Side-Channel Attack Protection Techniques in FPGA Systems using Enhanced Dual-Rail Solutions

Esta tesis doctoral se centra principalmente en técnicas de ataque y contramedidas relacionadas con ataques de canal lateral (SCA por sus siglas en inglés), que han sido propuestas dentro del campo de investigación académica desde hace 17 años. Las investigaciones relacionadas han experimentado un notable crecimiento en las últimas décadas, mientras que los diseños enfocados en la protección sólida y eficaz contra dichos ataques aún se mantienen como un tema de investigación abierto, en el que se necesitan iniciativas más confiables para la protección de la información persona de empresa y de datos nacionales. El primer uso documentado de codificación secreta se remonta a alrededor de 1700 B.C., cuando los jeroglíficos del antiguo Egipto eran descritos en las inscripciones. La seguridad de la información siempre ha supuesto un factor clave en la transmisión de datos relacionados con inteligencia diplomática o militar. Debido a la evolución rápida de las técnicas modernas de comunicación, soluciones de cifrado se incorporaron por primera vez para garantizar la seguridad, integridad y confidencialidad de los contextos de transmisión a través de cables sin seguridad o medios inalámbricos. Debido a las restricciones de potencia de cálculo antes de la era del ordenador, la técnica de cifrado simple era un método más que suficiente para ocultar la información. Sin embargo, algunas vulnerabilidades algorítmicas pueden ser explotadas para restaurar la regla de codificación sin mucho esfuerzo. Esto ha motivado nuevas investigaciones en el área de la criptografía, con el fin de proteger el sistema de información ante sofisticados algoritmos. Con la invención de los ordenadores se ha acelerado en gran medida la implementación de criptografía segura, que ofrece resistencia eficiente encaminada a obtener mayores capacidades de computación altamente reforzadas. Igualmente, sofisticados cripto-análisis han impulsado las tecnologías de computación. Hoy en día, el mundo de la información ha estado involucrado con el campo de la criptografía, enfocada a proteger cualquier campo a través de diversas soluciones de cifrado. Estos enfoques se han fortalecido debido a la unificación optimizada de teorías matemáticas modernas y prácticas eficaces de hardware, siendo posible su implementación en varias plataformas (microprocesador, ASIC, FPGA, etc.). Las necesidades y requisitos de seguridad en la industria son las principales métricas de conducción en el diseño electrónico, con el objetivo de promover la fabricación de productos de gran alcance sin sacrificar la seguridad de los clientes. Sin embargo, una vulnerabilidad en la implementación práctica encontrada por el Prof. Paul Kocher, et al en 1996 implica que un circuito digital es inherentemente vulnerable a un ataque no convencional, lo cual fue nombrado posteriormente como ataque de canal lateral, debido a su fuente de análisis. Sin embargo, algunas críticas sobre los algoritmos criptográficos teóricamente seguros surgieron casi inmediatamente después de este descubrimiento. En este sentido, los circuitos digitales consisten típicamente en un gran número de celdas lógicas fundamentales (como MOS - Metal Oxide Semiconductor), construido sobre un sustrato de silicio durante la fabricación. La lógica de los circuitos se realiza en función de las innumerables conmutaciones de estas células. Este mecanismo provoca inevitablemente cierta emanación física especial que puede ser medida y correlacionada con el comportamiento interno del circuito. SCA se puede utilizar para revelar datos confidenciales (por ejemplo, la criptografía de claves), analizar la arquitectura lógica, el tiempo e incluso inyectar fallos malintencionados a los circuitos que se implementan en sistemas embebidos, como FPGAs, ASICs, o tarjetas inteligentes. Mediante el uso de la comparación de correlación entre la cantidad de fuga estimada y las fugas medidas de forma real, información confidencial puede ser reconstruida en mucho menos tiempo y computación. Para ser precisos, SCA básicamente cubre una amplia gama de tipos de ataques, como los análisis de consumo de energía y radiación ElectroMagnética (EM). Ambos se basan en análisis estadístico y, por lo tanto, requieren numerosas muestras. Los algoritmos de cifrado no están intrínsecamente preparados para ser resistentes ante SCA. Es por ello que se hace necesario durante la implementación de circuitos integrar medidas que permitan camuflar las fugas a través de "canales laterales". Las medidas contra SCA están evolucionando junto con el desarrollo de nuevas técnicas de ataque, así como la continua mejora de los dispositivos electrónicos. Las características físicas requieren contramedidas sobre la capa física, que generalmente se pueden clasificar en soluciones intrínsecas y extrínsecas. Contramedidas extrínsecas se ejecutan para confundir la fuente de ataque mediante la integración de ruido o mala alineación de la actividad interna. Comparativamente, las contramedidas intrínsecas están integradas en el propio algoritmo, para modificar la aplicación con el fin de minimizar las fugas medibles, o incluso hacer que dichas fugas no puedan ser medibles. Ocultación y Enmascaramiento son dos técnicas típicas incluidas en esta categoría. Concretamente, el enmascaramiento se aplica a nivel algorítmico, para alterar los datos intermedios sensibles con una máscara de manera reversible. A diferencia del enmascaramiento lineal, las operaciones no lineales que ampliamente existen en criptografías modernas son difíciles de enmascarar. Dicho método de ocultación, que ha sido verificado como una solución efectiva, comprende principalmente la codificación en doble carril, que está ideado especialmente para aplanar o eliminar la fuga dependiente de dato en potencia o en EM. En esta tesis doctoral, además de la descripción de las metodologías de ataque, se han dedicado grandes esfuerzos sobre la estructura del prototipo de la lógica propuesta, con el fin de realizar investigaciones enfocadas a la seguridad sobre contramedidas de arquitectura a nivel lógico. Una característica de SCA reside en el formato de las fuentes de fugas. Un típico ataque de canal lateral se refiere al análisis basado en la potencia, donde la capacidad fundamental del transistor MOS y otras capacidades parásitas son las fuentes esenciales de fugas. Por lo tanto, una lógica robusta resistente a SCA debe eliminar o mitigar las fugas de estas micro-unidades, como las puertas lógicas básicas, los puertos I/O y las rutas. Las herramientas EDA proporcionadas por los vendedores manipulan la lógica desde un nivel más alto, en lugar de realizarlo desde el nivel de puerta, donde las fugas de canal lateral se manifiestan. Por lo tanto, las implementaciones clásicas apenas satisfacen estas necesidades e inevitablemente atrofian el prototipo. Por todo ello, la implementación de un esquema de diseño personalizado y flexible ha de ser tomado en cuenta. En esta tesis se presenta el diseño y la implementación de una lógica innovadora para contrarrestar SCA, en la que se abordan 3 aspectos fundamentales: I. Se basa en ocultar la estrategia sobre el circuito en doble carril a nivel de puerta para obtener dinámicamente el equilibrio de las fugas en las capas inferiores; II. Esta lógica explota las características de la arquitectura de las FPGAs, para reducir al mínimo el gasto de recursos en la implementación; III. Se apoya en un conjunto de herramientas asistentes personalizadas, incorporadas al flujo genérico de diseño sobre FPGAs, con el fin de manipular los circuitos de forma automática. El kit de herramientas de diseño automático es compatible con la lógica de doble carril propuesta, para facilitar la aplicación práctica sobre la familia de FPGA del fabricante Xilinx. En este sentido, la metodología y las herramientas son flexibles para ser extendido a una amplia gama de aplicaciones en las que se desean obtener restricciones mucho más rígidas y sofisticadas a nivel de puerta o rutado. En esta tesis se realiza un gran esfuerzo para facilitar el proceso de implementación y reparación de lógica de doble carril genérica. La viabilidad de las soluciones propuestas es validada mediante la selección de algoritmos criptográficos ampliamente utilizados, y su evaluación exhaustiva en comparación con soluciones anteriores. Todas las propuestas están respaldadas eficazmente a través de ataques experimentales con el fin de validar las ventajas de seguridad del sistema. El presente trabajo de investigación tiene la intención de cerrar la brecha entre las barreras de implementación y la aplicación efectiva de lógica de doble carril. En esencia, a lo largo de esta tesis se describirá un conjunto de herramientas de implementación para FPGAs que se han desarrollado para trabajar junto con el flujo de diseño genérico de las mismas, con el fin de lograr crear de forma innovadora la lógica de doble carril. Un nuevo enfoque en el ámbito de la seguridad en el cifrado se propone para obtener personalización, automatización y flexibilidad en el prototipo de circuito de bajo nivel con granularidad fina. Las principales contribuciones del presente trabajo de investigación se resumen brevemente a continuación: Lógica de Precharge Absorbed-DPL logic: El uso de la conversión de netlist para reservar LUTs libres para ejecutar la señal de precharge y Ex en una lógica DPL. Posicionamiento entrelazado Row-crossed con pares idénticos de rutado en redes de doble carril, lo que ayuda a aumentar la resistencia frente a la medición EM selectiva y mitigar los impactos de las variaciones de proceso. Ejecución personalizada y herramientas de conversión automática para la generación de redes idénticas para la lógica de doble carril propuesta. (a) Para detectar y reparar conflictos en las conexiones; (b) Detectar y reparar las rutas asimétricas. (c) Para ser utilizado en otras lógicas donde se requiere un control estricto de las interconexiones en aplicaciones basadas en Xilinx. Plataforma CPA de pruebas personalizadas para el análisis de EM y potencia, incluyendo la construcción de dicha plataforma, el método de medición y análisis de los ataques. Análisis de tiempos para cuantificar los niveles de seguridad. División de Seguridad en la conversión parcial de un sistema de cifrado complejo para reducir los costes de la protección. Prueba de concepto de un sistema de calefacción auto-adaptativo para mitigar los impactos eléctricos debido a la variación del proceso de silicio de manera dinámica. La presente tesis doctoral se encuentra organizada tal y como se detalla a continuación: En el capítulo 1 se abordan los fundamentos de los ataques de canal lateral, que abarca desde conceptos básicos de teoría de modelos de análisis, además de la implementación de la plataforma y la ejecución de los ataques. En el capítulo 2 se incluyen las estrategias de resistencia SCA contra los ataques de potencia diferencial y de EM. Además de ello, en este capítulo se propone una lógica en doble carril compacta y segura como contribución de gran relevancia, así como también se presentará la transformación lógica basada en un diseño a nivel de puerta. Por otra parte, en el Capítulo 3 se abordan los desafíos relacionados con la implementación de lógica en doble carril genérica. Así mismo, se describirá un flujo de diseño personalizado para resolver los problemas de aplicación junto con una herramienta de desarrollo automático de aplicaciones propuesta, para mitigar las barreras de diseño y facilitar los procesos. En el capítulo 4 se describe de forma detallada la elaboración e implementación de las herramientas propuestas. Por otra parte, la verificación y validaciones de seguridad de la lógica propuesta, así como un sofisticado experimento de verificación de la seguridad del rutado, se describen en el capítulo 5. Por último, un resumen de las conclusiones de la tesis y las perspectivas como líneas futuras se incluyen en el capítulo 6. Con el fin de profundizar en el contenido de la tesis doctoral, cada capítulo se describe de forma más detallada a continuación: En el capítulo 1 se introduce plataforma de implementación hardware además las teorías básicas de ataque de canal lateral, y contiene principalmente: (a) La arquitectura genérica y las características de la FPGA a utilizar, en particular la Xilinx Virtex-5; (b) El algoritmo de cifrado seleccionado (un módulo comercial Advanced Encryption Standard (AES)); (c) Los elementos esenciales de los métodos de canal lateral, que permiten revelar las fugas de disipación correlacionadas con los comportamientos internos; y el método para recuperar esta relación entre las fluctuaciones físicas en los rastros de canal lateral y los datos internos procesados; (d) Las configuraciones de las plataformas de pruebas de potencia / EM abarcadas dentro de la presente tesis. El contenido de esta tesis se amplia y profundiza a partir del capítulo 2, en el cual se abordan varios aspectos claves. En primer lugar, el principio de protección de la compensación dinámica de la lógica genérica de precarga de doble carril (Dual-rail Precharge Logic-DPL) se explica mediante la descripción de los elementos compensados a nivel de puerta. En segundo lugar, la lógica PA-DPL es propuesta como aportación original, detallando el protocolo de la lógica y un caso de aplicación. En tercer lugar, dos flujos de diseño personalizados se muestran para realizar la conversión de doble carril. Junto con ello, se aclaran las definiciones técnicas relacionadas con la manipulación por encima de la netlist a nivel de LUT. Finalmente, una breve discusión sobre el proceso global se aborda en la parte final del capítulo. El Capítulo 3 estudia los principales retos durante la implementación de DPLs en FPGAs. El nivel de seguridad de las soluciones de resistencia a SCA encontradas en el estado del arte se ha degenerado debido a las barreras de implantación a través de herramientas EDA convencionales. En el escenario de la arquitectura FPGA estudiada, se discuten los problemas de los formatos de doble carril, impactos parásitos, sesgo tecnológico y la viabilidad de implementación. De acuerdo con estas elaboraciones, se plantean dos problemas: Cómo implementar la lógica propuesta sin penalizar los niveles de seguridad, y cómo manipular un gran número de celdas y automatizar el proceso. El PA-DPL propuesto en el capítulo 2 se valida con una serie de iniciativas, desde características estructurales como doble carril entrelazado o redes de rutado clonadas, hasta los métodos de aplicación tales como las herramientas de personalización y automatización de EDA. Por otra parte, un sistema de calefacción auto-adaptativo es representado y aplicado a una lógica de doble núcleo, con el fin de ajustar alternativamente la temperatura local para equilibrar los impactos negativos de la variación del proceso durante la operación en tiempo real. El capítulo 4 se centra en los detalles de la implementación del kit de herramientas. Desarrollado sobre una API third-party, el kit de herramientas personalizado es capaz de manipular los elementos de la lógica de circuito post P&R ncd (una versión binaria ilegible del xdl) convertido al formato XDL Xilinx. El mecanismo y razón de ser del conjunto de instrumentos propuestos son cuidadosamente descritos, que cubre la detección de enrutamiento y los enfoques para la reparación. El conjunto de herramientas desarrollado tiene como objetivo lograr redes de enrutamiento estrictamente idénticos para la lógica de doble carril, tanto para posicionamiento separado como para el entrelazado. Este capítulo particularmente especifica las bases técnicas para apoyar las implementaciones en los dispositivos de Xilinx y su flexibilidad para ser utilizado sobre otras aplicaciones. El capítulo 5 se enfoca en la aplicación de los casos de estudio para la validación de los grados de seguridad de la lógica propuesta. Se discuten los problemas técnicos detallados durante la ejecución y algunas nuevas técnicas de implementación. (a) Se discute el impacto en el proceso de posicionamiento de la lógica utilizando el kit de herramientas propuesto. Diferentes esquemas de implementación, tomando en cuenta la optimización global en seguridad y coste, se verifican con los experimentos con el fin de encontrar los planes de posicionamiento y reparación optimizados; (b) las validaciones de seguridad se realizan con los métodos de correlación y análisis de tiempo; (c) Una táctica asintótica se aplica a un núcleo AES sobre BCDL estructurado para validar de forma sofisticada el impacto de enrutamiento sobre métricas de seguridad; (d) Los resultados preliminares utilizando el sistema de calefacción auto-adaptativa sobre la variación del proceso son mostrados; (e) Se introduce una aplicación práctica de las herramientas para un diseño de cifrado completa. Capítulo 6 incluye el resumen general del trabajo presentado dentro de esta tesis doctoral. Por último, una breve perspectiva del trabajo futuro se expone, lo que puede ampliar el potencial de utilización de las contribuciones de esta tesis a un alcance más allá de los dominios de la criptografía en FPGAs. ABSTRACT This PhD thesis mainly concentrates on countermeasure techniques related to the Side Channel Attack (SCA), which has been put forward to academic exploitations since 17 years ago. The related research has seen a remarkable growth in the past decades, while the design of solid and efficient protection still curiously remain as an open research topic where more reliable initiatives are required for personal information privacy, enterprise and national data protections. The earliest documented usage of secret code can be traced back to around 1700 B.C., when the hieroglyphs in ancient Egypt are scribed in inscriptions. Information security always gained serious attention from diplomatic or military intelligence transmission. Due to the rapid evolvement of modern communication technique, crypto solution was first incorporated by electronic signal to ensure the confidentiality, integrity, availability, authenticity and non-repudiation of the transmitted contexts over unsecure cable or wireless channels. Restricted to the computation power before computer era, simple encryption tricks were practically sufficient to conceal information. However, algorithmic vulnerabilities can be excavated to restore the encoding rules with affordable efforts. This fact motivated the development of modern cryptography, aiming at guarding information system by complex and advanced algorithms. The appearance of computers has greatly pushed forward the invention of robust cryptographies, which efficiently offers resistance relying on highly strengthened computing capabilities. Likewise, advanced cryptanalysis has greatly driven the computing technologies in turn. Nowadays, the information world has been involved into a crypto world, protecting any fields by pervasive crypto solutions. These approaches are strong because of the optimized mergence between modern mathematical theories and effective hardware practices, being capable of implement crypto theories into various platforms (microprocessor, ASIC, FPGA, etc). Security needs from industries are actually the major driving metrics in electronic design, aiming at promoting the construction of systems with high performance without sacrificing security. Yet a vulnerability in practical implementation found by Prof. Paul Kocher, et al in 1996 implies that modern digital circuits are inherently vulnerable to an unconventional attack approach, which was named as side-channel attack since then from its analysis source. Critical suspicions to theoretically sound modern crypto algorithms surfaced almost immediately after this discovery. To be specifically, digital circuits typically consist of a great number of essential logic elements (as MOS - Metal Oxide Semiconductor), built upon a silicon substrate during the fabrication. Circuit logic is realized relying on the countless switch actions of these cells. This mechanism inevitably results in featured physical emanation that can be properly measured and correlated with internal circuit behaviors. SCAs can be used to reveal the confidential data (e.g. crypto-key), analyze the logic architecture, timing and even inject malicious faults to the circuits that are implemented in hardware system, like FPGA, ASIC, smart Card. Using various comparison solutions between the predicted leakage quantity and the measured leakage, secrets can be reconstructed at much less expense of time and computation. To be precisely, SCA basically encloses a wide range of attack types, typically as the analyses of power consumption or electromagnetic (EM) radiation. Both of them rely on statistical analyses, and hence require a number of samples. The crypto algorithms are not intrinsically fortified with SCA-resistance. Because of the severity, much attention has to be taken into the implementation so as to assemble countermeasures to camouflage the leakages via "side channels". Countermeasures against SCA are evolving along with the development of attack techniques. The physical characteristics requires countermeasures over physical layer, which can be generally classified into intrinsic and extrinsic vectors. Extrinsic countermeasures are executed to confuse the attacker by integrating noise, misalignment to the intra activities. Comparatively, intrinsic countermeasures are built into the algorithm itself, to modify the implementation for minimizing the measurable leakage, or making them not sensitive any more. Hiding and Masking are two typical techniques in this category. Concretely, masking applies to the algorithmic level, to alter the sensitive intermediate values with a mask in reversible ways. Unlike the linear masking, non-linear operations that widely exist in modern cryptographies are difficult to be masked. Approved to be an effective counter solution, hiding method mainly mentions dual-rail logic, which is specially devised for flattening or removing the data-dependent leakage in power or EM signatures. In this thesis, apart from the context describing the attack methodologies, efforts have also been dedicated to logic prototype, to mount extensive security investigations to countermeasures on logic-level. A characteristic of SCA resides on the format of leak sources. Typical side-channel attack concerns the power based analysis, where the fundamental capacitance from MOS transistors and other parasitic capacitances are the essential leak sources. Hence, a robust SCA-resistant logic must eliminate or mitigate the leakages from these micro units, such as basic logic gates, I/O ports and routings. The vendor provided EDA tools manipulate the logic from a higher behavioral-level, rather than the lower gate-level where side-channel leakage is generated. So, the classical implementations barely satisfy these needs and inevitably stunt the prototype. In this case, a customized and flexible design scheme is appealing to be devised. This thesis profiles an innovative logic style to counter SCA, which mainly addresses three major aspects: I. The proposed logic is based on the hiding strategy over gate-level dual-rail style to dynamically overbalance side-channel leakage from lower circuit layer; II. This logic exploits architectural features of modern FPGAs, to minimize the implementation expenses; III. It is supported by a set of assistant custom tools, incorporated by the generic FPGA design flow, to have circuit manipulations in an automatic manner. The automatic design toolkit supports the proposed dual-rail logic, facilitating the practical implementation on Xilinx FPGA families. While the methodologies and the tools are flexible to be expanded to a wide range of applications where rigid and sophisticated gate- or routing- constraints are desired. In this thesis a great effort is done to streamline the implementation workflow of generic dual-rail logic. The feasibility of the proposed solutions is validated by selected and widely used crypto algorithm, for thorough and fair evaluation w.r.t. prior solutions. All the proposals are effectively verified by security experiments. The presented research work attempts to solve the implementation troubles. The essence that will be formalized along this thesis is that a customized execution toolkit for modern FPGA systems is developed to work together with the generic FPGA design flow for creating innovative dual-rail logic. A method in crypto security area is constructed to obtain customization, automation and flexibility in low-level circuit prototype with fine-granularity in intractable routings. Main contributions of the presented work are summarized next: Precharge Absorbed-DPL logic: Using the netlist conversion to reserve free LUT inputs to execute the Precharge and Ex signal in a dual-rail logic style. A row-crossed interleaved placement method with identical routing pairs in dual-rail networks, which helps to increase the resistance against selective EM measurement and mitigate the impacts from process variations. Customized execution and automatic transformation tools for producing identical networks for the proposed dual-rail logic. (a) To detect and repair the conflict nets; (b) To detect and repair the asymmetric nets. (c) To be used in other logics where strict network control is required in Xilinx scenario. Customized correlation analysis testbed for EM and power attacks, including the platform construction, measurement method and attack analysis. A timing analysis based method for quantifying the security grades. A methodology of security partitions of complex crypto systems for reducing the protection cost. A proof-of-concept self-adaptive heating system to mitigate electrical impacts over process variations in dynamic dual-rail compensation manner. The thesis chapters are organized as follows: Chapter 1 discusses the side-channel attack fundamentals, which covers from theoretic basics to analysis models, and further to platform setup and attack execution. Chapter 2 centers to SCA-resistant strategies against generic power and EM attacks. In this chapter, a major contribution, a compact and secure dual-rail logic style, will be originally proposed. The logic transformation based on bottom-layer design will be presented. Chapter 3 is scheduled to elaborate the implementation challenges of generic dual-rail styles. A customized design flow to solve the implementation problems will be described along with a self-developed automatic implementation toolkit, for mitigating the design barriers and facilitating the processes. Chapter 4 will originally elaborate the tool specifics and construction details. The implementation case studies and security validations for the proposed logic style, as well as a sophisticated routing verification experiment, will be described in Chapter 5. Finally, a summary of thesis conclusions and perspectives for future work are included in Chapter 5. To better exhibit the thesis contents, each chapter is further described next: Chapter 1 provides the introduction of hardware implementation testbed and side-channel attack fundamentals, and mainly contains: (a) The FPGA generic architecture and device features, particularly of Virtex-5 FPGA; (b) The selected crypto algorithm - a commercially and extensively used Advanced Encryption Standard (AES) module - is detailed; (c) The essentials of Side-Channel methods are profiled. It reveals the correlated dissipation leakage to the internal behaviors, and the method to recover this relationship between the physical fluctuations in side-channel traces and the intra processed data; (d) The setups of the power/EM testing platforms enclosed inside the thesis work are given. The content of this thesis is expanded and deepened from chapter 2, which is divided into several aspects. First, the protection principle of dynamic compensation of the generic dual-rail precharge logic is explained by describing the compensated gate-level elements. Second, the novel DPL is originally proposed by detailing the logic protocol and an implementation case study. Third, a couple of custom workflows are shown next for realizing the rail conversion. Meanwhile, the technical definitions that are about to be manipulated above LUT-level netlist are clarified. A brief discussion about the batched process is given in the final part. Chapter 3 studies the implementation challenges of DPLs in FPGAs. The security level of state-of-the-art SCA-resistant solutions are decreased due to the implementation barriers using conventional EDA tools. In the studied FPGA scenario, problems are discussed from dual-rail format, parasitic impact, technological bias and implementation feasibility. According to these elaborations, two problems arise: How to implement the proposed logic without crippling the security level; and How to manipulate a large number of cells and automate the transformation. The proposed PA-DPL in chapter 2 is legalized with a series of initiatives, from structures to implementation methods. Furthermore, a self-adaptive heating system is depicted and implemented to a dual-core logic, assumed to alternatively adjust local temperature for balancing the negative impacts from silicon technological biases on real-time. Chapter 4 centers to the toolkit system. Built upon a third-party Application Program Interface (API) library, the customized toolkit is able to manipulate the logic elements from post P&R circuit (an unreadable binary version of the xdl one) converted to Xilinx xdl format. The mechanism and rationale of the proposed toolkit are carefully convoyed, covering the routing detection and repairing approaches. The developed toolkit aims to achieve very strictly identical routing networks for dual-rail logic both for separate and interleaved placement. This chapter particularly specifies the technical essentials to support the implementations in Xilinx devices and the flexibility to be expanded to other applications. Chapter 5 focuses on the implementation of the case studies for validating the security grades of the proposed logic style from the proposed toolkit. Comprehensive implementation techniques are discussed. (a) The placement impacts using the proposed toolkit are discussed. Different execution schemes, considering the global optimization in security and cost, are verified with experiments so as to find the optimized placement and repair schemes; (b) Security validations are realized with correlation, timing methods; (c) A systematic method is applied to a BCDL structured module to validate the routing impact over security metric; (d) The preliminary results using the self-adaptive heating system over process variation is given; (e) A practical implementation of the proposed toolkit to a large design is introduced. Chapter 6 includes the general summary of the complete work presented inside this thesis. Finally, a brief perspective for the future work is drawn which might expand the potential utilization of the thesis contributions to a wider range of implementation domains beyond cryptography on FPGAs.

Thermal, stress, and traps effects in AlGaN/GaN HEMTs

GaN y AlN son materiales semiconductores piezoeléctricos del grupo III-V. La heterounión AlGaN/GaN presenta una elevada carga de polarización tanto piezoeléctrica como espontánea en la intercara, lo que genera en su cercanía un 2DEG de grandes concentración y movilidad. Este 2DEG produce una muy alta potencia de salida, que a su vez genera una elevada temperatura de red. Las tensiones de puerta y drenador provocan un stress piezoeléctrico inverso, que puede afectar a la carga de polarización piezoeléctrica y así influir la densidad 2DEG y las características de salida. Por tanto, la física del dispositivo es relevante para todos sus aspectos eléctricos, térmicos y mecánicos. En esta tesis se utiliza el software comercial COMSOL, basado en el método de elementos finitos (FEM), para simular el comportamiento integral electro-térmico, electro-mecánico y electro-térmico-mecánico de los HEMTs de GaN. Las partes de acoplamiento incluyen el modelo de deriva y difusión para el transporte electrónico, la conducción térmica y el efecto piezoeléctrico. Mediante simulaciones y algunas caracterizaciones experimentales de los dispositivos, hemos analizado los efectos térmicos, de deformación y de trampas. Se ha estudiado el impacto de la geometría del dispositivo en su auto-calentamiento mediante simulaciones electro-térmicas y algunas caracterizaciones eléctricas. Entre los resultados más sobresalientes, encontramos que para la misma potencia de salida la distancia entre los contactos de puerta y drenador influye en generación de calor en el canal, y así en su temperatura. El diamante posee une elevada conductividad térmica. Integrando el diamante en el dispositivo se puede dispersar el calor producido y así reducir el auto-calentamiento, al respecto de lo cual se han realizado diversas simulaciones electro-térmicas. Si la integración del diamante es en la parte superior del transistor, los factores determinantes para la capacidad disipadora son el espesor de la capa de diamante, su conductividad térmica y su distancia a la fuente de calor. Este procedimiento de disipación superior también puede reducir el impacto de la barrera térmica de intercara entre la capa adaptadora (buffer) y el substrato. La muy reducida conductividad eléctrica del diamante permite que pueda contactar directamente el metal de puerta (muy cercano a la fuente de calor), lo que resulta muy conveniente para reducir el auto-calentamiento del dispositivo con polarización pulsada. Por otra parte se simuló el dispositivo con diamante depositado en surcos atacados sobre el sustrato como caminos de disipación de calor (disipador posterior). Aquí aparece una competencia de factores que influyen en la capacidad de disipación, a saber, el surco atacado contribuye a aumentar la temperatura del dispositivo debido al pequeño tamaño del disipador, mientras que el diamante disminuiría esa temperatura gracias a su elevada conductividad térmica. Por tanto, se precisan capas de diamante relativamente gruesas para reducer ele efecto de auto-calentamiento. Se comparó la simulación de la deformación local en el borde de la puerta del lado cercano al drenador con estructuras de puerta estándar y con field plate, que podrían ser muy relevantes respecto a fallos mecánicos del dispositivo. Otras simulaciones se enfocaron al efecto de la deformación intrínseca de la capa de diamante en el comportamiento eléctrico del dispositivo. Se han comparado los resultados de las simulaciones de la deformación y las características eléctricas de salida con datos experimentales obtenidos por espectroscopía micro-Raman y medidas eléctricas, respectivamente. Los resultados muestran el stress intrínseco en la capa producido por la distribución no uniforme del 2DEG en el canal y la región de acceso. Además de aumentar la potencia de salida del dispositivo, la deformación intrínseca en la capa de diamante podría mejorar la fiabilidad del dispositivo modulando la deformación local en el borde de la puerta del lado del drenador. Finalmente, también se han simulado en este trabajo los efectos de trampas localizados en la superficie, el buffer y la barrera. Las medidas pulsadas muestran que tanto las puertas largas como las grandes separaciones entre los contactos de puerta y drenador aumentan el cociente entre la corriente pulsada frente a la corriente continua (lag ratio), es decir, disminuir el colapse de corriente (current collapse). Este efecto ha sido explicado mediante las simulaciones de los efectos de trampa de superficie. Por su parte, las referidas a trampas en el buffer se enfocaron en los efectos de atrapamiento dinámico, y su impacto en el auto-calentamiento del dispositivo. Se presenta también un modelo que describe el atrapamiento y liberación de trampas en la barrera: mientras que el atrapamiento se debe a un túnel directo del electrón desde el metal de puerta, el desatrapamiento consiste en la emisión del electrón en la banda de conducción mediante túnel asistido por fonones. El modelo también simula la corriente de puerta, debida a la emisión electrónica dependiente de la temperatura y el campo eléctrico. Además, también se ilustra la corriente de drenador dependiente de la temperatura y el campo eléctrico. ABSTRACT GaN and AlN are group III-V piezoelectric semiconductor materials. The AlGaN/GaN heterojunction presents large piezoelectric and spontaneous polarization charge at the interface, leading to high 2DEG density close to the interface. A high power output would be obtained due to the high 2DEG density and mobility, which leads to elevated lattice temperature. The gate and drain biases induce converse piezoelectric stress that can influence the piezoelectric polarization charge and further influence the 2DEG density and output characteristics. Therefore, the device physics is relevant to all the electrical, thermal, and mechanical aspects. In this dissertation, by using the commercial finite-element-method (FEM) software COMSOL, we achieved the GaN HEMTs simulation with electro-thermal, electro-mechanical, and electro-thermo-mechanical full coupling. The coupling parts include the drift-diffusion model for the electron transport, the thermal conduction, and the piezoelectric effect. By simulations and some experimental characterizations, we have studied the device thermal, stress, and traps effects described in the following. The device geometry impact on the self-heating was studied by electro-thermal simulations and electrical characterizations. Among the obtained interesting results, we found that, for same power output, the distance between the gate and drain contact can influence distribution of the heat generation in the channel and thus influence the channel temperature. Diamond possesses high thermal conductivity. Integrated diamond with the device can spread the generated heat and thus potentially reduce the device self-heating effect. Electro-thermal simulations on this topic were performed. For the diamond integration on top of the device (top-side heat spreading), the determinant factors for the heat spreading ability are the diamond thickness, its thermal conductivity, and its distance to the heat source. The top-side heat spreading can also reduce the impact of thermal boundary resistance between the buffer and the substrate on the device thermal behavior. The very low electrical conductivity of diamond allows that it can directly contact the gate metal (which is very close to the heat source), being quite convenient to reduce the self-heating for the device under pulsed bias. Also, the diamond coated in vias etched in the substrate as heat spreading path (back-side heat spreading) was simulated. A competing mechanism influences the heat spreading ability, i.e., the etched vias would increase the device temperature due to the reduced heat sink while the coated diamond would decrease the device temperature due to its higher thermal conductivity. Therefore, relative thick coated diamond is needed in order to reduce the self-heating effect. The simulated local stress at the gate edge of the drain side for the device with standard and field plate gate structure were compared, which would be relevant to the device mechanical failure. Other stress simulations focused on the intrinsic stress in the diamond capping layer impact on the device electrical behaviors. The simulated stress and electrical output characteristics were compared to experimental data obtained by micro-Raman spectroscopy and electrical characterization, respectively. Results showed that the intrinsic stress in the capping layer caused the non-uniform distribution of 2DEG in the channel and the access region. Besides the enhancement of the device power output, intrinsic stress in the capping layer can potentially improve the device reliability by modulating the local stress at the gate edge of the drain side. Finally, the surface, buffer, and barrier traps effects were simulated in this work. Pulsed measurements showed that long gates and distances between gate and drain contact can increase the gate lag ratio (decrease the current collapse). This was explained by simulations on the surface traps effect. The simulations on buffer traps effects focused on illustrating the dynamic trapping/detrapping in the buffer and the self-heating impact on the device transient drain current. A model was presented to describe the trapping and detrapping in the barrier. The trapping was the electron direct tunneling from the gate metal while the detrapping was the electron emission into the conduction band described by phonon-assisted tunneling. The reverse gate current was simulated based on this model, whose mechanism can be attributed to the temperature and electric field dependent electron emission in the barrier. Furthermore, the mechanism of the device bias via the self-heating and electric field impact on the electron emission and the transient drain current were also illustrated.

Compositional differences within and between eukaryotic genomes

Eukaryotic genome similarity relationships are inferred using sequence information derived from large aggregates of genomic sequences. Comparisons within and between species sample sequences are based on the profile of dinucleotide relative abundance values (The profile is ρ*XY = f*XY/f*Xf*Y for all XY, where f*X denotes the frequency of the nucleotide X and f*XY denotes the frequency of the dinucleotide XY, both computed from the sequence concatenated with its inverted complement). Previous studies with respect to prokaryotes and this study document that profiles of different DNA sequence samples (sample size ≥50 kb) from the same organism are generally much more similar to each other than they are to profiles from other organisms, and that closely related organisms generally have more similar profiles than do distantly related organisms. On this basis we refer to the collection {ρ*XY} as the genome signature. This paper identifies ρ*XY extremes and compares genome signature differences for a diverse range of eukaryotic species. Interpretations on the mechanisms maintaining these profile differences center on genome-wide replication, repair, DNA structures, and context-dependent mutational biases. It is also observed that mitochondrial genome signature differences between species parallel the corresponding nuclear genome signature differences despite large differences between corresponding mitochondrial and nuclear signatures. The genome signature differences also have implications for contrasts between rodents and other mammals, and between monocot and dicot plants, as well as providing evidence for similarities among fungi and the diversity of protists.

The atypical codon usage of the plant psbA gene may be the remnant of an ancestral bias

The psbA gene of the chloroplast genome has a codon usage that is unusual for plant chloroplast genes. In the present study the evolutionary status of this codon usage is tested by reconstructing putative ancestral psbA sequences to determine the pattern of change in codon bias during angiosperm divergence. It is shown that the codon biases of the ancestral genes are much stronger than all extant flowering plant psbA genes. This is related to previous work that demonstrated a significant increase in synonymous substitution in psbA relative to other chloroplast genes. It is suggested, based on the two lines of evidence, that the codon bias of this gene currently is not being maintained by selection. Rather, the atypical codon bias simply may be a remnant of an ancestral codon bias that now is being degraded by the mutation bias of the chloroplast genome, in other words, that the psbA gene is not at equilibrium. A model for the evolution of selective pressure on the codon usage of plant chloroplast genes is discussed.

A physical basis for protein secondary structure

A physical theory of protein secondary structure is proposed and tested by performing exceedingly simple Monte Carlo simulations. In essence, secondary structure propensities are predominantly a consequence of two competing local effects, one favoring hydrogen bond formation in helices and turns, the other opposing the attendant reduction in sidechain conformational entropy on helix and turn formation. These sequence specific biases are densely dispersed throughout the unfolded polypeptide chain, where they serve to preorganize the folding process and largely, but imperfectly, anticipate the native secondary structure.

LC2, the Chlamydomonas Homologue of the t Complex-encoded Protein Tctex2, Is Essential for Outer Dynein Arm Assembly

Tctex2 is thought to be one of the distorter genes of the mouse t haplotype. This complex greatly biases the segregation of the chromosome that carries it such that in heterozygous +/t males, the t haplotype is transmitted to >95% of the offspring, a phenomenon known as transmission ratio distortion. The LC2 outer dynein arm light chain of Chlamydomonas reinhardtii is a homologue of the mouse protein Tctex2. We have identified Chlamydomonas insertional mutants with deletions in the gene encoding LC2 and demonstrate that the LC2 gene is the same as the ODA12 gene, the product of which had not been identified previously. Complete deletion of the LC2/ODA12 gene causes loss of all outer arms and a slow jerky swimming phenotype. Transformation of the deletion mutant with the cloned LC2/ODA12 gene restores the outer arms and rescues the motility phenotype. Therefore, LC2 is required for outer arm assembly. The fact that LC2 is an essential subunit of flagellar outer dynein arms allows us to propose a detailed mechanism whereby transmission ratio distortion is explained by the differential binding of mutant (t haplotype encoded) and wild-type dyneins to the axonemal microtubules of t-bearing or wild-type sperm, with resulting differences in their motility.

A preference for own-subspecies' song guides vocal learning in a song bird

In many song birds, males develop their songs as adults by imitating the songs of one or more tutors, memorized previously during a sensitive phase early in life. Previous work using two assays, the production of imitations by adult males and playback-induced calling by young birds during the sensitive phase for memorization, has shown that song birds can discriminate between their own and other species' songs. Herein I use both assays to show that male mountain white-crowned sparrows, Zonotrichia leucophrys oriantha, must learn to sing but have a genetic predisposition to memorize and learn the songs of their own subspecies. Playback tests to young naive birds before they even begin to sing reveal that birds give begging calls more in response to oriantha song than to songs of another species. After 10 days of tutoring with songs of either their own or another subspecies, birds continue to give stronger call responses to songs of their own subspecies, irrespective of whether they were tutored with them, and are more discriminating in distinguishing between different dialects of their own subspecies. The memory processes that facilitate recognition and discrimination of own-subspecies' song may also mediate the preferential imitation of song of a bird's own subspecies. Such perceptual biases could constrain the direction and rate of cultural evolution of learned songs.

Comparison of intron-containing and intron-lacking human genes elucidates putative exonic splicing enhancers

Of the rules used by the splicing machinery to precisely determine intron–exon boundaries only a fraction is known. Recent evidence suggests that specific short sequences within exons help in defining these boundaries. Such sequences are known as exonic splicing enhancers (ESE). A possible bioinformatical approach to studying ESE sequences is to compare genes that harbor introns with genes that do not. For this purpose two non-redundant samples of 719 intron-containing and 63 intron-lacking human genes were created. We performed a statistical analysis on these datasets of intron-containing and intron-lacking human coding sequences and found a statistically significant difference (P = 0.01) between these samples in terms of 5–6mer oligonucleotide distributions. The difference is not created by a few strong signals present in the majority of exons, but rather by the accumulation of multiple weak signals through small variations in codon frequencies, codon biases and context-dependent codon biases between the samples. A list of putative novel human splicing regulation sequences has been elucidated by our analysis.

Estimating risks in declining populations with poor data

Census data on endangered species are often sparse, error-ridden, and confined to only a segment of the population. Estimating trends and extinction risks using this type of data presents numerous difficulties. In particular, the estimate of the variation in year-to-year transitions in population size (the “process error” caused by stochasticity in survivorship and fecundities) is confounded by the addition of high sampling error variation. In addition, the year-to-year variability in the segment of the population that is sampled may be quite different from the population variability that one is trying to estimate. The combined effect of severe sampling error and age- or stage-specific counts leads to severe biases in estimates of population-level parameters. I present an estimation method that circumvents the problem of age- or stage-specific counts and is markedly robust to severe sampling error. This method allows the estimation of environmental variation and population trends for extinction-risk analyses using corrupted census counts—a common type of data for endangered species that has hitherto been relatively unusable for these analyses.

Comparação entre prótese plástica e metálica na paliação endoscópica da obstrução maligna da via biliar: revisão sistemática e metanálise baseadas em estudos randomizados

INTRODUÇÃO: A prótese biliar endoscópica é aceita em todo o mundo como a primeira escolha de tratamento paliativo na obstrução biliar maligna. Atualmente ainda persistem dois tipos de materiais utilizados em sua confecção: plástico e metal. Consequentemente, muitas dúvidas surgem quanto a qual deles é o mais benéfico para o paciente. Esta revisão reúne as informações disponíveis da mais alta qualidade sobre estes dois tipos de prótese, fornecendo informações em relação à disfunção, complicação, taxas de reintervenção, custos, sobrevida e tempo de permeabilidade; e pretende ajudar a lidar com a prática clínica nos dias de hoje. OBJETIVO: Analisar, através de metanálise, os benefícios de dois tipos de próteses na obstrução biliar maligna inoperável. MÉTODOS: Uma revisão sistemática de ensaios clínicos randomizados (RCT) foi conduzida, com a última atualização em março de 2015, utilizando EMBASE, CINAHL (EBSCO), Medline, Lilacs / Centro (BVS), Scopus, o CAPES (Brasil), e literatura cinzenta. As informações dos estudos selecionados foram extraídas tendo em vista seis desfechos: primariamente disfunção, taxas de reintervenção e complicações; e, secundariamente, custos, sobrevivência e tempo de permeabilidade. Os dados sobre as características dos participantes do RCT, critérios de inclusão e exclusão e tipos de próteses também foram extraídos. Os vieses foram avaliados principalmente através da escala de Jadad. Esta metanálise foi registrada no banco de dados PROSPERO pelo número CRD42014015078. A análise do risco absoluto dos resultados foi realizada utilizando o software RevMan 5, calculando as diferenças de risco (RD) de variáveis dicotômicas e média das diferenças (MD) de variáveis contínuas. Os dados sobre a RD e MD para cada desfecho primário foram calculados utilizando o teste de Mantel-Haenszel e a inconsistência foi avaliada com o teste Qui-quadrado (Chi2) e o método de Higgins (I2). A análise de sensibilidade foi realizada com a retirada de estudos discrepantes e a utilização do efeito aleatório. O teste t de Student foi utilizado para a comparação das médias aritméticas ponderadas, em relação aos desfechos secundários. RESULTADOS: Inicialmente foram identificados 3660 estudos; 3539 foram excluídos por título ou resumo, enquanto 121 estudos foram totalmente avaliados e foram excluídos, principalmente por não comparar próteses metálicas (SEMS) e próteses plásticas (PS), levando a treze RCT selecionados e 1133 indivíduos metanálise. A média de idade foi de 69,5 anos, e o câncer mais comum foi de via biliar (proximal) e pancreático (distal). O diâmetro de SEMS mais utilizado foi de 10 mm (30 Fr) e o diâmetro de PS mais utilizado foi de 10 Fr. Na metanálise, SEMS tiveram menor disfunção global em comparação com PS (21,6% versus 46,8% p < 0,00001) e menos reintervenções (21,6% versus 56,6% p < 0,00001), sem diferença nas complicações (13,7% versus 15,9% p = 0,16). Na análise secundária, a taxa média de sobrevida foi maior no grupo SEMS (182 contra 150 dias - p < 0,0001), com um período maior de permeabilidade (250 contra 124 dias - p < 0,0001) e um custo semelhante por paciente, embora menor no grupo SEMS (4.193,98 contra 4.728,65 Euros - p < 0,0985). CONCLUSÃO: SEMS estão associados com menor disfunção, menores taxas de reintervenção, melhor sobrevida e maior tempo de permeabilidade. Complicações e custos não apresentaram diferença

Development of data reduction pipelines for GTC instruments at the UCM

The UCM Instrumentation Group (GUAIX) is developing currently Data Reduction Pipelines (DRP) for four instruments of the GTC: EMIR, FRIDA, MEGARA and MIRADAS. The purpose of the DRPs is to provide astronomers scientific quality data, removing instrumental biases, calibrating the images in physical units and providing a estimation of the associated uncertainties.

Estudo de transistores de tunelamento induzido por efeito de campo (TFET) construídos em nanofio.

Esse trabalho de mestrado teve como estudo o transistor Túnel-FET (TFET) fabricado em estrutura de nanofio de silício. Este estudo foi feito de forma teórica (simulação numérica) e experimental. Foram estudadas as principais características digitais e analógicas do dispositivo e seu potencial para uso em circuitos integrados avançados para a próxima década. A análise foi feita através da extração experimental e estudo dos principais parâmetros do dispositivo, tais como inclinação de sublimiar, transcondutância (gm), condutância de saída (gd), ganho intrínseco de tensão (AV) e eficiência do transistor. As medidas experimentais foram comparadas com os resultados obtidos pela simulação. Através do uso de diferentes parâmetros de ajuste e modelos de simulação, justificou-se o comportamento do dispositivo observado experimentalmente. Durante a execução deste trabalho estudou-se a influência da escolha do material de fonte no desempenho do dispositivo, bem como o impacto do diâmetro do nanofio nos principais parâmetros analógicos do transistor. Os dispositivos compostos por fonte de SiGe apresentaram valores maiores de gm e gd do que aqueles compostos por fonte de silício. A diferença percentual entre os valores de transcondutância para os diferentes materiais de fonte variou de 43% a 96%, sendo dependente do método utilizado para comparação, e a diferença percentual entre os valores de condutância de saída variou de 38% a 91%. Observou-se também uma degradação no valor de AV com a redução do diâmetro do nanofio. O ganho calculado a partir das medidas experimentais para o dispositivo com diâmetro de 50 nm é aproximadamente 45% menor do que o correspondente ao diâmetro de 110 nm. Adicionalmente estudou-se o impacto do diâmetro considerando diferentes polarizações de porta (VG) e concluiu-se que os TFETs apresentam melhor desempenho para baixos valores de VG (houve uma redução de aproximadamente 88% no valor de AV com o aumento da tensão de porta de 1,25 V para 1,9 V). A sobreposição entre porta e fonte e o perfil de dopantes na junção de tunelamento também foram analisados a fim de compreender qual combinação dessas características resultariam em um melhor desempenho do dispositivo. Observou-se que os melhores resultados estavam associados a um alinhamento entre o eletrodo de porta e a junção entre fonte e canal e a um perfil abrupto de dopantes na junção. Por fim comparou-se a tecnologia MOS com o TFET, obtendo-se como resultado um maior valor de AV (maior do que 40 dB) para o TFET.

The Perks of Playing Dress-Up: Psychological Impressions of Fashion and Feminism.

Research on the psychological impact of women's fashion has focused on fashion's negative influence over how women think and feel about themselves. Several studies have examined the relationship between fashion and women's self-appraisals (Martin & Gentry, 1997; Pinhas, Toner, Ali, Garfinkel, & Stuckless, 1999; Tiggemann, Polivy, & Hargreaves, 2009), although few investigations have explored the range of viewpoints that arise when women interact with their own personal style or with other forms of fashion media. This paper presents a narrative review of what has been written about fashion in clinical research. I briefly discuss why this is an important topic and why fashion has psychological meaning. Cognitive behavioral therapy (CBT) is considered in the exploration of fashion's impact on conjuring unproductive and productive schemas (Beck, 1976; Wright, Basco, & Thase, 2006). This discussion includes a presentation of interviews with female consultants, hypothetical examples, my own accounts, and feminist perspectives. While emphasizing the potential biases of women's interactions with fashion, I discuss matters of gender performance and reflections on clinical work. The purpose of this article is to present a pro-social defense of fashion. I do this by acquiring personal chronicles, applying those findings to the current body of research, and adding to the continued investigation of why women's fashion is still important in a postfeminist world.

What We Don't Know Can Hurt Us: Clinical Inference Bias and Constructed Memories of Sexual Abuse.

The false memory/recovered memory debate, research regarding the malleability of memory, and the current lack of methods for validating recovered memories all support the view that heightened care is required of therapists dealing with clients whom they suspect have been sexually abused. The judgmental heuristics that underlie the major clinical inference biases of confirmatory bias, biased covariation, base rate fallacies, and schematic processing errors are all relevant to the processes leading to therapist-client constructions of memories of sexual abuse. Suggestions for minimizing each of these biases are offered. Personal motivations of the client and client suggestibility are factors that may contribute to the construction of memories of sexual abuse, and suggestions for minimizing the impact of these motivations are offered. In conclusion, general suggestions for minimizing the impact of clinical inference biases within the sexual abuse treatment context are summarized.

Investigating evidence for different black hole accretion modes since redshift z ∼ 1

Chandra data in the COSMOS, AEGIS-XD and 4 Ms Chandra Deep Field South are combined with multiwavelength photometry available in those fields to determine the rest-frame U − V versus V − J colours of X-ray AGN hosts in the redshift intervals 0.1 < z < 0.6 (mean z¯=0.40) and 0.6 < z < 1.2 (mean z¯=0.85). This combination of colours provides an effective and least model-dependent means of separating quiescent from star-forming, including dust reddened, galaxies. Morphological information emphasizes differences between AGN populations split by their U − V versus V − J colours. AGN in quiescent galaxies consist almost exclusively of bulges, while star-forming hosts are equally split between early- and late-type hosts. The position of AGN hosts on the U − V versusV − J diagram is then used to set limits on the accretion density of the Universe associated with evolved and star-forming systems independent of dust induced biases. It is found that most of the black hole growth at z≈ 0.40 and 0.85 is associated with star-forming hosts. Nevertheless, a non-negligible fraction of the X-ray luminosity density, about 15–20 per cent, at both z¯=0.40 and 0.85, is taking place in galaxies in the quiescent region of the U − V versus V − J diagram. For the low-redshift sub-sample, 0.1 < z < 0.6, we also find tentative evidence, significant at the 2σ level, that AGN split by their U − V and V − J colours have different Eddington ratio distributions. AGN in blue star-forming hosts dominate at relatively high Eddington ratios. In contrast, AGN in red quiescent hosts become increasingly important as a fraction of the total population towards low Eddington ratios. At higher redshift, z > 0.6, such differences are significant at the 2σ level only for sources with Eddington ratios ≳ 10^− 3. These findings are consistent with scenarios in which diverse accretion modes are responsible for the build-up of supermassive black holes at the centres of galaxies. We compare these results with the predictions of theGALFORM semi-analytic model for the cosmological evolution of AGN and galaxies. This model postulates two black hole fuelling modes, the first is linked to star formation events and the second takes place in passive galaxies. GALFORM predicts that a substantial fraction of the black hole growth at z < 1 is associated with quiescent galaxies, in apparent conflict with the observations. Relaxing the strong assumption of the model that passive AGN hosts have zero star formation rate could bring those predictions in better agreement with the data.