TFG: Development of an Artificial Intelligent system based on low-resource Edge Computing for autonomous vehicles applications

Nowadays, there are plenty of IoT devices which make our everyday life easier thanks to their intelligent tasks: data capture, process automation… However, the increase of these devices is turning out to be somehow risky in terms of latency or bandwidth. That is the reason why some alternatives that may solve these problems are being searched, and one of them is Edge Computing technologies.

Edge Computing devices are those who are able to process the information captured without connecting to the network. Due to that, the latency and bandwidth issues that may occur can be significantly reduced, allowing the radio spectrum to decrease its saturation as well as improving the latency and consumption performance.

In this project, the main goal is to create a system that is able to develop and execute Artificial Intelligence algorithms designed for autonomous driving and assistance to the driver, always taking care of Edge Computing philosophy. In order to do that, we have used Google Coral, a hardware platform that perfectly adapts to our needs, allowing us to develop all the Edge Computing algorithms as well as offering appropriate consumption and processing characteristics.

Finally, we have tested our system in a real situation, evaluating the quality of the results as well as the resources used (latency, bandwidth…) and the advantages and disadvantages in relation to the existing technologies is this area. After these experiments, we have concluded that the quality of our Edge Computing system is enough to carry out the tasks it has been designed for. Also, all the resources used have been optimized in relation to Cloud Computing alternatives, turning this project into a faster, more effcient and economic alternative.

TFM: Design and implementation of an intelligent system for eldercare support

The number of elderly people in Spain is in a constant process of growth. It is frequent that in this age range problems appear that do not allow people to live their daily lives in a totally independent way. The health care system has limited resources to solve these problems, and the families of these people are having problems reconciling their work and family life.

The motivation for this Master’s Thesis has been to design a solution for people in this situation. In order to do this, a study of the technological solutions present on the market has been carried out first to obtain a list of functionalities that help to deal with the problem. Then, an architecture has been designed that allows the incorporation of these functionalities. Finally, a proof of concept for one of them has been implemented.

The designed architecture is composed of three elements. The terminal is placed at the patient’s home, and provides the patient with different utilities. The web client allows the caregiver to access the different functionalities that arise from monitoring the patient from the terminal. The server is responsible for managing communications between the terminal and the client, as well as managing access to the architecture’s resources, such as the database.

To implement the proof of concept, the utility that has been chosen is the dispensing of medicines. On the patient side, the terminal will alert you when it is time to take the medication, making it accessible at that time. The dispenser has a sensor on the lid that allows you to know when the medication has been accessed, allowing you to monitor your medication intake.

On the caregiver’s side, the status of the intake can be checked through a web client. Once the caregiver has logged into the application, he or she can access the associated terminal and check whether the programmed intakes have been taken in the time interval defined for them, or whether they have been forgotten. On the other hand, the web client also allows the caregiver to schedule the intakes from the terminal.

TFG: Implementation and evaluation of ANNs in microcontroller-based systems

Nowadays, artificial neural networks (ANNs) are computational models that, while they have solved many different problems, they require a large amount of memory to execute those solutions. Therefore, their implementation is more common in systems with high-performance capabilities, such as data centers or servers. However, there is an increasing interest in developing these solutions on devices with fewer resources, such as personal computers, mobile devices, and microcontrollers.  This situation has led to the proliferation of multiple techniques and tools to reduce the requirements of these models and allow their implementation on those platforms.

The main objective of this project is the evaluation of a tool for the implementation of artificial neural networks in microcontroller-based systems. For this purpose, a practical use case has been defined.

To achieve this objective, the classification of types of tremors in Parkinson’s patients was chosen as the practical use case. In addition, the Arduino Nano 33 BLE Sense has been also chosen as the microcontroller-based platform to implement the solutions of the practical use case.

Later, a dataset has been generated with real measurements from a sensor on that platform. With this dataset various experiments have been carried out to determine how the different structures of artificial neural networks deal with the chosen use case.

Then, based on those experiments, some appropriate ANNs for classifying types of tremors have been designed, trained, and evaluated in a system with graphics and tensor processing units and in a microcontroller-based system.

Finally, based on the results obtained, the trade-offs between classification accuracy, memory footprint and inference latency involved in the implementation of ANNs solutions in microcontroller-based systems have been determined. In addition, a discussion of the strengths and weaknesses of the selected tool for the implementation of artificial neural networks in microcontroller-based systems has been presented.

Gated Recurrent Unit Neural Networks for Automatic Modulation Classification With Resource-Constrained End-Devices

The article “Gated Recurrent Unit Neural Networks for Automatic Modulation Classification With Resource-Constrained End-Devices” by our lab member Ramiro Utrilla has just been published in the IEEE Access, a high-impact open-access journal.

This work has been carried out in collaboration with researchers from the CONNECT – Centre for Future Networks and Communications in Dublin (Ireland), where Ramiro carried out a research stay of 3 months.

In this article, they focus on the Automatic Modulation Classification (AMC). AMC is essential to carry out multiple CR techniques, such as dynamic spectrum access, link adaptation and interference detection, aimed at improving communications throughput and reliability and, in turn, spectral efficiency. In recent years, multiple Deep Learning (DL) techniques have been proposed to address the AMC problem. These DL techniques have demonstrated better generalization, scalability and robustness capabilities compared to previous solutions. However, most of these techniques require high processing and storage capabilities that limit their applicability to energy- and computation-constrained end-devices.

In this work, they propose a new gated recurrent unit neural network solution for AMC that has been specifically designed for resource-constrained IoT devices.

The proposed GRU network model for AMC.

They trained and tested their solution with over-the-air measurements of real radio signals, which were acquired with the MIGOU platform.

Dataset generation scenario set up.

Comparison of signals recorded at (a) 1 and (b) 6 meters. The signals in the bottom row are the normalized version of those in the top row.

Their results show that the proposed solution has a memory footprint of 73.5 kBytes, 51.74% less than the reference model, and achieves a classification accuracy of 92.4%.

Increasing the training set can lead to improvements in the performance of a model without increasing its complexity. These improvements allow developers to reduce the complexity of the model and, therefore, the device resources it requires. However, longer training processes can lead to fitting and gradient problems. These tradeoffs should be explored when developing neural network-based solutions for resource-constrained end-devices.

Research visit at UC Berkeley

Our lab member Alba Rozas has recently completed a PhD research visit at the Berkeley Wireless Research Center (BWRC), part of the University of California at Berkeley, under the supervision of renowned Profesor Jan Rabaey. This group carries out world-leading research in the fields of radio communications and wireless electronics, with a particular recent interest in Body Area Networks and the Human Intranet.

World Map with location pins for every BWRC member

The main research line of Alba’s PhD is focused on QoS-aware and energy efficient routing strategies for WSNs. This 5-month stay took place within the last stage of her PhD, and its main goal was the study and research of over-the-body communication aspects and the Human Intranet paradigm. These fields have unique characteristics that differentiate them from traditional WSNs, presenting new challenges and opportunities. Thus, in this research stay, Alba has focused on applying the core ideas of her PhD to the field of body area networks. As a result of the work carried out during the visit, she has ultimately developed an energy-efficient and QoS-aware strategy for on-body wireless communication, based on dynamic human activity detection.

In addition to this main research objective, the stay has also strengthened the already existing working relationship between BWRC and B105. Both labs are already collaborating in the development of health-related solutions and systems, initiated by Alvaro Araujo‘s two research visits at BWRC.

Applied Science: Special Issue “Wireless Sensor Networks: Technologies, Applications, Prospects”

The main objective of this Special Issue is to provide a common space for WSNs researchers to share their high quality research and outcomes, and disseminate them to the rest of the world. The topics include novel designs,
developments, and management of smart systems with a focus on new applications. In addition to these, notable advancements in the performance of WSN are welcome.

El B105 participa en el Reto ONCE de Teleco Emprende

Este pasado miércoles tuvo lugar el concurso Reto ONCE de Teleco Emprende. Desde el B105 presentamos una propuesta en el ámbito de la realidad aumentada, vía tecnología SLAM (Simultaneous Localization and Mapping).

La propuesta es una aplicación móvil orientada a usuarios que aún mantienen un cierto grado de visión, p.e. en casos de miopía magna, pérdida de la visión periférica, etc. Esta aplicación ofrecería :

  • Mejora de la visión residual, adaptando la imagen del escenario a un formato más fácilmente perceptible, y filtrando cualquier elemento no deseado.
  • Guía a un punto destino en escenarios previamente modelados en 3D, sin necesidad de instalación previa.
  • Información de puntos de interés, ofreciendo una interfaz intuitiva y accesible a bases de datos de terceros (p.e. Google Places).

A continuación os mostramos un vídeo de ejemplo de la primera funcionalidad. En este caso mostramos únicamente los elementos marcados por el usuario y con un ángulo de visión más amplio:

Desde el B105 queremos agradecer al equipo gestor de este evento, y a la ONCE por su confianza en el potencial de los alumnos de la ETSIT. ¡Esperamos que vuelvan pronto!

Visita del CIDAT

Como podrán apreciar los lectores frecuentes de nuestro blog, las lineas de investigación de nuestro grupo abordan temáticas muy diversas. La que trataremos en este artículo está orientada mejorar la percepción espacial por medios no visuales. Para ello, utilizamos una red de dispositivos “wearables” que generan estímulos hápticos y acústicos de acuerdo a teorías recientes en materia de sustitución sensorial.

El objetivo principal es que una persona con discapacidad visual grave o ceguera tenga menos dificultades a la hora de desplazarse por la ciudad, en interiores, etc. Nuestro primer prototipo, Virtually Enhanced Senses (VES), virtualiza las características más importantes de un escenario real desde una perspectiva de orientación y movilidad, y proporciona la información al usuario de forma intuitiva.

Recientemente hemos tenido la suerte de contar con personal del CIDAT para la evaluación y posterior perfeccionamiento del prototipo. Durante las reuniones y demostraciones de la tecnología, los usuarios finales pudieron experimentar de primera mano el sistema tanto en escenarios virtuales como reales.

Desde aquí queríamos agradecer a nuestros invitados por su tiempo, esfuerzo e ilusión, esperando vernos de nuevo en un futuro próximo.

El B105 participa en ITEMAS

Nuestro laboratorio pasa a formar parte de la plataforma ITEMAS. La Plataforma de Innovación en Tecnologías Médicas y Sanitarias (ITEMAS) es una estructura de apoyo a la innovación sanitaria promovida por el Instituto de Salud Carlos III (ISCIII). Su objetivo es facilitar que las ideas innovadoras de los profesionales sanitarios lleguen a generar valor para el sistema, a través de favorecer la  transferencia de tecnología, la cultura de la innovación y la comunicación con el resto de la sociedad.

Desde el B105 queremos contribuir tenemos una gran experiencia en labores de I+D+i en el diseño de sistemas electrónicos en diferentes ámbitos. Uno de los ámbitos que más ha trabajado, especialmente en los últimos tiempos, es el sanitario. Desde nuestro punto de vista es fundamental contar con el usuario de la tecnología desde antes incluso de la idea del proyecto, por lo que la sinergia con los miembros de la plataforma es imprescindible tanto para la buena consecución de los proyectos como para su paso al tejido empresarial.

A Methodology for Choosing Time Synchronization Strategies for Wireless IoT Networks

This summer we have published a new article about time synchronization for wireless sensor networks, applied to the field of IoT, in Sensors Open Access Journal. This journal has these statistics:

  • 2018 Impact Factor: 3.031
  • 5-year Impact Factor: 3.302
  • JCR category rank: 15/61 (Q1) in ‘Instruments & Instrumentation’

This article belongs to the Special Issue Topology Control and Protocols in Sensor Network and IoT Applications.

This article has a direct relationship with the thesis of our colleague Francisco Tirado-Andrés. This thesis investigates a methodology, and associated tools, to make it easier for all researchers to choose time synchronization protocols for specific WSNs.

For more information about this article please visit MDPI webpage.