Publication: Nueva metodología para el endurecimiento óptimo de sistemas digitales con distribución de la funcionalidad, trabajando en entornos sometidos a la radiación ionizante
Loading...
Identifiers
Publication date
2016-06
Defense date
2016-07-19
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Debido al extenso uso de los sistemas digitales distribuidos en las diferentes áreas tecnológicas sometidas a entornos agresivos, como por ejemplo la automoción o el espacio, actualmente es de vital importancia la aplicación de técnicas de endurecimiento ante los efectos de radiación ionizante. Esta tesis se inició con el proyecto de investigación RENASER+ “TEC2010-22095-C03-03”, financiado por el Ministerio Español de Ciencia y Tecnología. El objetivo principal del subproyecto liderado por la Universidad Carlos III de Madrid, consistía en la evaluación de la sensibilidad de los sistemas digitales a los efectos de la radiación ionizante. Para ello, se planificaron una serie de tareas destinadas a establecer un protocolo y desarrollar un sistema de pruebas válido para realizar campañas de irradiación en los aceleradores instalados en el Centro Nacional de Aceleradores de la Universidad de Sevilla (CNA). Con el fin de alcanzar los objetivos de este proyecto, en los trabajos de la tesis doctoral, se estudiaron y aplicaron técnicas de tolerancia a fallos para sistemas distribuidos, y se comprobó la sensibilidad del sistema con el método basado en la inyección de los fallos por emulación. En concreto, utilizando la herramienta de Emulación Autónoma desarrollada en el grupo DMA de la Universidad Carlos III. A lo largo del desarrollo de la tesis doctoral, en concordancia con las tareas y objetivos del proyecto RENASER+, se han introducido mejoras en el sistema de inyección de fallos transitorios por emulación, para permitir la inyección de fallos múltiples en los elementos de memoria y para permitir el análisis detallado del efecto acumulado de los fallos detectados mediante técnicas de mitigación de fallos, y técnicas de rastreo del fallo y su propagación. Con el objetivo de generar un método integral de análisis, validación y calificación de los sistemas distribuidos robustos, además de inyectar fallos por emulación se indujeron fallos reales en dichos sistemas, mediante el uso del acelerador de partículas tipo ciclotrón del CNA de Sevilla. La evaluación de la sensibilidad del sistema distribuido prototipado requirió una cualificación mediante varias campañas de irradiación, en esta tesis se ha propuesto un método de planificación y realización de una campaña que permite la monitorización continúa del efecto de los fallos y la generación automática de informes finales del proceso completo. En la realización de la tesis doctoral se planteó un objetivo más ambicioso, que consistía en la generación de un método de comprobación continua, durante el tiempo de funcionamiento normal de los sistemas distribuidos, de la presencia de fallos que pudieran causar un mal funcionamiento, e incluso averías graves. Así, se amplió el método de calificación de la robustez de sistemas distribuidos a un método de test on-line de dichos sistemas. Durante el estudio de la robustez de los sistemas distribuidos, principalmente para aplicaciones terrestres, se identificó y estudió el problema de los fallos debidos a envejecimiento de los componentes electrónicos digitales. Por lo tanto, el método propuesto es también adecuado para la realización del test on-line durante el funcionamiento normal del sistema distribuido, para la detección de los fallos procedentes de la radiación ionizante (fallos transitorios) y los fallos debidos al envejecimiento de los dispositivos (fallos permanentes).
Los sistemas digitales distribuidos, aparte de los nodos que los componen, tienen un protocolo de comunicación que puede fallar igual que el mismo nodo. En esta tesis doctoral se han estudiado varios protocolos de comunicación utilizados en sistemas distribuidos de aplicaciones aeroespaciales y de automoción; así mismo, se han comparado los protocolos CAN y LIN, y se ha propuesto, diseñado y prototipado un sistema distribuido básico, para la comprobación de la metodología de calificación y test on-line y de las técnicas de endurecimiento propuestas.
Como aportación original de esta tesis doctoral, se propone una nueva metodología para comprobar y asegurar el endurecimiento efectivo de un sistema digital distribuido, que incluya cualquier bus de comunicación. Otra contribución original de esta tesis doctoral, consiste en una técnica que permite la comparación de los datos obtenidos tras la irradiación de una forma directa y transparente al sistema de test, basada en un bloque de detección de minoría para rastreo de la propagación de los fallos. Se han analizado distintas tecnologías de FPGA basadas en memoria Flash, con el propósito de una mejor caracterización de los dispositivos a iradiar. Para el sistema final se ha optado por la utilización de una FPGA Igloo de Microsemi®, debido a que es una tecnología más robusta y basada en la tecnología Flash. Se ha desarrollado un software de control para el sistema distribuido, el cual se ejecuta en un microprocesador contenido en uno de los nodos del sistema y el que envía los resultados obtenidos por medio de un bus SPI a un PC. El software automatiza el proceso de recolección de los datos, proporcionando el resultado de cuál de los nodos ha fallado, si se ha recuperado, si ha fallado el sistema de comunicación, etc. Estos resultados permiten validar experimentalmente el método propuesto para los sistemas distribuidos digitales. Además de cumplir con los objetivos del proyecto, se han resuelto dos de los problemas típicos de este tipo de sistemas. Estos corresponden al reset síncrono del sistema y a la monitorización del mantenimiento continuo del tiempo real del sistema. Los distintos bloques propuestos y utilizados en este método integral de calificación de sistemas distribuidos son adecuados para la realización de test on-line de sistemas distribuidos, con el objeto de detectar la presencia de fallos permanentes (debidos al envejecimiento de los dispositivos) y de fallos transitorios (debidos a la radiación ionizante).
Finalmente, los últimos aportes de la investigación de este trabajo de tesis, correspondientes al análisis detallado del material del encapsulado del dispositivo a irradiar, y su comportamiento ante el paso de protones, se han podido obtener gracias a la ayuda del Departamento de Ciencia e Ingeniería de los Materiales e Ingeniería Química de la Universidad Carlos III de Madrid y del Centro Nacional de Aceleradores de la Universidad de Sevilla.
La mayoría de los resultados parciales y globales de este trabajo de tesis doctoral han sido publicados en Conferencias Internacionales (International On-Line Testing Symposium, Latin American Test Workshop, Digital Circuit and Integrated Systems) y en una revista internacional (IEEE Transactions on Nuclear Science).
Nowadays, due to the spread of distributed digital systems in the different technological areas, working in aggressive environments, such as automotive and aerospace applications, hardening techniques against ionizing radiation effects is crucial. This PhD work started with the research Project RENASER+ “TEC2010-22095-C03-03”, supported by the Spanish Ministry of Economy and Competitiveness. The main objective of the sub-project, managed by the Carlos III University of Madrid, was the assessment of the sensitiveness of digital systems against ionizing radiation effects. For this purpose, a number of tasks were planned, aimed to set a protocol and to develop a test system, able to run irradiation campaigns in the accelerators placed in the National Centre of Accelerators, University of Seville (CNA). Furthermore, in order to fulfill the project objectives, the works developed in the PhD were devoted to the study and application of Fault Tolerance Techniques for distributed digital systems, and to the prior analysis of the systems sensitivity with emulation-based fault injection campaigns. This last task used the Autonomous Emulation Tool developed in the DMA research group of Carlos III University of Madrid. Along the development of the work of this PdD, in agreement with the tasks and objectives of RENASER+ Project, new improvements have been proposed, developed and included in the emulation-based transient fault injection tool, in order to enable the injection of multiple faults in memory elements, to enable the detailed tracking of accumulated effect of faults injected and detected by mitigation techniques and to scan the fault propagation within the distributed system. With the main purpose of generating a comprehensive method for the analysis, assessment and qualification of robust digital distributed systems, apart from injecting faults via emulation, real faults have been injected thanks to the use of a particle accelerators, CNA-US cyclotron (proton beam). The assessment of the prototyped distributed digital systems, through proton beam irradiation campaigns, required three irradiation campaigns; in the PhD a planning method was proposed for this type of campaigns on distributed systems, which allows the continuous monitoring of fault effects and automatic final report generation. As a more ambitious objective, in the PhD the method was extended to an on-line test, for the continuous checking during normal operation of distributed digital systems, to detect faults which can cause a wrong behavior or, even, a serious failure. Therefore, the qualification method for the assessment of the robustness of distributed digital systems was extended to an on-line method for these systems. During the robustness study of distributed digital systems another problem was identified and analyzed: the permanent faults appearing in this type of systems due to digital devices aging. Indeed, the proposed method is suitable for on-line testing of the normal operation of the distributed digital systems, for the detection of faults due to ionizing radiation (transient faults) and the detection of faults due to the device aging (permanent faults). Distributed digital systems, apart the nodes composing them, use to include a communication protocol that can suffer also from transient or permanent faults. In this PhD different communication protocols, used in distributed digital systems for aerospace and automotive applications; also, CAN and LIN protocols have been compared and a basic distributed system has been proposed, designed and prototyped, for validating the qualification methodology and the test on-line method, as well as proposed hardening techniques. As original contribution of this PhD, a comprehensive methodology for assessing and assuring the effective hardening of distributed digital systems, including a communication protocol, is proposed. Another original contribution of this PhD is the technique that allows the analysis of the data obtained (transparently and directly whatever the test system used) from any irradiation campaign, based on a minority checker block for the fault tracking. Furthermore, in this PhD work, different Flash-based FPGA technologies have been analyzed, with the purpose of better characterizing the analyzed devices. For the final system implemented, FPGAs Igloo from Microsemi® have been selected, because is a more robust technology. Specific control software has been developed for the distributed system, which was run on a microprocessor included in one node of the system and that sends obtained results through a PCI bus to a Personal Computer. This software automates the data collecting process, telling which node is failing, its recovery state, as well as any communication fault occurring. These tools allow the experimental validation of proposed method for the distributed digital systems. Apart from fulfilling the Project objectives, two typical problems of this type of systems have been solved. They correspond to synchronous initialization of the system and continuous real-time maintenance. Different blocks proposed and used in this comprehensive method for the assessment of the sensitivity of distributed digital systems are also adequate for on-line testing of these systems, with the prupose of detecting permanent faults (due to device aging) and transient faults (due to ionizing radiation). Finally, the latest research works of this PhD work correspond to the detailed analysis of the packaging of the devices to be irradiated, as well as its behavior when proton particles are going through it. These works have been done thanks to the help of Science and Material Engineering and Chemistry Engineering Department of Carlos III University of Madrid, and the National Centre of Accelerators of University of Seville. The majority of partial and global results of this PhD have been published in International Conferences (International On-Line Testing Symposium, Latin American Test Workshop, Digital Circuit and Integrated Systems) and in the IEEE Transactions on Nuclear Science.
Nowadays, due to the spread of distributed digital systems in the different technological areas, working in aggressive environments, such as automotive and aerospace applications, hardening techniques against ionizing radiation effects is crucial. This PhD work started with the research Project RENASER+ “TEC2010-22095-C03-03”, supported by the Spanish Ministry of Economy and Competitiveness. The main objective of the sub-project, managed by the Carlos III University of Madrid, was the assessment of the sensitiveness of digital systems against ionizing radiation effects. For this purpose, a number of tasks were planned, aimed to set a protocol and to develop a test system, able to run irradiation campaigns in the accelerators placed in the National Centre of Accelerators, University of Seville (CNA). Furthermore, in order to fulfill the project objectives, the works developed in the PhD were devoted to the study and application of Fault Tolerance Techniques for distributed digital systems, and to the prior analysis of the systems sensitivity with emulation-based fault injection campaigns. This last task used the Autonomous Emulation Tool developed in the DMA research group of Carlos III University of Madrid. Along the development of the work of this PdD, in agreement with the tasks and objectives of RENASER+ Project, new improvements have been proposed, developed and included in the emulation-based transient fault injection tool, in order to enable the injection of multiple faults in memory elements, to enable the detailed tracking of accumulated effect of faults injected and detected by mitigation techniques and to scan the fault propagation within the distributed system. With the main purpose of generating a comprehensive method for the analysis, assessment and qualification of robust digital distributed systems, apart from injecting faults via emulation, real faults have been injected thanks to the use of a particle accelerators, CNA-US cyclotron (proton beam). The assessment of the prototyped distributed digital systems, through proton beam irradiation campaigns, required three irradiation campaigns; in the PhD a planning method was proposed for this type of campaigns on distributed systems, which allows the continuous monitoring of fault effects and automatic final report generation. As a more ambitious objective, in the PhD the method was extended to an on-line test, for the continuous checking during normal operation of distributed digital systems, to detect faults which can cause a wrong behavior or, even, a serious failure. Therefore, the qualification method for the assessment of the robustness of distributed digital systems was extended to an on-line method for these systems. During the robustness study of distributed digital systems another problem was identified and analyzed: the permanent faults appearing in this type of systems due to digital devices aging. Indeed, the proposed method is suitable for on-line testing of the normal operation of the distributed digital systems, for the detection of faults due to ionizing radiation (transient faults) and the detection of faults due to the device aging (permanent faults). Distributed digital systems, apart the nodes composing them, use to include a communication protocol that can suffer also from transient or permanent faults. In this PhD different communication protocols, used in distributed digital systems for aerospace and automotive applications; also, CAN and LIN protocols have been compared and a basic distributed system has been proposed, designed and prototyped, for validating the qualification methodology and the test on-line method, as well as proposed hardening techniques. As original contribution of this PhD, a comprehensive methodology for assessing and assuring the effective hardening of distributed digital systems, including a communication protocol, is proposed. Another original contribution of this PhD is the technique that allows the analysis of the data obtained (transparently and directly whatever the test system used) from any irradiation campaign, based on a minority checker block for the fault tracking. Furthermore, in this PhD work, different Flash-based FPGA technologies have been analyzed, with the purpose of better characterizing the analyzed devices. For the final system implemented, FPGAs Igloo from Microsemi® have been selected, because is a more robust technology. Specific control software has been developed for the distributed system, which was run on a microprocessor included in one node of the system and that sends obtained results through a PCI bus to a Personal Computer. This software automates the data collecting process, telling which node is failing, its recovery state, as well as any communication fault occurring. These tools allow the experimental validation of proposed method for the distributed digital systems. Apart from fulfilling the Project objectives, two typical problems of this type of systems have been solved. They correspond to synchronous initialization of the system and continuous real-time maintenance. Different blocks proposed and used in this comprehensive method for the assessment of the sensitivity of distributed digital systems are also adequate for on-line testing of these systems, with the prupose of detecting permanent faults (due to device aging) and transient faults (due to ionizing radiation). Finally, the latest research works of this PhD work correspond to the detailed analysis of the packaging of the devices to be irradiated, as well as its behavior when proton particles are going through it. These works have been done thanks to the help of Science and Material Engineering and Chemistry Engineering Department of Carlos III University of Madrid, and the National Centre of Accelerators of University of Seville. The majority of partial and global results of this PhD have been published in International Conferences (International On-Line Testing Symposium, Latin American Test Workshop, Digital Circuit and Integrated Systems) and in the IEEE Transactions on Nuclear Science.
Description
Keywords
Circuitos digitales, Radiación ionizante, Sistemas dinámicos, Emulación autónoma, Sistema de inyección de fallos