Universität Stuttgart
Permanent URI for this communityhttps://elib.uni-stuttgart.de/handle/11682/1
Browse
Search Results
Item Open Access Visualization of uncorrelated point data(2008) Reina, Guido; Ertl, Thomas (Prof. Dr.)Sciences are the most common application context for computer-generated visualization. Researchers in these areas have to work with large datasets of many different types, but the one trait that is common to all is that in their raw form they exceed the cognitive abilities of human beings. Visualization not only aims at enabling users to quickly extract as much information as possible from datasets, but also at allowing the user to work at all with those that are too large and complex to be directly grasped by human cognition. In this work, the focus is on uncorrelated point data, or point clouds, which is sampled from real-world measurements or generated by computer simulations. Such datasets are gridless and exhibit no connectivity, and each point represents an entity of its own. To effectively work with such datasets, two main problems must be solved: on the one hand, a large number of complex primitives with potentially many attributes must be visualized, and on the other hand the interaction with the datasets must be designed in an intuitive way. This dissertation will present novel methods which allow the handling of large, point-based data sets of high dimensionality. The contribution for the rendering of hundreds of thousands of application-specific glyphs is a Graphics-Processing-Unit(GPU)-based solution that allows the exploration of datasets that exhibit a moderate number of dimensions, but an extremely large number of points. These approaches are proven to be working for molecular dynamics(MD) datasets as well as for 3D tensor fields. Factors critical for the performance of these algorithms are thoroughly analyzed, the main focus being on the fast rendering of these complex glyphs in high quality. To improve the visualization of datasets with many attributes and only a moderate number of points, methods for the interactive reduction of dimensionality and analysis of the influences of different dimensions as well as of different metrics will be presented. The rendering of the resulting data in 3D similarity space is also addressed. A GPU-based reduction of dimensions has been implemented that allows interactive tweaking of the reduction parameters while observing the results in real time. With the availability of a fast and responsive visualization, the missing component for a complete system is the human-computer interaction. The user must be able to navigate the information space and interact with a dataset, selecting or filtering the items that are of interest to him, inspecting the attributes of particular data points. Today, one must distinguish between the application context and the modality of different interaction approaches. Current research ranges from keyboard-and-mouse desktop interaction over different haptic interfaces (also including feedback) up to tracked interaction for virtual reality(VR) installations. In the context of this work, the problem of interacting with point-based datasets is tackled for two different situations. The first is the workstation-based analysis of clustering mechanics in thermodynamics simulations, the second a VR immersive navigation and interaction with point cloud datasets.Item Open Access Multi-field visualization on graphics processing units(2008) Botchen, Ralf Peter; Ertl, Thomas (Prof. Dr.)The generation of multi-field data has become commonplace in many scientific disciplines and application areas today. While researchers have produced numerous techniques for analyzing a single scalar, vector, or tensor field over the last years, finding approaches for exploring multi-field datasets still forms one of the significant challenges in visualization and analytics. One crucial aspect for the growing demand of multi-field visualization techniques is the fact that scientists need to explore the interaction of these fields to gain deeper understanding of underlying processes and relationships. This work addresses the challenge of illustrating multi-field data and presents new approaches of visualization techniques for a variety of application areas, with the aim to map these algorithms to graphics hardware architectures to achieve interactive visualization. In particular, the main contributions of this thesis contain multi-field flow visualization with one focus on integrating an additional flow uncertainty value, based on measurement simulation, into visualization. Therefore, texture based advection techniques are extended for the transport and display of the additional information. The second focus lies on the illustration of multiple fields as one combined characteristic set to minimize memory usage and allow further feature extraction from the new unique representation. New techniques are developed for multi-field volume rendering in the area of medical applications, with the primary challenge to intermix volumetric data that was acquired by different medical imaging modalities. The proposed solutions give implementation details for raycasting and slice-based rendering of multiple overlapping volumes. The third application area is video visualization. This domain is a typical representative for multi-field visualization, as it combines both, flow fields and multi-volume data for illustration. The goal of the introduced video visualization techniques is to extract dynamic or still objects in a scene, detect their individual actions and the relations among each other and to display this filtered information as a continuous stream of signatures for analysis. Another problematic issue in multi-field visualization is the size of the data, which is usually rather large. Yet, data transfer to and memory size on GPUs are two major bottlenecks. To address this issue, throughout the thesis techniques for data reduction by combination and data bricking for continuous streaming are discussed. Finally, multi-field data encoding and visualization techniques are presented that utilize the advantages of radial basis functions to minimize the data size.Item Open Access Particle tracing methods for visualization and computer graphics(2008) Schafhitzel, Tobias; Weiskopf, Daniel (Prof. Dr.)This thesis discusses the broad variety of particle tracing algorithms with focus on flow visualization. Starting with a general overview of the basics of visualization and computer graphics, mathematics, and fluid dynamics, a number of methods using particle tracing for flow visualization and computer graphics are proposed. The first part of this thesis considers mostly texture-based techniques that are implemented on the graphics processing unit (GPU) in order to provide an interactive dense representation of 3D flow fields. This part considers particle tracing methods that can be applied on general vector fields and includes texture based visualization in volumes as well as on surfaces. Furthermore, it is described how particle tracing can be used for extracting flow structures, like path surfaces, of the given vector field. The second part of this thesis considers particle tracing on derived vector fields for flow visualization. Therefore, first a feature extraction criterion is applied on a fluid flow field. In most cases this results in a scalar field serving as base for the particle tracing methods. Here, it is shown how higher order derivatives of scalar fields can be used to extract flow features like 1D vortex core lines or 2D shear sheets. The extracted structures are further processed in terms of feature tracking. The third part generalizes particle tracing for arbitrary applications in visualization and computer graphics. Here, the particles' path either might be defined by the perspective of the human eye or by a force field that influences the particles' motion by considering second order ordinary differential equations. All three parts clarify the importance of particle tracing methods for a wide range of applications in flow visualization and computer graphics by various examples. Furthermore, it is shown how the flexibility of this method strongly depends on the underlying vector field, and how those vector fields can be generated in order to solve problems that go beyond traditional particle tracing in fluid flow fields.Item Open Access A cross-layer framework for sensor networks(2008) Lachenmann, Andreas; Rothermel, Kurt (Prof. Dr.)Cross-layer interactions are often used in wireless sensor networks. They help to optimize energy consumption, deal with memory limitations, and consider the special properties of wireless communication. However, cross-layer interactions have the disadvantage of negatively affecting desirable properties of the software design like modularity and reusability. In the extreme, applications consist of a monolithic piece of code that is hard to develop and impossible to maintain. Therefore, this thesis investigates different approaches to address the negative side-effects of cross-layer interactions. In particular, it develops a framework that pursues three different strategies. First, it tries to preserve modularity and increase reusability by decoupling components that exchange data. This strategy is realized by TinyXXL, a programming abstraction for cross-layer data exchange. This part of the framework has been created based on an analysis of cross-layer interactions in existing applications. With some compile-time optimizations TinyXXL can reduce both energy and memory consumption compared to an application built from reusable components. Using Neidas, a novel neighborhood data sharing algorithm, it offers a comprehensive system for data exchange among the layers of a single node and with neighboring nodes. Second, the framework relaxes one of the constraints that often lead to cross-layer interactions and, thus, reduces the need to apply them. Specifically, it includes ViMem, a flash-based virtual memory system that helps to reduce memory limitations and tries to optimize the memory layout. Finally, the third strategy is to partially move energy concerns into the system software. For this purpose the framework includes Levels, an abstraction to specify optional functionality which allows to accurately meet a user-defined lifetime goal. If necessary, Levels deactivates functionality in order to reach that target lifetime. Furthermore, it includes a distributed algorithm that helps to provide a constant application quality over the total network lifetime.Item Open Access Supporting business process fragmentation while maintaining operational semantics : a BPEL perspective(2008) Khalaf, Rania; Leymann, Frank (Prof. Dr.)Globalization and the increase of competitive pressures created the need for agility in business processes, including the ability to outsource, offshore, or otherwise distribute its once-centralized business processes or parts thereof. While hampered thus far by limited infrastructure capabilities, the increase in bandwidth and connectivity and decrease in communication cost have removed these limits. An organization that aims for such fragmentation of its business processes needs to be able to separate the process into different parts. Today, this is a manual, design-time endeavor. For example, it may use the concept of subprocesses as parts to be outsourced. However, there is often no way to foresee, in advance, which parts of the process need to be cut-off. Thus, today’s technology for outsourcing is static and not dynamic at all. Therefore, there is a growing need for the ability to fragment one’s business processes in an agile manner, and be able to distribute and wire these fragments so that their combined execution recreates the function of the original process. Additionally, this needs to be done in a networked environment, which is where ‘Service Oriented Architecture’ plays a vital role. ‘Service Oriented Architecture’ (SOA) is a relatively new approach to software that natively deals with the very dynamic, distributed, loosely coupled, and heterogeneous features of today’s networked environment, offering application functions as networked services. Web services is one instantiation of an SOA, consisting of a modular, layered stack of XML standards and corresponding implementations that address the different aspects of this environment. The standard covering business processes for Web services is the Business Process Execution Language for Web Services (also known as ‘BPEL’). Relevant characteristics of BPEL are that it is SOA-centric, has a scope construct that groups activities providing them with common behavior such as fault and compensation handlers, and combines graph and calculus based approaches to process modeling. This thesis describes how to identify, create, and execute process fragments without loosing the operational semantics of the original process models. It does so within the framework of the Web services stack of standards, BPEL in particular. The contributions are a categorization of existing Web services aggregation techniques, a meta-model of Web services business process mechanisms using a graph-based formalism, a solution for the automatic and operational semantics-preserving decomposition of such processes, and an architecture and implementation for a corresponding build-time and runtime environment.Item Open Access Analyse und Optimierung der Softwareschichten von wissenschaftlichen Anwendungen für Metacomputing(2008) Keller, Rainer; Resch, Michael (Prof. Dr.-Ing.)Für parallele Anwendungen ist das Message Passing Interface (MPI) das Programmierparadigma der Wahl für Höchstleistungsrechner mit verteiltem Speicher. Mittels des Konzeptes des MetaComputings wiederum können verschiedenste Rechenressourcen mit PACX-MPI gekoppelt werden. Dies ist einerseits von Interesse, weil Problemgrößen gelöst werden sollen, die nicht auf nur einem System ausgeführt werden könnten, andererseits, weil gekoppelte Simulationen gerechnet werden, die auf bestimmten Rechnerarchitekturen ausgeführt werden sollen oder weil Systeme mit bestimmten Eigenschaften wie Visualisierungs- mit parallelen Rechenressourcen verbunden werden müssen. Diese Koppelung stellt für die verteilten Anwendungen eine Barriere dar, da Kommunikation zu nicht-lokalen Prozessen weitaus langsamer ist, als über das rechnerinterne Netzwerk. In dieser Arbeit werden Lösungen auf den Software-Ebenen ausgehend von der Netzwerkschicht, durch Verbesserungen innerhalb der verwendeten Middleware, bis hin zur Optimierung innerhalb der Anwendungsschicht erarbeitet. In Bezug auf die unterste Softwareschicht wird für die Middleware PACX-MPI eine allgemeine Bibliothek zur Netzwerkkommunikation auf Basis von User Datagram Protocol (UDP) entwickelt. Somit können Limitierungen des Transport Control Protocols (TCP) umgangen werden, vor allem in Verbindung mit Netzwerken mit hoher Latenz und großer Bandbreite, so genannte Long Fat Pipes. Die hier implementierte Bibliothek ist portabel programmiert und durch die Verwendung von Threads effizient. Dieses Protokoll erreicht gute Werte für die Bandbreite im Local Area Network (LAN), aber auch im Wide Area Network (WAN). Getestet wird dieses Protokoll zur Veranschaulichung mittels einer Verbindung zwischen Rechnern in Stuttgart und Canberra, Australien. Innerhalb der Middleware wird die Optimierung der kollektiven Kommunikationsroutinen behandelt und am Beispiel der Funktion PACX_Alltoall die Verbesserung anhand des IMB Benchmarks auf einem Metacomputer gezeigt. Zur Analyse der Kommunikationseigenschaften wird die Erweiterung einer Tracing-Bibliothek für PACX-MPI, sowie die Implementierung einer generischen Schnittstelle zur Messung der Kommunikationscharakteristik auf MPI-Schicht erläutert. Weiterhin wird eine allgemeine MPI-Testsuite vorgestellt, die beim Auffinden von Fehlern sowohl in PACX-MPI, als auch innerhalb der Open MPI Implementierung hilfreich war. Auf der obersten Softwareschicht werden Optimierungsmöglichkeiten für Anwendungen für MetaComputing aufgezeigt. Beispielhaft wird die Analyse des Kommunikationsmusters einer Anwendung aus dem Bereich der Bioinformatik gezeigt. Weiterhin wird die Implementierung des Cachings und Prefetchings von vielfach kommunizierten Daten mit räumlicher und zeitlicher Lokalität vorgestellt. Erst die Methodik des Cachings und Prefetchings erlaubt die Ausführung der Anwendung in einem Metacomputer und ist exemplarisch für eine Klasse von Algorithmen mit ähnlichem Kommunikationsmuster.Item Open Access Exploiting programmable graphics hardware for interactive visualization of 3D data fields(2008) Klein, Thomas; Ertl, Thomas (Prof. Dr.)Modern numerical simulation and data acquisition techniques create a multitude of different data fields. The interactive visualization of these large, three-dimensional, and often also time-dependent scalar, vector, and tensor fields plays an integral part in analyzing and understanding this data. Although basic visualization techniques vary significantly depending on the type of the respective data fields, there is one key feature that is dominating in today's visualization research. Driven by the need for interactive data inspection and exploration and by the extraordinary rate of increase of the computational power provided by modern graphics processing units, the attempt for consequent application of graphics hardware in all stages of the visualization pipeline has become a central theme in order to cope with the challenges of data set sizes growing at an ever increasing pace and advancing demands on the accuracy and complexity of visualizations. Contemporary graphics processing units now have reached a level of programmability roughly resembling their CPU counterparts. However, there are still important differences that strongly influence the design and implementation of GPU-based visualization algorithms. This thesis addresses the problem of how to efficiently exploit the programmability and parallel processing capabilities of modern graphics processors for interactive visualization of three-dimensional data fields of varying data complexity and abstraction level. In particular new methods and GPU-based solutions for high-quality volume ray casting, the reconstruction of polygonal isosurfaces, and the point-based visualization of symmetric, second-order tensor fields, such as obtained by diffusion tensor imaging or resulting from CFD simulations, by means of ellipsoidal glyphs are presented that by combining the mapping and rendering stage onto the GPU result in an improved visualization cycle. Furthermore, a new approach for the topological analysis of noisy vector fields is described. Although this work is focused on a number of specific visualization problems, it also intends to identify general design principles for GPU-based visualization algorithms that may prove useful in the context of topics not covered by this thesis.Item Open Access Integrated management framework for dynamic virtual organisations(2008) Wesner, Stefan; Resch, Michael (Prof. Dr.-Ing.)This thesis describes an Service Level Agreement based model for dynamic virtual organisations and a corresponding management framework for service providers making them able to fullfill such SLAs. The proposed framework is realised as a hierachical model starting from low level management close the hardware and network primitives necessary to realise the services up to the business relationship management layer. The concept is instantiated for the scenario of a High Performance Computing service provider.Item Open Access Towards meshless volume visualization(2008) Tejada-Gamero, Eduardo Jose; Ertl, Thomas (Prof. Dr.)In this thesis, novel meshless methods for surface and volume data reconstruction and rendering are proposed. Surface reconstruction from unorganized point sets is first addressed with projection operators. Specifically, a curvature-driven projection operator is presented which defines an approximate surface for a given point cloud based on a diffusion equation and on curvature estimation for point sets. Implicit formulations for surface approximation are also addressed. An implicit surface definition based on approximate moving least-squares approximation is introduced, which is able to provide high-order local approximations to the surface without requiring to solve systems of equations. Bilateral filters are introduced into this surface definition in order to better represent sharp features by robustly estimating normal vectors. An adaptive implicit formulation based on partition of unity and orthogonal polynomials is also proposed. This formulation addresses approximation and robustness issues presented by previous work on partition of unity implicits. To accelerate the rendering of these surface definitions, hardware-accelerated ray-casting of implicit surfaces and surfaces defined by projection operators is also discussed. The results obtained for surface approximation are then applied to volume data in order to extract surfaces that represent some feature in the volume. Regarding scalar data, a moving least-squares surface definition is proposed which is able to approximate iso-surfaces and surfaces located are regions with high gradient magnitude. The rendering of such surfaces is performed on graphics hardware to accelerate the computations. Visualization of vector fields is also addressed, specifically the interactive computation and rendering of streamsurfaces and of the novel path-surfaces. To that end, a hardware-accelerated streamlines and path-lines generation process is presented, which is able to produce a quasi regular sampling of the surface. This allows the use of known point-based surface rendering algorithms to interactively visualize the streamsurface or path-surface. Volume visualization is then addressed using meshless methods. These visualization methods are based on a meshless volume model extracted from the data. This model is obtained using the moving least-squares approximation method. In order to preserve details in the reconstruction of the volumetric data, bilateral filtering is used which, together with the use of orthogonal polynomials, provides a matrix-free detail-preserving reconstruction of the volume data. To further accelerate the computation of the function reconstruction, the use of approximate approximation is also explored in this context. To that end, an anisotropic iterated approximate moving least-squares approximation of the volume data is defined, which converges to an ellipsoidal-basis-functions interpolation of the data. Finally, volume deformation by means of moving least-squares is addressed and a closed formulation for nonlinear polynomial deformations is proposed. An implementation of the set of moving least-squares deformations on hardware graphics is also presented and used to interactively compute volume deformations by means of displacement maps.