Desenvolvimento de arquitetura para sistemas de reconhecimento automático de voz baseados em modelos ocultos de Markov

Este trabalho foi realizado dentro da área de reconhecimento automático de voz (RAV). Atualmente, a maioria dos sistemas de RAV é baseada nos modelos ocultos de Markov (HMMs) [GOM 99] [GOM 99b], quer utilizando-os exclusivamente, quer utilizando-os em conjunto com outras técnicas e constituindo sistemas híbridos. A abordagem estatística dos HMMs tem mostrado ser uma das mais poderosas ferramentas disponíveis para a modelagem acústica e temporal do sinal de voz. A melhora da taxa de reconhecimento exige algoritmos mais complexos [RAV 96]. O aumento do tamanho do vocabulário ou do número de locutores exige um processamento computacional adicional. Certas aplicações, como a verificação de locutor ou o reconhecimento de diálogo podem exigir processamento em tempo real [DOD 85] [MAM 96]. Outras aplicações tais como brinquedos ou máquinas portáveis ainda podem agregar o requisito de portabilidade, e de baixo consumo, além de um sistema fisicamente compacto. Tais necessidades exigem uma solução em hardware. O presente trabalho propõe a implementação de um sistema de RAV utilizando hardware baseado em FPGAs (Field Programmable Gate Arrays) e otimizando os algoritmos que se utilizam no RAV. Foi feito um estudo dos sistemas de RAV e das técnicas que a maioria dos sistemas utiliza em cada etapa que os conforma. Deu-se especial ênfase aos Modelos Ocultos de Markov, seus algoritmos de cálculo de probabilidades, de treinamento e de decodificação de estados, e sua aplicação nos sistemas de RAV. Foi realizado um estudo comparativo dos sistemas em hardware, produzidos por outros centros de pesquisa, identificando algumas das suas características mais relevantes. Foi implementado um modelo de software, descrito neste trabalho, utilizado para validar os algoritmos de RAV e auxiliar na especificação em hardware. Um conjunto de funções digitais implementadas em FPGA, necessárias para o desenvolvimento de sistemas de RAV é descrito. Foram realizadas algumas modificações nos algoritmos de RAV para facilitar a implementação digital dos mesmos. A conexão, entre as funções digitais projetadas, para a implementação de um sistema de reconhecimento de palavras isoladas é aqui apresentado. A implementação em FPGA da etapa de pré-processamento, que inclui a pré-ênfase, janelamento e extração de características, e a implementação da etapa de reconhecimento são apresentadas finalmente neste trabalho.

Designing single event upset mitigation techniques for large SRAM-Based FPGA components

This thesis presents the study and development of fault-tolerant techniques for programmable architectures, the well-known Field Programmable Gate Arrays (FPGAs), customizable by SRAM. FPGAs are becoming more valuable for space applications because of the high density, high performance, reduced development cost and re-programmability. In particular, SRAM-based FPGAs are very valuable for remote missions because of the possibility of being reprogrammed by the user as many times as necessary in a very short period. SRAM-based FPGA and micro-controllers represent a wide range of components in space applications, and as a result will be the focus of this work, more specifically the Virtex® family from Xilinx and the architecture of the 8051 micro-controller from Intel. The Triple Modular Redundancy (TMR) with voters is a common high-level technique to protect ASICs against single event upset (SEU) and it can also be applied to FPGAs. The TMR technique was first tested in the Virtex® FPGA architecture by using a small design based on counters. Faults were injected in all sensitive parts of the FPGA and a detailed analysis of the effect of a fault in a TMR design synthesized in the Virtex® platform was performed. Results from fault injection and from a radiation ground test facility showed the efficiency of the TMR for the related case study circuit. Although TMR has showed a high reliability, this technique presents some limitations, such as area overhead, three times more input and output pins and, consequently, a significant increase in power dissipation. Aiming to reduce TMR costs and improve reliability, an innovative high-level technique for designing fault-tolerant systems in SRAM-based FPGAs was developed, without modification in the FPGA architecture. This technique combines time and hardware redundancy to reduce overhead and to ensure reliability. It is based on duplication with comparison and concurrent error detection. The new technique proposed in this work was specifically developed for FPGAs to cope with transient faults in the user combinational and sequential logic, while also reducing pin count, area and power dissipation. The methodology was validated by fault injection experiments in an emulation board. The thesis presents comparison results in fault coverage, area and performance between the discussed techniques.

Uma interface para o aumento da faixa de freqüências de operação de FPAAS

O crescente avanço nas mais diversas áreas da eletrônica, desde instrumentação em baixa freqüência até telecomunicações operando em freqüências muito elevadas, e a necessidade de soluções baratas em curto espaço de tempo que acompanhem a demanda de mercado, torna a procura por circuitos programáveis, tanto digitais como analógicos, um ponto comum em diversas pesquisas. Os dispositivos digitais programáveis, que têm como grande representante os Field Programmable Gate Arrays (FPGAs), vêm apresentando um elevado e contínuo crescimento em termos de complexidade, desempenho e número de transistores integrados, já há várias décadas. O desenvolvimento de dispositivos analógicos programáveis (Field Programmable Analog Arrays – FPAAs), entretanto, esbarra em dois pontos fundamentais que tornam sua evolução um tanto latente: a estreita largura de banda alcançada, conseqüência da necessidade de um grande número de chaves de programação e reconfiguração, e a elevada área consumida por componentes analógicos como resistores e capacitores, quando integrados em processos VLSI Este trabalho apresenta uma proposta para aumentar a faixa de freqüências das aplicações passíveis de serem utilizadas tanto em FPAAs comerciais quanto em outros FPAAs, através da utilização de uma interface de translação e seleção de sinais, mantendo características de programabilidade do FPAA em questão, sem aumentar em muito sua potência consumida. A proposta, a simulação e a implementação da interface são apresentadas ao longo desta dissertação. Resultados de simulação e resultados práticos obtidos comprovam a eficácia da proposta.

Controlador adaptativo backstepping a estrutura variável

Neste trabalho, um controlador adaptativo backstepping a estrutura variável (Variable Structure Adaptive Backstepping Controller, VS-ABC) é apresentado para plantas monovariáveis, lineares e invariantes no tempo com grau relativo unitário. Ao invés das tradicionais leis integrais para estimação dos parâmetros da planta, leis chaveadas são utilizadas com o objetivo de aumentar a robustez em relação a incertezas paramétricas e distúrbios externos, bem como melhorar o desempenho transitório do sistema. Adicionalmente, o projeto do novo controlador é mais intuitivo quando comparado ao controlador backstepping original, uma vez que os relés introduzidos apresentam amplitudes diretamente relacionadas com os parâmetros nominais da planta. Esta nova abordagem, com uso de estrutura variável, também reduz a complexidade das implementações práticas, motivando a utilização de componentes industriais, tais como, FPGAs (Field Programmable Gate Arrays ), MCUs (Microcontrollers) e DSPs (Digital Signal Processors). Simulações preliminares para um sistema instável de primeira e segunda ordem são apresentadas de modo a corroborar os estudos. Um dos exemplos de Rohrs é ainda abordado através de simulações, para os dois cenários adaptativos: o controlador backstepping adaptativo original e o VS-ABC

Common architecture for discrete wavelet transform analysis and synthesis with sequential and constant processing elements

This paper adresses the problem on processing biological data such as cardiac beats, audio and ultrasonic range, calculating wavelet coefficients in real time, with processor clock running at frequency of present ASIC's and FPGA. The Paralell Filter Architecture for DWT has been improved, calculating wavelet coefficients in real time with hardware reduced to 60%. The new architecture, which also processes IDWT, is implemented with the Radix-2 or the Booth-Wallace Constant multipliers. Including series memory register banks, one integrated circuit Signal Analyzer, ultrasonic range, is presented.

A petri net based timing model for hardware/software co-design of digital systems

We have recently proposed an extension to Petri nets in order to be able to directly deal with all aspects of embedded digital systems. This extension is meant to be used as an internal model of our co-design environment. After analyzing relevant related work, and presenting a short introduction to our extension as a background material, we describe the details of the timing model we use in our approach, which is mainly based in Merlin's time model. We conclude the paper by discussing an example of its usage. © 2004 IEEE.

An AC voltage regulator with high-power-factor, and control using a FPGA device

This paper presents the analysis, design, simulation, and experimental results for a high frequency high Power-Factor (PF) AC (Alternate Current) voltage regulator, using a Sepic converter as power stage. The control technique employed to impose a sinusoidal input current waveform, with low Total Harmonic Distortion (THD), is the sinusoidal variable hysteresis control. The control technique was implemented in a FPGA (Field Programmable Gate Array) device, using a Hardware Description Language (VHDL). Through the use of the proposed control technique, the AC voltage regulator performs active power-factor correction, and low THD in the input current, for linear and non-linear loads, satisfying the requirements of the EEC61000-3-2 standards. Experimental results from an example prototype, designed for 300W of nominal output power, 50kHz (switching frequency), and 127Vrms of nominal input and output voltages, are presented in order to validate the proposed AC regulator. © 2005 IEEE.

Conception of wheeled mobile robots with reconfigurable control using integrate prototyping

With the fast innovation of the hardware and software technologies using rapid prototyping devices, with application in the robotics and automation, more and more it becomes necessary the development of applications based on methodologies that facilitate future modifications, updates and enhancements in the original projected system. This paper presents a conception of mobile robots using rapid prototyping, distributing the several control actions in growing levels of complexity and using resources of reconfigurable computing proposal oriented to embed systems implementation. Software and the hardware are structuralized in independents blocks, with connection through common bus. The study and applications of new structures control that permits good performance in relation to the parameter variations. This kind of controller can be tested on different platform representing the wheeled mobile robots using reprogrammable logic components (FPGA). © 2006 IEEE.

Discrete wavelet transform signal analyzer

This paper addresses the problem of processing biological data, such as cardiac beats in the audio and ultrasonic range, and on calculating wavelet coefficients in real time, with the processor clock running at a frequency of present application-specified integrated circuits and field programmable gate array. The parallel filter architecture for discrete wavelet transform (DWT) has been improved, calculating the wavelet coefficients in real time with hardware reduced up to 60%. The new architecture, which also processes inverse DWT, is implemented with the Radix-2 or the Booth-Wallace constant multipliers. One integrated circuit signal analyzer in the ultrasonic range, including series memory register banks, is presented. © 2007 IEEE.

HW/SW for an intelligent transducer network based on IEEE 1451 standard

This work describes a hardware/software co-design system development, named IEEE 1451 platform, to be used in process automation. This platform intends to make easier the implementation of IEEE standards 1451.0, 1451.1, 1451.2 and 1451.5. The hardware was built using NIOS II processor resources on Alteras Cyclone II FPGA. The software was done using Java technology and C/C++ for the processors programming. This HW/SW system implements the IEEE 1451 based on a control module and supervisory software for industrial automation. © 2011 Elsevier B.V.

Approximation of hyperbolic tangent activation function using hybrid methods

Artificial Neural Networks are widely used in various applications in engineering, as such solutions of nonlinear problems. The implementation of this technique in reconfigurable devices is a great challenge to researchers by several factors, such as floating point precision, nonlinear activation function, performance and area used in FPGA. The contribution of this work is the approximation of a nonlinear function used in ANN, the popular hyperbolic tangent activation function. The system architecture is composed of several scenarios that provide a tradeoff of performance, precision and area used in FPGA. The results are compared in different scenarios and with current literature on error analysis, area and system performance. © 2013 IEEE.

Pré-regulador retificador boost com controle digital por valores médios, para sistema de iluminação fluorescente multi-lâmpadas, utilizando dispositivo FPGA e VHDL

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Modeling of state machines in VHDL for encapsulation of DNP3 Protocol in P2P ZigBee Network

Several countries have invested in technologies for Smart Grids. Among such protocols designed cover this area, highlights the DNP3 (Distributed Network Protocol version 3). Although the DNP3 be developed for operation over the serial interface, there is a trend in the literature to the use of other interfaces. The Zigbee wireless interface has become more popular in the industrial applications. In order to study the challenges of integrating of these two protocols, this article is presented the analysis of DNP3 protocol stack through state machines The encapsulation of DNP3 messages in P2P (point-to-point) ZigBee Network, may assist in the discovery and solution of failures of availability and security of this integration. The ultimate goal is to merge the features of DNP3 and Zigbee stacks, and display a solution that provides the benefits of wireless environment, without impairment of security required for Smart Grid applications.

LALP: a language to program custom FPGA-based acceleration engines

Field-Programmable Gate Arrays (FPGAs) are becoming increasingly important in embedded and high-performance computing systems. They allow performance levels close to the ones obtained with Application-Specific Integrated Circuits, while still keeping design and implementation flexibility. However, to efficiently program FPGAs, one needs the expertise of hardware developers in order to master hardware description languages (HDLs) such as VHDL or Verilog. Attempts to furnish a high-level compilation flow (e.g., from C programs) still have to address open issues before broader efficient results can be obtained. Bearing in mind an FPGA available resources, it has been developed LALP (Language for Aggressive Loop Pipelining), a novel language to program FPGA-based accelerators, and its compilation framework, including mapping capabilities. The main ideas behind LALP are to provide a higher abstraction level than HDLs, to exploit the intrinsic parallelism of hardware resources, and to allow the programmer to control execution stages whenever the compiler techniques are unable to generate efficient implementations. Those features are particularly useful to implement loop pipelining, a well regarded technique used to accelerate computations in several application domains. This paper describes LALP, and shows how it can be used to achieve high-performance computing solutions.