The common manufacturing process of prosthetic sockets is usually a time- and labor consuming activity. This project’s purpose was to look for alternative manufacturing methods that could speed up the process and enhance the experience for the patient for example make some personal design or make the socket lighter. The main goal was to investigate which properties could be achieved by applying Electron Beam Melting as an alternative manufacturing process for prosthetic sockets by applying an earlier developed methodology. An investigation of earlier scientific works with the keywords (additive manufacturing, free form fabrication, orthopedic, prosthetic sockets and rapid manufacturing) was done as well as gathering knowledge how to operate and handle the machines necessary to carry out the project. An updated version of the methodology was developed where the design was verified using finite element analysis. With the updated version the methodology contained nine steps, which in short was as follows. First apprehend an inner socket from an orthopedic clinic with a pattern drawn up on it, the pattern is then transferred to a computer environment and manipulated to the desired shape and thickness. A compressive strength test, both virtual and experimental, was designed by a modified version of the ISO-10328 standard and the virtual design was verified before the socket was manufactured in the Electron Beam Melting machine. The manufactured socket was tested in the experimental set-up to verify the virtual one. The result was a personal designed socket of Ti6Al4V including the male pyramid for connection and a suspension system, which consisted of an inner socket and a one-way valve. It was concluded that Electron Beam Melting could be used as an alternative manufacturing process of prosthetic sockets.

A multidatabase system (MDBMS) is a facility that allows users to access data located in multiple autonomous database management systems (DBMSs) at different sites. To ensure global atomicity for multidatabase transactions, a reliable global atomic commitment protocol is a possible solution. In this protocol a centralized transaction manager (TM) receives global transactions, submits subtransactions to the appropriate sites via AGENTS. An AGENT is a component of MDBS that runs on each site; AGENTS after receiving subtransactions from the transaction manager perform the transaction and send the results back to TM. We have presented a unique proof-of-concept, a JAVA application for an Agent Based Transaction Manager that preserves global atomicity. It provides a user friendly interface through which reliable atomic commitment protocol for global transaction execution in multidatabase environment can be visualized. We demonstrated with three different test case scenarios how the protocol works. This is useful in further research in this area where atomicity of transactions can be verified for protocol correctness.

Technische Univ., Diplomarbeit--Wien, 2005.

The infraorbital nerve (ION) transmits sensory information from mechanoreceptors of the upper lip and vibrissae (whiskers) to the brain via the infraorbital foramen (IOF). Vibrissae are special sensory hairs used by mammals to explore their surroundings. Researchers have used the size of the IOF to infer vibrissa count, which in turn has been incorporated into phylogenetic and ecological interpretations of fossil taxa. However, these interpretations are based on untested assumptions linking IOF size, ION size, vibrissae, and ecology. The purpose of this dissertation is to determine the functional significance of IOF size and to apply the results to the fossil record. It is hypothesized that ecological differences among mammals affects maxillary mechanoreceptivity (touch sensitivity of the rostrum), and that IOF area may be used as a measure of maxillary mechanoreceptivity. Three questions are posed to appraise this hypothesis: (1) Does IOF area correlate with ION area and vibrissa count? (2) How do the IOFs of primates differ from those of other mammals? (3) How do diet, substrate preference, and activity pattern affect IOF size? IOF area, ION area, and vibrissa count were collected from cadaver of extant mammals as well as museum osteological specimens. Results indicate that: (1) IOF and ION areas show a strong positive correlation. Based on this finding, it is hypothesized that IOF area may be a good measure of maxillary mechanoreception. (2) Vibrissae count significantly correlates with IOF area. (3) Euarchontans have relatively smaller IOFs than most other mammals. (4) The IOFs of primates co-vary with diet, where frugivores have relatively larger IOFs than both insectivores and folivores. Infraorbital foramen areas of 14 adapoid, six omomyoid, and 15 plesiadapiform species were measured. Two questions were addressed: (1) Do the sampled fossils share a similar reduction in IOF area to extant primates? (2) Do extinct frugivores have larger IOFs than insectivores and folivores? Results show that, adapoids and omomyoids have relatively small IOFs similar to euarchontans, but plesiadapiforms retain larger IOFs, comparable to most non-euarchontan mammals. Dietary analyses indicate that both frugivorous adapoids and omomyoids have larger IOFs than both insectivorous and folivorous species. / text

Ingeniero Civil en Computación / Los servidores DNS son máquinas que resuelven consultas sobre nombres de dominio y se caracterizan por atender grandes cantidades de pequeñas peticiones que usualmente caben en una única unidad de comunicación en Internet llamada paquete . Para aumentar la cantidad de respuestas, el software encargado de esto explota las máquinas con múltiples procesadores, paralelizando las atenciones, lo cual ha mostrado no generar las ganancias esperadas. Para estudiar este problema se replicó la operación usando núcleos de Sistemas Operativos modernos e intentando leer concurrentemente desde un socket, identificando así los posibles puntos de falla: la implementación de read en la libc, el mecanismo de atención de las llamadas al sistema, o alguna porción de código ejecutado al recibir un paquete desde Internet. Los primeros dos posibles orígenes fueron descartados con pruebas directas mediante la comparación del rendimiento de read al leer desde distintas fuentes y determinando cómo se comporta la atención de las llamadas a sistema, mediante la creación de una de estas con fines de prueba. Así, el estudio se concentra en la tercera posible fuente del problema: el núcleo de Linux. Para estudiar el comportamiento de cómo es recibido un paquete, se investigó cómo fluye la información a través del stack de red desde que ésta arriba al dispositivo. Se descubrió que la información era encolada en estructuras de datos compartidas, requiriendo sincronización, e identificando, por lo tanto, un posible punto de falla. Para corroborarlo, se modificó un núcleo con el fin de determinar cómo la sincronización afectaba en la serialización de los accesos a un socket. Los resultados de las pruebas anteriores ejecutadas sobre el núcleo modificado arrojaron que el esquema de sincronización utilizado no permitía las lecturas concurrentes, por lo que se propuso cambiar el esquema de encolamiento de los paquetes en el socket, introduciendo estructuras que sí permitan la paralelización de read. Para simplificar la situación, el esquema de encolamiento de paquetes fue modelado en una implementación en C en espacio usuario, replicando estructuras y sincronización presentes en el núcleo. Sobre este modelo fue implementada una solución con múltiples colas de recepción de paquetes, creando colas por cada lector concurrente desde el lado de la aplicación. Finalmente, el modelo arrojó que esta solución permite efectivamente paralelizar los accesos, llegando a duplicar el throughput alcanzado actualmente por los sockets en determinadas configuraciones de threads.

Magíster en Ciencias, Mención Computación / Ingeniero Civil en Computación / La proliferación de sistemas con múltiples núcleos de procesamiento ha transformado la aplicación de técnicas de programación paralela en uno de los tópicos más estudiados en los últimos años, ello en pos de mejorar los rendimientos generales de cualquier operación. Una aplicación práctica de dicho enfoque es en el procesamiento de requerimientos DNS, los cuales han evidenciado un gran aumento de la mano del desarrollo de la Internet y la masificación de distintos tipos de dispositivos que demandan conectividad, y que por sus características son un buen candidato a un enfoque de procesamiento paralelo. Sin embargo, distintas investigaciones han revelado que la aplicación de técnicas multithreading per se no son una estrategia que garantice un escalamiento en los resultados. Distintas empresas de carácter global (como Google, Facebook y Toshiba) así como investigaciones locales (de la mano de NIC Chile) han reconocido la existencia de este problema, reiterando responsabilidades a los Internet Sockets provistos por el kernel de Linux, los cuales al ser expuestos a un consumo concurrente degradan su capacidad de consumo de datos y rendimiento general, aún cuando existe poder de cómputo excedente. Así estas empresas y otros trabajos de investigación han planteado varias hipótesis teóricamente razonables para explicar dicho comportamiento pero sin concretar las mismas en términos experimentales que permitan confirmarlas o desmentirlas. La presente investigación plantea un estudio experimental del caso de los Sockets UDP que combina técnicas de profiling y testing de distinto nivel, a fin de verificar las principales sospechas vigentes que den explicación al problema en cuestión, reunidas en 3 líneas de trabajo: Problemas de distribución de carga, degradación del rendimiento por mecanismos de bloqueo y problemas de manejo de caché y defectos de contención de recursos. En la misma línea, se ilustra el impacto de fenómeno de contención de recursos en un escenario concurrente y su repercusión en los canales de comunicación en el procesamiento de datos en arquitecturas modernas multiprocesador como la estudiada. Es fruto de este trabajo un veredicto por cada estudio que concluya explicando las características inherentes a los Internet Sockets que expliquen su mal desempeño, bajo los distintos análisis efectuados. Posteriormente, la investigación se traslada a estudiar la técnica denominada reuseport, un desarrollo de ingenieros de Google que plantea una solución al problema presentado y que promete mitigar el efecto negativo causado por el uso de multithreading en el contexto estudiado. Así también, se repasan aspectos como el rendimiento de este enfoque, y ciertas debilidades del mismo. Finalmente, es producto del presente trabajo el planteamiento e implementación de una solución propia al problema que, inspirado en el diseño de reuseport e implementado como un módulo del kernel, provee un mecanismo de distribución de paquetes que permite optar a mejoras de desempeño en el procesamiento de los mismos usando técnicas de procesamiento paralelo clásicas. Una solución que --en su configuración estándar-- provee un rendimiento competitivo a reuseport, pero que gracias a ser ampliamente configurable permite postular a mejores resultados que reuseport en contextos no uniformes de distribución de paquetes.

This thesis explores the possibilities of implementation of methods for data acquisition in wireless access point MikroTik while using virtualized OpenWRT system. First part of the paper contains short introduction into device’s features and extention of the features by virtualizing the operation system. Design and implementation of the scripts for average load of interfaces data acquisition is in the second part of the paper. The next part deals with the transmission of the gathered data and implementation of the proposed programs in C language. The last part of the paper covers the implementation of a complete program to collect, send and receive data with the option of parameters configuration, and automatic acquisition of the configuration.

The coding of network applications requires a mastery of communication APIs. One important set of communication APIs, socket APIs, are commonly used for coding C++ network applications. Socket APIs, unfortunately, are difficult to master, due to their procedural nature and inherent complexity. PISCES (Platform Independent Socket ClassES), is a C++ socket library that provides a simplified cross-platform interface to BSD-style socket APIs. This thesis analyzes the use of the PISCES library as a teaching tool. The PISCES library was refactored, then used as a basis for a series of introductory undergraduate assignments on network programming. Survey results show that the PISCES-based assignments were effective for introducing novice programmers to socket programming. The comparison of PISCES's series of assignments with those of NetCp and Toll's proposals, provide the inherent simplicity and safety involved in using PISCES's series assignments.

The need of smart devices that uses sensors have never been higher and by the year 2020 it will be over 50 billion devices connected to the internet. All these devices that uses a sensor and are connected to the internet are a part of something called Internet of Things. The purpose of this study has therefore been to implement a stripped IoT platform that doesn’t use any external libraries to lower the cost for minor companies that doesn’t need the more advanced and expensive platforms. After the implementation various stress test will be performed to see the performance of the platform. The study has been done through web- based sources and as a programming language Java has been used in the development environment NetBeans, the database has been made with MySQL workbench. The result of the study has been a platform that uses REST to post and get data from the database. The external library mysql- connector-java-5.1.45 was essential for a connection to the database and therefore had to be used. The result of the stress test was that the platform performed well and could handle at least 500 REST calls per second with a small increase in response time, but the standard deviation was considerably higher. The conclusion was that the platform performed stable at 50 – 250 calls per second and because of it being stripped several platforms could be used in a company to divide the work load between them resulting in a both stable and scalable solution. / Behovet för smarta enheter som använder sensorer har aldrig varit högre och det är trott att vid år 2020 kommer mer än 50 miljarder enheter vara uppkopplad mot internet. Alla dessa enheter med sensorer som är anslutna mot internet går under namnet Internet of Things. Syftet med denna studie har därför varit att skapa en avskalad IoT plattform som inte använder externa bibliotek för att hålla ned kostnaderna för de mindre företagen som inte behöver de mer avancerade och dyrare plattformarna. Efter att plattformen blivit implementerad skulle stresstester utförs för att avgöra hur bra den presterar. Studien har genomförts med hjälp av webbaserade källor och programmeringen av plattformen har utförts i programmeringsmiljön NetBeans i språket Java och databasen är skapad i MySQL workbench. Resultatet av studien har gett en plattform som använder REST för att skicka till och hämta data från databasen. Att göra implementeringen utan att använda externa bibliotek gick inte då biblioteket mysql-connector-java-5.1.45 var essentiellt för uppkoppling mot databasen. Stresstesterna gav att plattformen presterade stabilt och kunde hantera åtminstone 500 REST förfrågningar per sekund med endast en liten ökning i svarstiden, dock blev standardavvikelsen för svarstiden betydligt högre. Slutsatsen av studien blev att eftersom plattformen fungerar stabilt för 50 – 250 förfrågningar per sekund och då den tar upp lite processorkraft kan flera plattformar användas i ett företag för att då fördela arbetskraften emellan dem vilket resulterar i en lösning som både är skalbar samt stabil.

Com a heterogeneidade de tecnologias de comunicação sem fio presentes na borda de redes de acesso, serviços providos na Internet podem ser acessados de forma quasi ubíqua através de dispositivos móveis ou portáteis. O acesso a esses serviços, contudo, está associado a atrasos e rupturas frequentes na comunicação devido a razões inerentes à mobilidade do dispositivo, como: i) perda de sinal em locais onde há pouca ou nenhuma cobertura de acesso móvel; ii) erros no quadro de dados durante a transmissão e, consequentemente, perdas de pacotes, que podem ser ocasionados por interferência no sinal ou enfraquecimento deste pelo distanciamento do dispositivo em relação à Estação Base; iii) mudanças de endereços IP durante transmissões em andamento causadas pela migração do dispositivo entre diferentes redes. Como consequência, aplicações falham com a ruptura de comunicações orientadas a conexão. Tratar a mobilidade de forma transparente à aplicação é um dos desafios da Computação Móvel e Ubíqua que vem sendo pesquisado ao longo da última década. Soluções foram propostas para operarem desde a Camada de Enlace à Aplicação. Muitas delas, entretanto, exigem modificações na pilha de protocolos TCP/IP e adição de infraestrutura específica de rede no suporte à comunicação fim-a-fim. Além de elevar o custo das etapas de implantação e manutenção, estratégias intrusivas e dependentes de infraestrutura adicional podem não apresentar desempenho satisfatório. Nesse contexto, propomos tratar a mobilidade no nível da própria aplicação através de Sessões de Comunicação que não falham com atrasos e desconexões. Operando somente nos nós-fim e de modo transparente às Camadas adjacentes de Aplicação e Transporte, as sessões não requerem infraestrutura adicional para intermediar ou controlar a comunicação entre pares, tampouco modificações em protocolos legados da pilha TCP/IP. O conceito de Sessões Tolerantes a Rupturas é implementado através de uma API de propósito geral em sistemas Linux que estende a interface de Sockets. A API é, na prática, uma camada transparente sobre o Socket que provê Ciência de Mobilidade à aplicação através de mecanismos para: acompanhar a localização de nós ao longo da duração de uma sessão; detectar rupturas nas transmissões causadas pela mobilidade do nó ou de seu par remoto; suspender e retomar sessões de forma eficiente, segura e confiável. Experimentos conduzidos em ambientes emulados e reais com equipamentos de uso comercial mostram a eficiência das sessões. Além de introduzir baixa degradação na vazão fim-a-fim, rupturas na transmissão podem ser detectadas em microssegundos e sessões suspensas são reabertas em milissegundos. Com um desempenho superior a solução de mobilidade geral da Camada IP, as sessões não necessitam de adaptações de software em equipamentos de rede / Nowadays services available on the Internet can be accessed from mobile devices while they roam across heterogeneous wireless networks. Due to the inherent reasons of device mobility, however, the access to such services is frequently involved with delay and disruptions. The most common reasons are: i) losing radio signal at places where mobile access coverage area is not available; ii) frame error, losses, and fading on the radio signal when the mobile device moves away from the Base Station; iii) changes on the devices IP address over ongoing transmission, while the mobile node migrates among different wireless networks. As result, networked application fails with disruptions on TCP connections established in the mobile users path. Handling seamlessly mobility on the Internet is a technical challenge of the Mobile Computing Paradigm. It has been widely researched over the last decade. Several solutions have been proposed to work from the Link Layer to the Application Layer. Most of them, however, work intrusively and require modifications in the classical TCP/IP protocol stack, as well as rely on additional network infrastructure to support mobile end-to-end communication. Besides increasing the cost of deployment and maintenance, intrusive and infrastructure dependent strategies may not present suitable performance. In this sense, we devised an architecture to handle mobility at the Application level by means of communication sessions that do not fail with delay, disruption or disconnection. Such sessions work only at the end-systems in a such way that: are fully transparent to the adjacent layers of Transport and Application; do not require additional network infrastructure to forward and manage the communication between two mobile peers; and do not impose any modification on the legacy protocols from the TCP/IP stack. The concept of Disruption-Tolerant Sessions is implemented in Linux by means of a general purpose API extended from the Socket interface. Such API is a transparent layer placed on top of the Socket to provide mobility awareness to the Application Layer. To do so, session services are provided for: tracking mobile peers along the session duration; detecting disruptions over TCP connection caused by mobility of the local or remote peer; suspending and resuming sessions with efficiency, security and reliability. Experiments conducted in emulated and real systems (off-the-shelf hardware and open source software) showed the desired efficiency. Besides introducing little overhead on the goodput, disruptions are detected in a range of microseconds and suspended sessions are resumed in milliseconds. With performance greater than the general IP layer mobility solution, the proposed sessions do not require software adaptation in the core of the network infrastructure