Adaptive Scalable SVD Unit for Fast Processing of Large LSE Problems

Singular Value Decomposition (SVD) is a key linear algebraic operation in many scientific and engineering applications. In particular, many computational intelligence systems rely on machine learning methods involving high dimensionality datasets that have to be fast processed for real-time adaptability. In this paper we describe a practical FPGA (Field Programmable Gate Array) implementation of a SVD processor for accelerating the solution of large LSE problems. The design approach has been comprehensive, from the algorithmic refinement to the numerical analysis to the customization for an efficient hardware realization. The processing scheme rests on an adaptive vector rotation evaluator for error regularization that enhances convergence speed with no penalty on the solution accuracy. The proposed architecture, which follows a data transfer scheme, is scalable and based on the interconnection of simple rotations units, which allows for a trade-off between occupied area and processing acceleration in the final implementation. This permits the SVD processor to be implemented both on low-cost and highend FPGAs, according to the final application requirements.

Optimisation of the first principle code Octopus for massive parallel architectures: application to light harvesting complexes

150 p.

Asymmetric flow networks

This paper provides a new model of network formation that bridges the gap between the two benchmark models by Bala and Goyal, the one-way flow model, and the two-way flow model, and includes both as particular extreme cases. As in both benchmark models, in what we call an "asymmetric flow" network a link can be initiated unilaterally by any player with any other, and the flow through a link towards the player who supports it is perfect. Unlike those models, in the opposite direction there is friction or decay. When this decay is complete there is no flow and this corresponds to the one-way flow model. The limit case when the decay in the opposite direction (and asymmetry) disappears, corresponds to the two-way flow model. We characterize stable and strictly stable architectures for the whole range of parameters of this "intermediate" and more general model. We also prove the convergence of Bala and Goyal's dynamic model in this context.

Open Personalization: Involving Third Parties in Improving the User Experience of Websites

Traditional software development captures the user needs during the requirement analysis. The Web makes this endeavour even harder due to the difficulty to determine who these users are. In an attempt to tackle the heterogeneity of the user base, Web Personalization techniques are proposed to guide the users’ experience. In addition, Open Innovation allows organisations to look beyond their internal resources to develop new products or improve existing processes. This thesis sits in between by introducing Open Personalization as a means to incorporate actors other than webmasters in the personalization of web applications. The aim is to provide the technological basis that builds up a trusty environment for webmasters and companion actors to collaborate, i.e. "an architecture of participation". Such architecture very much depends on these actors’ profile. This work tackles three profiles (i.e. software partners, hobby programmers and end users), and proposes three "architectures of participation" tuned for each profile. Each architecture rests on different technologies: a .NET annotation library based on Inversion of Control for software partners, a Modding Interface in JavaScript for hobby programmers, and finally, a domain specific language for end-users. Proof-of-concept implementations are available for the three cases while a quantitative evaluation is conducted for the domain specific language.

Contributions to unsupervised and supervised learning with applications in digital image processing

311 p. : il.

FTT-MA: A Flexible Time-Triggered Middleware Architecture for Time Sensitive, Resource-Aware AmI Systems

There is an increasing number of Ambient Intelligence (AmI) systems that are time-sensitive and resource-aware. From healthcare to building and even home/office automation, it is now common to find systems combining interactive and sensing multimedia traffic with relatively simple sensors and actuators (door locks, presence detectors, RFIDs, HVAC, information panels, etc.). Many of these are today known as Cyber-Physical Systems (CPS). Quite frequently, these systems must be capable of (1) prioritizing different traffic flows (process data, alarms, non-critical data, etc.), (2) synchronizing actions in several distributed devices and, to certain degree, (3) easing resource management (e.g., detecting faulty nodes, managing battery levels, handling overloads, etc.). This work presents FTT-MA, a high-level middleware architecture aimed at easing the design, deployment and operation of such AmI systems. FTT-MA ensures that both functional and non-functional aspects of the applications are met even during reconfiguration stages. The paper also proposes a methodology, together with a design tool, to create this kind of systems. Finally, a sample case study is presented that illustrates the use of the middleware and the methodology proposed in the paper.

Sliding mode position control for real-time control of induction motors

A sliding mode position control for high-performance real-time applications of induction motors in developed in this work. The design also incorporates a simple flux estimator in order to avoid the flux sensors. Then, the proposed control scheme presents a low computational cost and therefore can be implemented easily in a real-time applications using a low cost DSP-processor. The stability analysis of the controller under parameter uncertainties and load disturbances in provided using Lyapunov stability theory. Finally, simulated and experimental results show that the proposed controller with the proposed observer provides a good trajectory tracking and that this scheme is robust with respect to plant parameter variations and external load disturbances.

Control de un sistema de accionamientos de traslación basado en correa para un manipulador de cinemática paralela

[ES]Este Trabajo de Fin de Grado “Control de un sistema de accionamientos de traslación basado en correa para un manipulador de cinemática paralela” tiene como objetivo principal la implementación de un sistema de control que nos permita manejar un manipulador de cinemática paralela de dos grados de libertad accionado mediante dos motores eléctricos de corriente continua. Como componente central de este sistema de control, se dispondrá de un ordenador portátil cuyo procesador será el encargado de ejecutar las acciones necesarias para que pueda llevarse a cabo esta actividad de control. De esta forma, la tarea más importante y laboriosa a llevar cabo en este proyecto será el desarrollo de un aplicación de control que, corriendo en el citado ordenador, permitirá al usuario manejar el manipulador de cinemática paralela en cuestión. Para ello, esta aplicación deberá ser capaz de interpretar las ordenes de movimiento dadas por el usuario y transmitirlas al procesador del mencionado ordenador. Además de todo lo anterior, para completar el desarrollo del sistema de control, será necesaria la implementación de diversos sensores que se encargarán de detectar y transmitir las señales necesarias para evitar situaciones de emergencia en el que el manipulador estuviese a punto de chocar con algún objeto o persona. En conclusión, mediante el cumplimiento de los objetivos de este Trabajo de Fin de Grado, se va a disponer de un sistema de control sencillo, intuitivo y fácilmente operable, que va a permitir a cualquier futuro usuario del mismo el manejo de un robot de cinemática paralela.

Design and evaluation of an optical switching node

[EN]For a good development of elastic optical networks, the design of flexible optical switching nodes is required. This work analyses the previously proposed flexible architectures and, based on the most appropriate, which is the Architecture on Demand (AoD), proposes a specific configuration of the node that includes spatial and spectral switching and the wavelength conversion functionality with a low blocking probability and the minimum amount of modules; the characteristics of the traffic that the designed node is able to cope with are specified in the last chapter. An evaluation of the designed node is also done, and, compared to the other architectures, it is shown that the Architecture on Demand gives better results than others and that it has a higher potential for future developments.

Antikörper: Diseño e implementación de un sistema de detección de intrusos inalámbrico basado en Network IDS

[ES]Este proyecto, fruto de mi Trabajo Fin de Grado, pretende describir los aspectos más relevantes de los Wireless Intrusion Detection System (WIDS), tales como su historia, funcionamiento, arquitecturas, implementación o futuro. Tras esto, se expondrá el objetivo principal del proyecto. Proponer una arquitectura genérica para los WIDS y crear Antikörper, un WIDS basado en Network IDS completamente funcional, que cubra totalmente las necesidades de seguridad actuales en las redes Wireless LAN y sea adaptable a futuras revisiones.

Radio-aware service-level scheduling to minimize downlink traffic delay through Mobile Edge Computing

One of the most challenging problems in mobile broadband networks is how to assign the available radio resources among the different mobile users. Traditionally, research proposals are either speci c to some type of traffic or deal with computationally intensive algorithms aimed at optimizing the delivery of general purpose traffic. Consequently, commercial networks do not incorporate these mechanisms due to the limited hardware resources at the mobile edge. Emerging 5G architectures introduce cloud computing principles to add flexible computational resources to Radio Access Networks. This paper makes use of the Mobile Edge Computing concepts to introduce a new element, denoted as Mobile Edge Scheduler, aimed at minimizing the mean delay of general traffic flows in the LTE downlink. This element runs close to the eNodeB element and implements a novel flow-aware and channel-aware scheduling policy in order to accommodate the transmissions to the available channel quality of end users.

Deep learning review and its applications

Deep neural networks have recently gained popularity for improv- ing state-of-the-art machine learning algorithms in diverse areas such as speech recognition, computer vision and bioinformatics. Convolutional networks especially have shown prowess in visual recognition tasks such as object recognition and detection in which this work is focused on. Mod- ern award-winning architectures have systematically surpassed previous attempts at tackling computer vision problems and keep winning most current competitions. After a brief study of deep learning architectures and readily available frameworks and libraries, the LeNet handwriting digit recognition network study case is developed, and lastly a deep learn- ing network for playing simple videogames is reviewed.

A Real-Time Sliding Mode Control for a Wind Energy System Based on a Doubly Fed Induction Generator

In this paper, a real time sliding mode control scheme for a variable speed wind turbine that incorporates a doubly feed induction generator is described. In this design, the so-called vector control theory is applied, in order to simplify the system electrical equations. The proposed control scheme involves a low computational cost and therefore can be implemented in real-time applications using a low cost Digital Signal Processor (DSP). The stability analysis of the proposed sliding mode controller under disturbances and parameter uncertainties is provided using the Lyapunov stability theory. A new experimental platform has been designed and constructed in order to analyze the real-time performance of the proposed controller in a real system. Finally, the experimental validation carried out in the experimental platform shows; on the one hand that the proposed controller provides high-performance dynamic characteristics, and on the other hand that this scheme is robust with respect to the uncertainties that usually appear in the real systems.

Contributions to the efficient use of general purpose coprocessors: kernel density estimation as case study

142 p.

Estudio de señales cerebrales y aplicación a cojín elevador para discapacitados.

[ES]Estudio de obtención de señales cerebrales, creación de sensores de electroencefalograma y circuito que procesa frecuencias cerebrales para control de un prototipo de cojín elevador que ayude a personas discapacitadas a levantarse de un asiento con el fin de dar los primeros pasos en la creación de un exoesqueleto controlado cerebralmente.