally, a performance analysis of the implemented virtualization solution is conducted in terms of resource utilization by each service. Abstract years resourceconstrained constrained devices CPU memory storage hypervisor portability deployments context like containers infrastructure platform multicontainer multi container tool technology Finally

Resumen: En los últimos años, la virtualization de servicios y funciones de red ha permitido optimizar los recursos de hardware, como CPU, memoria y almacenamiento, en equipos con limitaciones de recursos. La virtualization tradicional se lleva a cabo mediante máquinas virtuales, utilizando una capa conocida como hipervisor. A pesar de que esta forma de virtualization ofrece ventajas como escalabilidad y portabilidad, presenta desventajas en términos de rendimiento en comparación con un despliegue no virtualizado. En este contexto, han surgido tecnologías alternativas de virtualization (como los contenedores) que permiten la virtualization en la misma infraestructura física, mejorando el rendimiento general, la portabilidad y la escalabilidad de los servicios. En este artículo se implementa el despliegue de servicios de red en la plataforma de desarrollo Raspberry Pi que cuenta con recursos limitados. Esto se logra mediante una solución de virtualization multicontenedor utilizando la herramienta Docker Compose, basada en la tecnología de contenerización Docker. Finalmente se lleva a cabo un análisis del rendimiento de la solución de virtualization implementada en términos de la utilization de recursos por parte de cada uno de los servicios. Resumen años hardware CPU almacenamiento virtuales hipervisor virtualizado contexto contenedores física general limitados Compose

Abstract. - In the field of medical care, hospitals face numerous challenges in effectively managing biomedical data. This can lead to a decrease in the efficiency of care, as not all institutions have efficient methods for managing such data. In addition to personal data, physiological constants, such as heart rate and oxygen levels, need to be constantly monitored in order to detect any changes. However, obtaining this data from different instruments and ensuring its constant recording can be problematic. To address these challenges, a system based on the Internet of Things (IoT) has been developed. This system utilizes sensors connected to ESP32 cards, which are in constant communication, to obtain physiological constants and other relevant data. A prototype has been designed, which includes sensors placed on the wrist to measure three physiological constants. The MAX30102 pulse sensor is used to measure blood oxygenation and heart rate. This sensor can be placed on the fingers, lobe, or wrist to obtain accurate readings. Additionally, the MLX90614 sensor is used for temperature acquisition. All the data collected by these sensors is managed by an ESP32 card, which acquires the information and sends it for further use. They employ communication protocols that enable the simultaneous reading of multiple sensors for the parallel monitoring of more than one patient, a capability not addressed in current prehospital care systems. To ensure constant monitoring of physiological constants, a master-slave configuration is utilized. Each slave module collects information from individual patients and sends it to a master card. The data is encrypted during transmission. These devices can be used in various healthcare settings, including prehospital care, and can be carried by the patients themselves. The collected data is then transmitted to a central system using the MQTT protocol. A master ESP32, connected to a Raspberry Pi 4, acts as the main console, where the data is centralized. Once the data is in the MQTT broker, it can be accessed and analyzed from various devices for traceability purposes. Real-time data recording is achieved by utilizing Google services, specifically Firebase, which stores the data in a database.

Resumen. - En la atención médica dentro de un hospital son múltiples los eventos que se presentan a diario, en donde el manejo de datos biomédicos es compartido en todas las áreas, lo que se torna complicado y no en todas las instituciones se tienen métodos adecuados para la gestión de estos datos, lo que provoca una disminución en la eficiencia de la atención. Además de los datos personales, las constantes fisiológicas son datos que deben llevar un monitoreo constante de lo contrario es un problema al tener signos cambiantes, estos datos se obtienen de diferentes instrumentos y se lleva el registro de una sola lectura sin ser constante. Debido a esto se desarrolla un sistema basado en el internet de las cosas, en el cual la obtención de las constantes fisiológicas y otros datos se realiza a través de sensores conectados a tarjetas ESP32 que están en constante comunicación. Para la adquisición de datos, se diseña un prototipo el cual se coloca en la muñeca con los sensores incluidos para la toma de 3 constantes fisiológicas. Se emplea el sensor de pulso MAX30102 que permite obtener la oxigenación de la sangre y la frecuencia cardiaca, para obtener la lectura del pulso y/o ritmo cardiaco se coloca el dispositivo en los dedos, lóbulo o muñeca. Además, se cuenta con el sensor MLX90614 para la adquisición de la temperatura. El manejo de datos se realiza por medio de una tarjeta ESP32 que adquiere la información de los sensores y la envía para su utilización. Se utilizan protocolos de comunicación que permite la lectura de múltiples sensores para la lectura paralela de más de un paciente, esto no se ha tratado en los sistemas actuales de atención prehospitalaria. El constante monitoreo independiente de las constantes fisiológicas se comunica a una central por medio de una configuración esclavos maestro, en donde cada esclavo obtiene la información de cada módulo en el paciente y es enviada a una tarjeta maestra, todos los datos son encriptados en cada envió, estos dispositivos pueden ser utilizados en atención prehospitalaria y dentro de los hospitales portados por el propio paciente. La concentración de los datos es recabada en una ESP32 maestra la cual se conecta de forma serial a la consola principal que es una Raspberry Pi 4, a través del protocolo MQTT, una vez que se tienen los datos en MQTT broker se pueden visualizar en diferentes dispositivos para la trazabilidad de estos datos. Los datos son registrados en una base de datos en tiempo real usando los servicios de Google en Firebase.

Abstract: This paper presents a prototype of an automobile driver assistance system based on YOLOv3. The system detects car types, traffic signs, and traffic lights in real-time and warns the driver accordingly. In the learning phase of the YOLO algorithm, the standard weights are learned first, followed by transfer learning to the objects of interest. The retraining phase uses 2,800 images obtained from the Internet of three countries of the real-life, and the testing phase uses real-time videos of Mexico City roads. In the validation phase, the proposed system achieves 95%, 37%, and 40% performance on the compiled dataset for the detection of road elements. The results obtained are comparable and in some cases better than those reported in previous works. Using a Raspberry Pi 4, the prototype was tested in real-life, generating visual and audible warnings for the driver, with an object recognition rate of 0.4 fps. A mean average precision (mAP) of 53% was reached by the proposed system. The experiments showed that the prototype achieved a poor recognition rate and required high computational processing for object recognition. However, YOLO is a model that can have good performance on low-resource hardware.

Abstract This work evaluates the feasibility of integrating lowcost Raspberry pi boards and ESP8266 microcontrollers with industrial equipment of Programmable Logic Controllers (PLC) in a decentralized network. These devices will be the nodes that participate in the exchange of data between the manufacturing floor and business services in a simple, reliable and economical way. The network nodes produce and consume data that is exchanged by an open-source protocol manager node called Node-RED. The protocol manager is a server with a Linux kernel on a RISC (Reduced instruction set computing) microprocessor. The sensors in the manufacturing nodes use SoC (System On Chip) microcontrollers and through the concept of Edge computing they acquire, process and send their data to the protocol manager. Thus, using the Iperf3, Wireshark and MTR tools, we will measure the performance of the communications link and the status of the participating nodes in their data production/consumption processes.

Resumen Este trabajo evalúa la viabilidad de integrar en una red descentralizada las placas de bajo costo Raspberry pi, microcontroladores ESP8266 con equipos industriales de Controladores Lógicos Programables (PLC). Estos dispositivos serán los nodos que participan en el intercambio de datos entre el piso de manufactura y los servicios empresariales de manera simple, fiable y económica. Los nodos de la red producen y consumen datos que son intercambiados por un nodo gestor de protocolos open source llamado Node-RED. El gestor de protocolos es un servidor con núcleo de Linux sobre un microprocesador RISC (Reduced instruction set computing). Los sensores en los nodos de manufactura utilizan microcontroladores SoC (System On Chip) y mediante el concepto de Edge computing adquieren, procesan y envían sus datos al gestor de protocolos. Así, mediante la herramienta Iperf3, Wireshark y MTR, mediremos el rendimiento del enlace de comunicaciones y el estado que guardan los nodos participantes en sus procesos producción/consumo de datos.

Resumen: Los Sistemas de Tiempo Real no críticos pueden desarrollarse integrando tres componentes básicos: una computadora de placa reducida Raspberry Pi, un sistema operativo con una distribución GNU/Linux y la extensión de tiempo real PREEMPT_RT; para su desarrollo, dimensionamiento y correcto funcionamiento, es necesario conocer el comportamiento de los tiempos de respuesta de sus tareas para cumplir con la condición de predecibilidad y cumplimiento de plazos. En este sentido, es necesario definir ciertas condiciones de operación como: tarea en estudio, prioridad, complejidad del algoritmo en ejecución, ésto con el objetivo de conocer la dinámica de la tarea en estudio y determinar su comportamiento.En este trabajo se presenta una memoria de Tesis Doctoral donde se desarrollaron dos modelos, el primero denominado ab initio y el segundo basado en el cociente de esparanzas matemáticas recursivas, a partir del cual se generaron dos algoritmos de estimación. Los tiempos de respuesta son generados por la tarea en estudio que consiste en la inversión de matrices de dimensiones 32×32, 64×64 y 128×128. La tarea está asociada a la política de planificación Round Robin con la mayor prioridad.

Abstract: Non-critical real-time systems can be developed by integrating three essential components: a Raspberry Pi single board computer, an operating system with a GNU/Linux distribution, and the PREEMPT_RT real-time kernel extension; for its development, sizing, and correct operation, it is necessary to know the behavior of the response times of its tasks to meet the condition of predictability and compliance with deadlines. For this, it is necessary to define certain operating conditions such as the task to be studied, the priority, and complexity of the algorithm in execution, to know the dynamics of the task under study and determine its behavior. This paper presents a report on Ph.D. Thesis where two models were developed, the first one called ab initio and the second one based on the recursive mathematical expectation quotient, from which two estimation algorithms were generated. The response times under study were obtained from a task of the inversión of matrices with dimensions: 32×32, 64×64, and 128×128. The task under study is associated with the Round Robin scheduling policy with the highest priority.

With the ever-growing traffic density, the number of road accidents has continued to increase. Finding solutions to reduce road accidents and improve traffic safety has become a top priority for Kiira Motors Corporation, a Ugandan state-owned automotive company. The company seeks to develop intelligent driver assistance systems for its market entry product, the Kayoola EVS bus. A machine learning-based driver monitoring system that would monitor driver drowsiness and send out an alarm in case drowsiness is detected has been developed in an attempt to reduce drowsiness-related accidents. The system consists of a camera positioned in such a way as to keep track of the driver's face. The camera is interfaced with a Raspberry Pi minicomputer which carries out the computations and analysis and when drowsiness is detected, an alarm is triggered. Dangerous driver behavior including distraction and fatigue has long been recognized as the main contributing factor in traffic accidents. This paper therefore presents the development of a driver monitoring system for the Kayoola Electric City Bus to address the increasing occurrences of road accidents. The machine learning-based driver monitoring system is designed to be non-intrusive with continuous real-time operation.

ABSTRACT The use of mechanistic plant growth models relies on the availability of high-quality inputs to reduce uncertainty in estimates. Measurements of photosynthetically active radiation inside a protected environment are either more expensive to obtain or dependent on assumptions regarding external measurements. This study aimed to reduce the influence of uncertainty in the measurements of low-cost lux meters by using a data assimilation strategy. We first determined, by simulation, the impact of different sensors on the estimates. We then used the Ensemble Kalman Filter to assimilate artificial observations of tomato growth in the Reduced-State Tomgro model, in simulations for which the solar radiation inputs were obtained from a low-cost lux meter. We compared the assimilated estimates to the simulations that used solar radiation obtained with a scientific-grade quantum sensor. For periods of larger radiation intensity, in which the differences in measurements from both instruments are larger, assimilation of observations with low errors lead to estimates that are closer to the ones obtained by scientific grade sensors. These results suggest that low-cost sensors could be used to obtain inputs for growth models in protected environments, provided there are also imperfect observations of the state. highquality high quality lowcost cost strategy determined simulation ReducedState Reduced State model meter scientificgrade sensor intensity environments state

ABSTRACT All horticultural regions of the world have fundamentally developed in the programmatic era through farmer impact and imaginative development practices. The methods described are used to monitor satisfaction and crop yields. Because of the quality of the soil and the nourishment of the land, their cultivation has improved and created more money. The disadvantage is that it took them a long speculation to acquire the crops, and the level of nutrition was not usually put at its obvious level. In addition, many areas were devoted to production, which required a lot of work to treat the whole area. In order to control the time and methods well, the majority of the regions switched to careful development principles with IoT structures. Growing on water is the most progressive strategy to grow natural plants, vegetables and fruits without using land. The use of Rockwool in agricultural strategies where water contamination is possible for a certain period will result in huge yields and the requirement of longer growing times will be waived. Most of the countries that have practiced smart and economic development with little external intervention. IoT sensors are used in the water cultivation development system to test the situation and quality of yields continuously. They will effectively provide information to the whole system when the water or nutrient level has dropped. In the beginning, the development of hydroponics was done horizontally in small spaces in order to maintain the water flow. Today, it is applied on a vertical structure to save space and water flows when needed. With this method, yields are likely to be achieved even more space-efficiently and with little external intervention. Vertical hydroponics performs better than previous conventional approaches; perhaps the farmers in the extension unit have considered the cost of the total layout. This evaluation paper describes the use of procedures and development of automated methods using IoT platforms. Many reference materials can be used big data. practices land money crops addition production area well structures plants waived intervention continuously dropped beginning flow Today needed method spaceefficiently efficiently approaches layout platforms data

Resumen Debido al creciente desarrollo tecnológico de los sistemas embebidos y el internet de las cosas (IoT), los dispositivos basados en microcontroladores se aplican cada vez más en diversas áreas del conocimiento. En este contexto, las plataformas y productos educativos online se consideran laboratorios virtuales remotos, ya que los estudiantes pueden acceder a los dispositivos físicos en cualquier lugar siempre que dispongan de un sistema informático conectado a internet. En este sentido, este trabajo describe el diseño y desarrollo de una plataforma con cuatro productos educativos en línea para la enseñanza de microcontroladores e IoT. Estos productos educativos están basados en software de código abierto, lo que permite su distribución gratuita en línea y el acceso a ellos desde un servidor en la nube. El enfoque de código abierto y multiplataforma (Linux, Windows® y Mac OSX) permite una mayor interacción y accesibilidad del usuario. Los productos educativos en línea hacen posible la programación del firmware del microcontrolador ESP32 de forma remota a través de OTA (Over the Air) y de sistemas embebidos Linux basados en Raspberry Pi (Rpi), permitiendo aplicaciones virtuales de laboratorio de microcontroladores. Además, los productos educativos en línea permiten manipular los pines vía internet a través de la interfaz gráfica de los microcontroladores ESP32 y ESP8266, así como del Rpi. Como resultado, se probó la plataforma en línea propuesta, ejecutándola en un servidor en la nube, y se validaron los cuatro productos educativos de aprendizaje a distancia y la aplicación real de microcontroladores e IoT, que se encuentran operativos, tal como fueron diseñados.

Abstract Due to the increasing technological development in embedded systems and the Internet of Things (IoT), devices based on microcontrollers are being increasingly applied in various areas of knowledge. In this context, online educational platforms and products are considered virtual remote laboratories, because students can access the physical devices anywhere as long as they have a computer system connected to the internet. In this sense, this work describes the design and development of a platform with four online educational products for teaching microcontrollers and IoT. These educational products are based on open-source software, which enables their free online distribution and the access to them from a cloud server. The open-source and multiplatform (Linux, Windows®, and Mac OSX) approach enables a greater user interaction and accessibility. The online educational products make possible the remote programming of the ESP32 microcontroller firmware via OTA (over the air) and Linux embedded systems based on Raspberry Pi (Rpi), enabling virtual microcontroller laboratory applications. In addition, the online educational products enable the manipulation of GPIO pins via the Internet through the graphical interface of the ESP32 and ESP8266 microcontrollers and the Rpi. As a result, the proposed online platform was tested, running it on a cloud server, and the four-distance learning educational products and the actual microcontroller and IoT application have been validated, and are operating as designed.

Resumen Este documento presenta una prueba de rendimiento creada para evaluar el desempeño de diferentes plataformas en la ejecución de un algoritmo de reducción de niebla basado en el “Dark Channel Prior” (DCP) (1). El parámetro utilizado para la evaluación fue el número de cuadros por segundo (FPS, por sus siglas en inglés) que el dispositivo es capaz de procesar. Con esta herramienta se logra determinar aquellas arquitecturas que son aptas para ejecutar el algoritmo en tiempo real. El ambiente de pruebas se ejecutó en cuatro plataformas, un Google Pixel 3a, una Raspberry Pi 3B+, una GPU de NVIDIA y un procesador Intel x86. Se usaron los siguientes kits de desarrollo de software (SDK, por sus siglas en inglés) según la plataforma: Android NDK, Yocto Poky, CUDA y la cadena de herramientas GCC. La herramienta permitió recopilar, para cada plataforma, los FPS para distintos tamaños de imagen, con estos resultados se pueden escoger la arquitectura más idónea según el área de implementación (e.g., bajo consumo o HPC).

Abstract In this document we present a benchmark to evaluate the performance of different platforms in the execution of a dehazing algorithm based on the Dark Channel Prior (DCP) (1). The parameter used for the evaluation was the number of frames per second (FPS) that the device was able to process. This tool allows to determine which architectures can execute the algorithm in real time. The testing environment was executing in four platforms, a Google Pixel 3a, a Raspberry Pi 3B+, a GPU by NVIDIA, and an Intel x86 processor. The following software development kits (SDK’s) where used for each of the platforms: Android NDK, Yocto Poky, CUDA, and the GCC toolchain. The tool allowed us to collect, for each platform, the FPS for different image sizes, these results allow the selection of an ideal architecture depending on a specific application (e.g., low power, HPC).

Sustainable analytical methods are highly demanded in the modern society. Within the green principles, novel procedures that attend the sustainable development goals have been proposed and the internet of things (IoT) can play key role to achieve this goal. The association of IoT with analytical chemistry enables the real-time obtaining of analytical data to control industrial processes as well as for monitoring different environmental scenarios and human health by accessing remotely analytical information. In this context, this review presents the main IoT technologies used for analytical chemistry as well as a tutorial for beginners in the field.

Resumen Se propone un sistema de medición y monitoreo automático de la temperatura en experimentos termosolares, basados en un computador de placa única - Raspberry Pi y un sistema de sensores. Los resultados de diferentes conjuntos de datos obtenidos por sensores, termopares y termómetros son comparados y se verifica la consistencia del sistema automático de medición de la temperatura.

Abstract For automatic temperature measuring and logging purposes in solar thermal experiments, a single board computer - Raspberry Pi based and sensor system is proposed. Different data sets are observed for sensors, thermometers, and thermocouples and they are compared. The consistency of the automated temperature measurement system is also verified.

ABSTRACT Indoor occupancy estimation is a process that contributes to maintaining quality standards in areas, and, nowadays, it serves as a reference to identify possible foci of infectious respiratory diseases contagion. This paper presents a tool for indoor occupancy estimation with Simple Mail Transfer Protocol notification, using a Raspberry Pi embedded board. The system is presented as an alternative to conventional occupancy estimation systems by measuring CO2 levels in the area. The method is based on image processing by applying the background subtraction technique using Python programming language. Initially, the area where the system is tested is characterized, and preprocessing, filtering, and thresholding stages are applied, as well of notification via e-mail for SMTP. The developed system is compared with a CO2 measurement system by applying a prioritization matrix comparing factors such as detection time, hit rate, and implementation costs. The proposed method presented better performance in the totality of the comparison parameters, with a prioritization of 87.972%. Basing the system in open-source software tools and high level and low-cost hardware tools allows for the system to be replicated and implemented on a large scale in controlled environments.

RESUMEN La estimación de ocupación en espacios interiores es un proceso que contribuye a mantener estándares de calidad en las zonas, y que, al día de hoy, sirve como referencia para identificar posibles focos de contagio de enfermedades respiratorias infecciosas. Este artículo presenta una herramienta para la estimación de ocupación en espacios interiores con notificación mediante el Protocolo Simple de Transferencia de Correo (SMTP) usando una placa embebida Raspberry Pi. El sistema se presenta como alternativa a los sistemas convencionales de estimación de ocupación midiendo los niveles de CO en la zona. El método se basa en procesamiento de imágenes aplicando la técnica de sustracción de fondo mediante lenguaje de programación Python. Inicialmente, se caracteriza la zona donde se prueba el sistema, y se aplican etapas de preprocesamiento, filtrados y umbralización, además de notificación vía correo electrónico por SMTP. El sistema desarrollado se compara con un sistema de medición de CO2 aplicando una matriz de priorización comparando factores como el tiempo de detección, tasa de aciertos y costos de implementación. El método propuesto presentó mejor rendimiento en la totalidad de los parámetros de comparación, con una priorización de 87.972 %. Basar el sistema en herramientas de software de código abierto y herramientas de hardware de alto nivel y bajo costo permite replicar e implementar el sistema a gran escala en ambientes controlados.

Abstract: This study presents the design of a face mask detector that uses a camera, controlled by Raspberry PI with Python, which generates alerts when detecting people not properly wearing face masks during the COVID-19 pandemic. The system uses a security camera and GSM (Global System for Mobile Communication) wireless communication. The face mask detector equipped with artificial intelligence is implemented in the "La Joya" building at the Salesian Polytechnic University (Ecuador). The learning model is developed in stages. The results show that at stage 20, the accuracy and the test accuracy of the model reaches one. Similarly, data loss achieves zero at stage 20. It is concluded that the face mask detection system successfully monitors a specific area, improving the safety of students attending classes during the COVID-19 pandemic.

Resumen: En este estudio se presenta el diseño de un detector de mascarilla por medio de una cámara controlada por un Raspberry PI con Python que genera alertas en caso de detectar a personas que no usen correctamente la mascarilla durante la pandemia de COVID-19. Se utiliza una cámara de seguridad y comunicación inalámbrica a través de GSM (Global System for Mobile Communication). El detector con inteligencia artificial se implementa en el edificio “La Joya” de la Universidad Politécnica Salesiana (Ecuador). El aprendizaje del modelo se realiza por épocas. Los resultados muestran que en la época 20, la precisión y prueba de exactitud del modelo llegan a 1. De igual forma, la pérdida (Loss) de datos llega en la época 20 a tener una pérdida de datos de 0. Se concluye que el sistema de detección de mascarilla permite monitorear un área determinada, proporcionando un mejor cuidado de los estudiantes durante la pandemia de COVID-19.

Resumen El controlador Proporcional Integral Derivativo (PID) ha sido por décadas el controlador más usado en la industria debido a sus diversas ventajas, siendo una de las más importantes la relación costo/beneficio que puede proveer, además de que es una metodología que se ha ido adaptando a las nuevas tecnologías a través de los años. Actualmente con el auge de la industria 4.0 es imperativo proporcionar herramientas educacionales asequibles que permitan adentrar a los educandos en estas tecnologías. La implementación de un controlador PID en un servidor OPC UA permite eliminar las barreras de propietario y eficientizar los procesos productivos a controlar y monitorear. En este trabajo se presenta la metodología de implementación de un control PID didáctico con un servidor OPC UA montado en una Raspberry Pi, que da la oportunidad de introducir a los jóvenes universitarios a un entorno de industria 4.0 para que en su futuro profesional logren una rápida adaptación al sector productivo.

Abstract The Proportional Integral Derivative (PID) controller has been the most widely used controller in the industry for decades due to its various advantages, one of the most important being the cost/benefit ratio it can provide, in addition to the fact that it is a methodology that has been adapted to new technologies over the years. Currently, with the rise of industry 4.0, it is imperative to provide affordable educational tools that allow students to delve into these technologies. Implementation of a PID control in an OPC UA server allows to eliminate proprietary barriers and streamline monitoring and controlling in production processes usisng it. This paper presents the methodology for implementing a didactic PID control with an OPC UA server mounted on a Raspberry Pi, which gives the opportunity to introduce young university students to an industry 4.0 environment so that in their professional future they can achieve rapid adaptation to the productive sector.