A utilização de várias ferramentas de injeção de falhas em um mesmo experimento de testes fornece mais subsídios para os resultados alcançados, tornando a atividade mais efetiva e menos sujeita a erros de interpretação. Neste sentido, as cargas de falhas possuem um importante papel, visto que elas compõem a principal entrada a ser fornecida nestas ferramentas. No entanto, os mecanismos oferecidos, nas ferramentas de injeção de falhas existentes, para esta especificação de cargas de falhas, possuem um baixo grau de usabilidade e expressividade. Por este motivo, o presente trabalho aborda uma metodologia, na qual cenários detalhados de testes, que envolvam experimentos com injeção de falhas, possam ser especificados de maneira simples, homogênea e padronizada. Para isso, é proposta a criação de um ambiente para a especificação destas cargas de falhas, denominado como jFaultload. Este ambiente, por sua vez, utiliza-se de um subconjunto da linguagem Java para a especificação destas cargas de falhas, ficando responsável ainda pela tradução, desta carga em Java, para os respectivos formatos de carga referentes a cada injetor de falhas utilizado em um dado experimento. Para efeito de exemplo e validação do ambiente proposto, as ferramentas FIRMAMENT, MENDOSUS e FAIL/FCI são integradas neste ambiente, tornando assim o cenário de testes amplamente detalhado. O serviço a ser testado, visando a demonstração da usabilidade e expressividade da solução proposta, foi uma sessão de video streaming, utilizando-se para isso do protocolo RTP, onde uma campanha de testes foi realizada com o injetor FIRMAMENT. / Use of two or more fault injection tools in a test campaign enriches the scenario obtained from a test execution. Faultloads represent the main input for these tools but their specification mechanisms lack usability and expressiveness. This thesis presents a full test scenario featuring the use of jFaultload, which applies Java for the specification of faultloads and translates them to specific formats that are appropriate to each available fault injector. FIRMAMENT, MENDOSUS and FAIL/FCI, fault injectors for communication systems, were integrated in the environment and complete the test scenario. The service under test used to demonstrate the usability and expressiveness of our solution is a video streaming session using RTP Protocol, which a test campaign was executed through the FIRMAMENT fault injector.

A complexidade dos sistemas de visualização imersivos pode variar tremendamente conforme a sua aplicação. Algumas ferramentas mais simples fazem uso de um único óculos de Realidade Virtual como infraestrutura de visualização. No entanto, aplicações mais complexas, como simuladores e outras ferramentas de treinamento, podem necessitar de uma infraestrutura distribuída, contendo diversos computadores e telas. Alguns simuladores e outras aplicações de treinamento fazem uso frequente de periféricos sofisticados de interação, que reproduzem de maneira fiel os elementos encontrados no cenário real. Além disto, o espaço de treinamento pode ser compartilhado por dois ou mais usuários. Estes requisitos acabam por impor o uso de sistemas de visualização complexos e distribuídos, que visam cobrir de maneira quase completa o campo de vis~ao destes usuários. Por causa das características deste tipo de sistema, as aplicações desenvolvidas nestes cenários são inerentemente complexas, pois frequentemente consideram aspectos específicos da infraestrutura para realizar a distribuição e o sincronismo da cena virtual. Esta complexidade dificulta o desenvolvimento, a manutenção e a interoperabilidade destas ferramentas. Este trabalho apresenta uma arquitetura de comunicação para promover o uso de sistemas imersivos de forma simples e transparente para as aplicações, viabilizando o uso de infraestruturas complexas e distribuídas. A arquitetura proposta utiliza o mecanismo de substituição do driver OpenGL para obter, de forma automática, a distribuição do aspecto gráfico das aplicações. Apesar deste conceito já ter sido discutido na literatura, esta proposta apresenta um conjunto de técnicas para contornar as limitações inerentes desta abordagem e obter ganhos de desempenho significativos, com resultados consistentes em um amplo conjunto de infraestruturas. As técnicas apresentadas neste trabalho sugerem, entre outras coisas, o uso de recursos modernos do padrão OpenGL para reduzir o volume de comunicação entre CPU e GPU. Um dos recursos avaliados foi o uso de mecanismos de renderização indireta, onde a aplicação armazena os comandos de renderização na memória da placa gráfica. Juntamente com esta técnica, o trabalho também investigou o uso de um algoritmo de culling na própria GPU, o que permitiu que esta otimização fosse utilizada mesmo em sistemas com arranjos mais complexos de tela. Os resultados obtidos mostram que a aplicação pode exibir o seu conteúdo em um conjunto amplo de sistemas imersivos, contendo mais resolução e mais geometria visível, sem deteriorar o seu desempenho. Os testes foram conduzidos em diferentes infraestruturas e com cenas de tamanhos variáveis. Nos casos mais complexos, as técnicas propostas podem reduzir em 86% o tempo médio de renderização, quando comparadas com as abordagens tradicionais. / The complexity of immersive visualization systems can vary tremendously depending on their application. Some simple tools might only require a conventional virtual reality goggle as a visualization infrastructure. However, more complex applications, such as simulators and other training tools, might require a distributed infrastructure, containing several computers and screens. Some training applications and simulators invariably make use of physical peripherals for interaction, which are designed to faithfully reproduce the elements found in real scenarios. Furthermore, the training area may be shared by two or more users. These requirements usually impose the use of complex and distributed imaging systems, which are intended to cover almost the entire field of view of the users involved. Because of the characteristics of this type of system, the applications developed for these infrastructures are inherently complex. They are required to consider specific aspects of the infrastructure itself to carry out the distribution and synchronization of the virtual scene. This complexity hampers the development, maintenance and interoperability of these tools. This work presents a communication architecture to promote the use of immersive systems by allowing applications to use complex and distributed infrastructures in a simple and transparent way. The proposed architecture uses the approach of replacing the OpenGL driver to transparently achieve graphics distribution. Although this has already been discussed in the literature, this document presents a set of techniques to overcome the inherent limitations of this approach and ultimately achieve significant performance gains, with consistent results across a broad range of infrastructures. The techniques presented here suggest, among other things, the use of modern features of the OpenGL standard to reduce the communication overhead between CPU and GPU. One of the features evaluated was the usage of indirect rendering, where the application stores all the rendering commands in the graphics card dedicated memory. Along with this feature, the work also investigated the use of a culling algorithm on the GPU itself, which allowed this optimization to be used even on systems containing screens with a more complex layout. The results show that the application can render its content in a wide range of immersive systems, with higher resolution and more visible geometry, without degrading its performance. The tests were conducted at different infrastructures and scenes with variable sizes. In the more complex use cases, the proposed techniques can reduce by up to 86% the average rendering time, when compared to the traditional approaches.

O surgimento de volumes de dados na ordem de petabytes cria a necessidade de desenvolver-se novas soluções que viabilizem o tratamento dos dados através do uso de sistemas de computação intensiva, como o MapReduce. O MapReduce é um framework de programação que apresenta duas funções: uma de mapeamento, chamada Map, e outra de redução, chamada Reduce, aplicadas a uma determinada entrada de dados. Este modelo de programação é utilizado geralmente em grandes clusters e suas tarefas Map ou Reduce são normalmente independentes entre si. O programador é abstraído do processo de paralelização como divisão e distribuição de dados, tolerância a falhas, persistência de dados e distribuição de tarefas. A motivação deste trabalho é aplicar o modelo de computação intensiva do MapReduce com grande volume de dados para uso em ambientes desktop grid. O objetivo então é investigar os algoritmos do MapReduce para adequar a computação intensiva aos ambientes heterogêneos. O trabalho endereça o problema da heterogeneidade de recursos, não tratando neste momento a volatilidade das máquinas. Devido às deficiências encontradas no MapReduce em ambientes heterogêneos foi proposto o MR-A++, que é um MapReduce com algoritmos adequados ao ambiente heterogêneo. O modelo do MR-A++ cria uma tarefa de medição para coletar informações, antes de ocorrer a distribuição dos dados. Assim, as informações serão utilizadas para gerenciar o sistema. Para avaliar os algoritmos alterados foi empregada a Análise 2k Fatorial e foram executadas simulações com o simulador MRSG. O simulador MRSG foi construído para o estudo de ambientes (homogêneos e heterogêneos) em larga escala com uso do MapReduce. O pequeno atraso introduzido na fase de setup da computação é compensado com a adequação do ambiente heterogêneo à capacidade computacional das máquinas, com ganhos de redução de tempo de execução dos jobs superiores a 70 % em alguns casos. / The emergence of data volumes in the order of petabytes creates the need to develop new solutions that make possible the processing of data through the use of intensive computing systems, as MapReduce. MapReduce is a programming framework that has two functions: one called Map, mapping, and another reducing called Reduce, applied to a particular data entry. This programming model is used primarily in large clusters and their tasks are normally independent. The programmer is abstracted from the parallelization process such as division and data distribution, fault tolerance, data persistence and distribution of tasks. The motivation of this work is to apply the intensive computation model of MapReduce with large volume of data in desktop grid environments. The goal then is to investigate the intensive computing in heterogeneous environments with use MapReduce model. First the problem of resource heterogeneity is solved, not treating the moment of the volatility. Due to deficiencies of the MapReduce model in heterogeneous environments it was proposed the MR-A++; a MapReduce with algorithms adequated to heterogeneous environments. The MR-A++ model creates a training task to gather information prior to the distribution of data. Therefore the information will be used to manager the system. To evaluate the algorithms change it was employed a 2k Factorial analysis and simulations with the simulant MRSG built for the study of environments (homogeneous and heterogeneous) large-scale use of MapReduce. The small delay introduced in phase of setup of computing compensates with the adequacy of heterogeneous environment to computational capacity of the machines, with gains in the run-time reduction of jobs exceeding 70% in some cases.

Hydrometeorology researchers use data-products like rainfall maps, quarter-hourly, hourly & daily rainfall accumulation maps, and reflectivity maps to drive their hydrological models. There are many sources available from which to generate such products e.g. (1) Weather-radar scan data, (2) Rain-gauge data and (3) Satellite maps, to name a few. Researchers, however, prefer to deal directly with the data-product, rather than deal with the details of sensor data-collection, data-management, storage, quality-control, processing and format conversion. Many researchers would additionally like a real-time continuous 'feed' of customized hydrometeorology data-products to drive their hydrological models in a real-time continuous fashion. Hydro-NEXRAD II is a new system that is currently being developed at IIHR-Hydroscience and Engineering (at The University of Iowa) to address this need. The system provides rapid and user-friendly access to the aforementioned user-customized hydrometeorology data-products in file formats convenient to researchers. Products are generated using near real-time Weather Surveillance Doppler Radar (WSR-88D) data from the National Weather Service's NEXRAD (NEXt generation RADar) radar-network. The architecture, design and engineering details of Hydro-NEXRAD II are the subject of this thesis.

Ubiquitous computing has become a widespread phenomenon in today's modern world, with the computing technology integrating with our daily life in an invisible manner. Embedded systems and wireless sensor networks are popular choices to achieve this. Programming embedded and sensor network systems has always been a challenge for the programmers due to the lack of sufficient high-level programming support. To deal with this serious limitation, we have developed DESAL (Dynamic Embedded Sensing and Actuation Language) which is a user-friendly high-level programming language for wireless sensor networks with an integrated middleware, which hides the low-level detail from the programmers. In this thesis we present the design and development of DESAL. We have made DESAL programs rule based. Programs are written in guard-action format defined in terms of the program states. There are established formal correctness proving methods that can work on guard-action formats to mathematically check a program for errors. Also, there is no hidden control context like events or interrupts. Time synchronization has been developed as part of the middleware that lets DESAL programs to coordinate through synchronized actions throughout the network. This facilitates classic coordination algorithms like clock synchronization, spanning tree construction and consensus. Also, synchronized wake up saves energy. Neighborhood management, including node discovery and monitoring, is also provided by the middleware. DESAL programs communicate via state sharing. There is no network programming required. The middleware provides that automatically. Combining all these features DESAL provides major network management services, and yet presents the users with a simple high-level programming interface. We implemented the DESAL compiler to convert DESAL programs to NesC on TinyOS and to Java. Another novel feature we have introduced in DESAL is a variable of type 'token'. The concept of token is commonly used in mutual exclusion algorithms. One of the case studies we have done uses the token variable to achieve increased lifetime of sensors in a ring topology. The working of token is hidden from the user. The token mechanism developed achieves self-stabilization in the system. Another case study with tokens involves selective activation of RFID tags in a scenario where among the three RFID tags present only one can work at a time. Struct is a new data structure introduced in DESAL. Sometimes we need to group together two or more variables. It is important to receive them at the same time. Hence, it is important to send these grouped data over the radio together. Struct does that. Function is another newly added feature to DESAL. Function is added to group together repeated statements in a program. The unique feature of function is that, it uses only global variables. No new local variable is declared. This can significantly reduce the stack overhead of the program, thus saving memory and running time. Case studies have been done to illustrate the features of DESAL and to find scope for improvement.

Solid cancer tumors must recruit new blood vessels for growth and maintenance. Discovering drugs that block this tumor-induced development of new blood vessels (angiogenesis) is an important approach in cancer treatment. However, the complexity of angiogenesis and the difficulty in implementing and evaluating medical changes prevent the discovery of novel and effective new therapies. This paper presents a massively parallel computational search-based approach for the discovery of novel potential cancer treatments, using a high fidelity simulation of angiogenesis. Discovering new therapies is viewed as multi-objective combinatorial optimization over two competing objectives: minimizing the medical cost of the intervention while minimizing the oxygen provided to the cancer tumor by angiogenesis. Results show the effectiveness of the search process in finding simple interventions that are currently in use and more interestingly, discovering some new approaches that are counterintuitive yet effective. Distributed systems are becoming more prevalent as the demand for connectivity increases. Developers are faced with the challenge of creating software systems that meet these demands and adhere to good software practices. Technologies of today aid developers in this, but they may cause applications to suffer performance problems and require developers to abandon basic software concepts, such as modularization, performance, and maintainability. This work presents the Vitruvian framework that provides solutions to common distribution goals, and distributes applications using replication and transparency at varying stages of application development.

This thesis describes the design and implementation of a distributed robot architecture, Distributed Field Robot Architecture. The approach taken in this thesis is threefold. First, the distributed architecture builds on existing hybrid deliberative/reactive architectures used for individual robots rather than creating a distributed architecture that requires re-engineering of existing robots. Second, the distributed layer of the architecture incorporates concepts from artificial intelligence and software agents. Third, the architecture is designed around Suns Jini middleware layer, rather than creating a middleware layer from scratch or attempting to adapt a software agent architecture. This thesis makes three primary contributions, both theoretical and practical, to intelligent robotics. First, the thesis defines key characteristics of a distributed robot architecture. Second, this thesis describes, implements, and validates a distributed robot architecture. Third, the implementation with a team of mobile ground robots interacting with an external software "mission controller" agent in a complex, outdoor task is itself a contribution. The architecture is validated with three existence proofs. First, an example is presented to show the implementation of a basic sensor service. Second, a basic behavior is presented to validate the reactive portion of the architecture. Finally, an intelligent agent is presented to validate the deliberative layer of the architecture and describe the integration with the distributed layer.

Enterprise information systems are increasingly being developed as distributed information systems. Quality attributes of distributed information systems, as in the centralised case, should be evaluated as early and as accurately as possible in the software engineering process. In particular, software measures associated with quality attributes of such systems should consider the characteristics of modern distributed technologies. Early design decisions have a deep impact on the implementation of distributed enterprise information systems and thus, on the ultimate quality of the software as an operational entity. Due to the fact that the distributed-software engineering process affords software engineers a number of design alternatives, it is important to develop tools and guidelines that can be used to assess and compare design artefacts quantitatively. This dissertation makes a contribution to the field of Software Engineering by proposing and evaluating software design measures for distributed enterprise information systems. In previous research, measures developed for distributed software have been focused in code attributes, and thus, only provide feedback towards the end of the software engineering process. In contrast, this thesis proposes a number of specific design measures that provide quantitative information before the implementation. These measures capture attributes of the structure and behaviour of distributed information systems that are deemed important to assess their quality attributes, based on the analysis of the problem domain. The measures were evaluated theoretically and empirically as part of a well defined methodology. On the one hand, we have followed a formal framework based on the theory of measurement, in order to carry out the theoretical validation of the proposed measures. On the other hand, the suitability of the measures, to be used as indicators of quality attributes, was evaluated empirically with a robust statistical technique for exploratory research. The data sets analysed were gathered after running several experiments and replications with a distributed enterprise information system. The results of the empirical evaluation show that most of the proposed measures are correlated to the quality attributes of interest, and that most of these measures may be used, individually or in combination, for the estimation of these quality attributes-namely efficiency, reliability and maintainability. The design of a distributed information system is modelled as a combination of its structure, which reflects static characteristics, and its behaviour, which captures complementary dynamic aspects. The behavioural measures showed slightly better individual and combined results than the structural measures in the experimentation. This was in line with our expectations, since the measures were evaluated as indicators of non-functional quality attributes of the operational system. On the other hand, the structural measures provide useful feedback that is available earlier in the software engineering process. Finally, we developed a prototype application to collect the proposed measures automatically and examined typical real-world scenarios where the measures may be used to make design decisions as part of the software engineering process.

There are many variants of leader election algorithm in distributed networks. In this research, an agent based approach to leader election in distributed networks is investigated. Agents have shown to be useful in several ways. In the theoretical perspective, agents sometime help in reducing the message complexity of the system and sometimes help in lowering time complexity. In a more practical sense, agents perform operations independent of the processors, thereby lending a more flexible algorithm supporting different types of networks.

The Internet grows extremely fast in terms of number of users and traffic volume, as well as in the number of services that must be supported. This development results in new requirements on routers—the main building blocks of the Internet. Existing router designs suffer from architectural limitations that make it difficult to meet future requirements, and the purpose of this thesis is to explore new ways of building routers. We take the approach to investigate distributed and modular router designs, where routers are composed of multiple modules that can be mapped onto different processing elements. The modules communicate through open well-defined interfaces over an internal network. Our overall hypothesis is that such a combination of modularization and decentralization is a promising way to improve scalability, flexibility, and robustness of Internet routers—properties that will be critical for new generations of routers. Our research methodology is based on design, implementation, and experimental verification. The design work has two main results: an overall system design and a distributed router control plane. The system design consists of interfaces, protocols, and internal mechanisms for physically separation of different components of a router. The distributed control plane is a decomposition of control software into independent modules mapped onto multiple distributed processing elements. Our design is evaluated and verified through the implementation of a prototype system. The experimental part of the work deals with two key issues. First, transport mechanisms for communication of internal control information between processing elements are evaluated. In particular, we investigate the use of reliable multicast protocols in this context. Results regarding communication overhead as well as overall performance of routing table dissemination and installation are presented. The results show that even though there are certain costs associated with using reliable multicast, there are large performance gains to be made when the number of processing elements increases. Second, we present performance results of processing routing information in a distributed control plane. These results show that the processing time can be significantly reduced by distributing the workload over multiple processing elements. This indicates that considerable performance improvements can be made through the use of the distributed control plane architecture proposed in this thesis. / QC 20100616