Refine
Year of publication
Document Type
- Doctoral Thesis (6517) (remove)
Language
Keywords
- climate change (56)
- Klimawandel (55)
- Modellierung (36)
- Nanopartikel (28)
- machine learning (23)
- Fernerkundung (20)
- Deutschland (19)
- Spracherwerb (19)
- Synchronisation (19)
- remote sensing (18)
Institute
- Institut für Biochemie und Biologie (1045)
- Institut für Physik und Astronomie (781)
- Institut für Chemie (679)
- Institut für Geowissenschaften (507)
- Wirtschaftswissenschaften (407)
- Institut für Ernährungswissenschaft (279)
- Öffentliches Recht (255)
- Bürgerliches Recht (220)
- Historisches Institut (216)
- Institut für Informatik und Computational Science (206)
Nowadays, graph data models are employed, when relationships between entities have to be stored and are in the scope of queries. For each entity, this graph data model locally stores relationships to adjacent entities. Users employ graph queries to query and modify these entities and relationships. These graph queries employ graph patterns to lookup all subgraphs in the graph data that satisfy certain graph structures. These subgraphs are called graph pattern matches. However, this graph pattern matching is NP-complete for subgraph isomorphism. Thus, graph queries can suffer a long response time, when the number of entities and relationships in the graph data or the graph patterns increases.
One possibility to improve the graph query performance is to employ graph views that keep ready graph pattern matches for complex graph queries for later retrieval. However, these graph views must be maintained by means of an incremental graph pattern matching to keep them consistent with the graph data from which they are derived, when the graph data changes. This maintenance adds subgraphs that satisfy a graph pattern to the graph views and removes subgraphs that do not satisfy a graph pattern anymore from the graph views.
Current approaches for incremental graph pattern matching employ Rete networks. Rete networks are discrimination networks that enumerate and maintain all graph pattern matches of certain graph queries by employing a network of condition tests, which implement partial graph patterns that together constitute the overall graph query. Each condition test stores all subgraphs that satisfy the partial graph pattern. Thus, Rete networks suffer high memory consumptions, because they store a large number of partial graph pattern matches. But, especially these partial graph pattern matches enable Rete networks to update the stored graph pattern matches efficiently, because the network maintenance exploits the already stored partial graph pattern matches to find new graph pattern matches. However, other kinds of discrimination networks exist that can perform better in time and space than Rete networks. Currently, these other kinds of networks are not used for incremental graph pattern matching.
This thesis employs generalized discrimination networks for incremental graph pattern matching. These discrimination networks permit a generalized network structure of condition tests to enable users to steer the trade-off between memory consumption and execution time for the incremental graph pattern matching. For that purpose, this thesis contributes a modeling language for the effective definition of generalized discrimination networks. Furthermore, this thesis contributes an efficient and scalable incremental maintenance algorithm, which updates the (partial) graph pattern matches that are stored by each condition test. Moreover, this thesis provides a modeling evaluation, which shows that the proposed modeling language enables the effective modeling of generalized discrimination networks. Furthermore, this thesis provides a performance evaluation, which shows that a) the incremental maintenance algorithm scales, when the graph data becomes large, and b) the generalized discrimination network structures can outperform Rete network structures in time and space at the same time for incremental graph pattern matching.
Indonesien zählt zu den weltweit führenden Ländern bei der Nutzung von geothermischer Energie. Die geothermischen Energiequellen sind im Wesentlichen an den aktiven Vulkanismus gebunden, der durch die Prozesse an der indonesischen Subduktionszone verursacht wird. Darüber hinaus sind geotektonische Strukturen wie beispielsweise die Sumatra-Störung als verstärkende Faktoren für das geothermische Potenzial von Bedeutung. Bei der geophysikalischen Erkundung der indonesischen Geothermie-Ressourcen konzentrierte man sich bisher vor allem auf die Magnetotellurik. Passive Seismologie wurde dahingegen ausschließlich für die Überwachung von im Betrieb befindlichen Geothermie-Anlagen verwendet. Jüngste Untersuchungungen z.B. in Island und in den USA haben jedoch gezeigt, dass seismologische Verfahren bereits in der Erkundungsphase wichtige Informationen zu den physikalischen Eigenschaften, zum Spannungsfeld und zu möglichen Fluid- und Wärmetransportwegen liefern können. In der vorgelegten Doktorarbeit werden verschiedene moderne Methoden der passiven Seismologie verwendet, um beispielhaft ein neues, von der indonesischen Regierung für zukünftige geothermische Energiegewinnung ausgewiesenes Gebiet im nördlichen Teil Sumatras (Indonesien) zu erkunden. Die konkreten Ziele der Untersuchungen umfassten (1) die Ableitung von 3D Strukturmodellen der P- und S-Wellen Geschwindigkeiten (Parameter Vp und Vs), (2) die Bestimmung der Absorptionseigenschaften (Parameter Qp), und (3) die Kartierung und Charakterisierung von Störungssystemen auf der Grundlage der Seismizitätsverteilung und der Herdflächenlösungen. Für diese Zwecke habe ich zusammen mit Kollegen ein seismologisches Netzwerk in Tarutung (Sumatra) aufgebaut und über einen Zeitraum von 10 Monaten (Mai 2011 – Februar 2012) betrieben. Insgesamt wurden hierbei 42 Stationen (jeweils ausgestattet mit EDL-Datenlogger, 3-Komponenten, 1 Hz Seismometer) über eine Fläche von etwa 35 x 35 km verteilt. Mit dem Netzwerk wurden im gesamten Zeitraum 2568 lokale Erdbeben registriert. Die integrierte Betrachtung der Ergebnisse aus den verschiedenen Teilstudien (Tomographie, Erdbebenverteilung) erlaubt neue Einblicke in die generelle geologische Stukturierung sowie eine Eingrenzung von Bereichen mit einem erhöhten geothermischen Potenzial. Das tomographische Vp-Modell ermöglicht eine Bestimmung der Geometrie von Sedimentbecken entlang der Sumatra-Störung. Für die Geothermie besonders interessant ist der Bereich nordwestlich des Tarutung-Beckens. Die dort abgebildeten Anomalien (erhöhtes Vp/Vs, geringes Qp) habe ich als mögliche Aufstiegswege von warmen Fluiden interpretiert. Die scheinbar asymetrische Verteilung der Anomalien wird hierbei im Zusammenhang mit der Seismizitätsverteilung, der Geometrie der Beben-Bruchflächen, sowie struktur-geologischen Modellvorstellungen diskutiert. Damit werden wesentliche Informationen für die Planung einer zukünftigen geothermischen Anlage bereitgestellt.
This work analyzes the saving and consumption behavior of agents faced with the possibility of unemployment in a dynamic and stochastic life cycle model. The intertemporal optimization is based on Dynamic Programming with a backward recursion algorithm. The implemented uncertainty is not based on income shocks as it is done in traditional life cycle models but uses Markov probabilities where the probability for the next employment status of the agent depends on the current status. The utility function used is a CRRA function (constant relative risk aversion), combined with a CES function (constant elasticity of substitution) and has several consumption goods, a subsistence level, money and a bequest function.
In the context of cosmological structure formation sheets, filaments and eventually halos form due to gravitational instabilities. It is noteworthy, that at all times, the majority of the baryons in the universe does not reside in the dense halos but in the filaments and the sheets of the intergalactic medium. While at higher redshifts of z > 2, these baryons can be detected via the absorption of light (originating from more distant sources) by neutral hydrogen at temperatures of T ~ 10^4 K (the Lyman-alpha forest), at lower redshifts only about 20 % can be found in this state. The remain (about 50 to 70 % of the total baryons mass) is unaccounted for by observational means. Numerical simulations predict that these missing baryons could reside in the filaments and sheets of the cosmic web at high temperatures of T = 10^4.5 - 10^7 K, but only at low to intermediate densities, and constitutes the warm-hot intergalactic medium (WHIM). The high temperatures of the WHIM are caused by the formation of shocks and the subsequent shock-heating of the gas. This results in a high degree of ionization and renders the reliable detection of the WHIM a challenging task. Recent high-resolution hydrodynamical simulations indicate that, at redshifts of z ~ 2, filaments are able to provide very massive galaxies with a significant amount of cool gas at temperatures of T ~ 10^4 K. This could have an important impact on the star-formation in those galaxies. It is therefore of principle importance to investigate the particular hydro- and thermodynamical conditions of these large filament structures. Density and temperature profiles, and velocity fields, are expected to leave their special imprint on spectroscopic observations. A potential multiphase structure may act as tracer in observational studies of the WHIM. In the context of cold streams, it is important to explore the processes, which regulate the amount of gas transported by the streams. This includes the time evolution of filaments, as well as possible quenching mechanisms. In this context, the halo mass range in which cold stream accretion occurs is of particular interest. In order to address these questions, we perform particular hydrodynamical simulations of very high resolution, and investigate the formation and evolution of prototype structures representing the typical filaments and sheets of the WHIM. We start with a comprehensive study of the one-dimensional collapse of a sinusoidal density perturbation (pancake formation) and examine the influence of radiative cooling, heating due to an UV background, thermal conduction, and the effect of small-scale perturbations given by the cosmological power spectrum. We use a set of simulations, parametrized by the wave length of the initial perturbation L. For L ~ 2 Mpc/h the collapse leads to shock-confined structures. As a result of radiative cooling and of heating due to an UV background, a relatively cold and dense core forms. With increasing L the core becomes denser and more concentrated. Thermal conduction enhances this trend and may lead to an evaporation of the core at very large L ~ 30 Mpc/h. When extending our simulations into three dimensions, instead of a pancake structure, we obtain a configuration consisting of well-defined sheets, filaments, and a gaseous halo. For L > 4 Mpc/h filaments form, which are fully confined by an accretion shock. As with the one-dimensional pancakes, they exhibit an isothermal core. Thus, our results confirm a multiphase structure, which may generate particular spectral tracers. We find that, after its formation, the core becomes shielded against further infall of gas onto the filament, and its mass content decreases with time. In the vicinity of the halo, the filament's core can be attributed to the cold streams found in other studies. We show, that the basic structure of these cold streams exists from the very beginning of the collapse process. Further on, the cross section of the streams is constricted by the outwards moving accretion shock of the halo. Thermal conduction leads to a complete evaporation of the cold stream for L > 6 Mpc/h. This corresponds to halos with a total mass higher than M_halo = 10^13 M_sun, and predicts that in more massive halos star-formation can not be sustained by cold streams. Far away from the gaseous halo, the temperature gradients in the filament are not sufficiently strong for thermal conduction to be effective.
With the emergence of the Internet of things (IoT), plenty of battery-powered and energy-harvesting devices are being deployed to fulfill sensing and actuation tasks in a variety of application areas, such as smart homes, precision agriculture, smart cities, and industrial automation. In this context, a critical issue is that of denial-of-sleep attacks. Such attacks temporarily or permanently deprive battery-powered, energy-harvesting, or otherwise energy-constrained devices of entering energy-saving sleep modes, thereby draining their charge. At the very least, a successful denial-of-sleep attack causes a long outage of the victim device. Moreover, to put battery-powered devices back into operation, their batteries have to be replaced. This is tedious and may even be infeasible, e.g., if a battery-powered device is deployed at an inaccessible location. While the research community came up with numerous defenses against denial-of-sleep attacks, most present-day IoT protocols include no denial-of-sleep defenses at all, presumably due to a lack of awareness and unsolved integration problems. After all, despite there are many denial-of-sleep defenses, effective defenses against certain kinds of denial-of-sleep attacks are yet to be found.
The overall contribution of this dissertation is to propose a denial-of-sleep-resilient medium access control (MAC) layer for IoT devices that communicate over IEEE 802.15.4 links. Internally, our MAC layer comprises two main components. The first main component is a denial-of-sleep-resilient protocol for establishing session keys among neighboring IEEE 802.15.4 nodes. The established session keys serve the dual purpose of implementing (i) basic wireless security and (ii) complementary denial-of-sleep defenses that belong to the second main component. The second main component is a denial-of-sleep-resilient MAC protocol. Notably, this MAC protocol not only incorporates novel denial-of-sleep defenses, but also state-of-the-art mechanisms for achieving low energy consumption, high throughput, and high delivery ratios. Altogether, our MAC layer resists, or at least greatly mitigates, all denial-of-sleep attacks against it we are aware of. Furthermore, our MAC layer is self-contained and thus can act as a drop-in replacement for IEEE 802.15.4-compliant MAC layers. In fact, we implemented our MAC layer in the Contiki-NG operating system, where it seamlessly integrates into an existing protocol stack.
Dark matter, DM, has not yet been directly observed, but it has a very solid theoretical basis. There are observations that provide indirect evidence, like galactic rotation curves that show that the galaxies are rotating too fast to keep their constituent parts, and galaxy clusters that bends the light coming from behind-lying galaxies more than expected with respect to the mass that can be calculated from what can be visibly seen. These observations, among many others, can be explained with theories that include DM. The missing piece is to detect something that can exclusively be explained by DM. Direct observation in a particle accelerator is one way and indirect detection using telescopes is another. This thesis is focused on the latter method.
The Very Energetic Radiation Imaging Telescope Array System, V ERITAS, is a telescope array that detects Cherenkov radiation. Theory predicts that DM particles annihilate into, e.g., a γγ pair and create a distinctive energy spectrum when detected by such telescopes, e.i., a monoenergetic line at the same energy as the particle mass. This so called ”smoking-gun” signature is sought with a sliding window line search within the sub-range ∼ 0.3 − 10 TeV of the VERITAS energy range, ∼ 0.01 − 30 TeV.
Standard analysis within the VERITAS collaboration uses Hillas analysis and look-up tables, acquired by analysing particle simulations, to calculate the energy of the particle causing the Cherenkov shower. In this thesis, an improved analysis method has been used. Modelling each shower as a 3Dgaussian should increase the energy recreation quality. Five dwarf spheroidal galaxies were chosen as targets with a total of ∼ 224 hours of data. The targets were analysed individually and stacked. Particle simulations were based on two simulation packages, CARE and GrISU.
Improvements have been made to the energy resolution and bias correction, up to a few percent each, in comparison to standard analysis. Nevertheless, no line with a relevant significance has been detected. The most promising line is at an energy of ∼ 422 GeV with an upper limit cross section of 8.10 · 10^−24 cm^3 s^−1 and a significance of ∼ 2.73 σ, before trials correction and ∼ 1.56 σ after. Upper limit cross sections have also been calculated for the γγ annihilation process and four other outcomes. The limits are in line with current limits using other methods, from ∼ 8.56 · 10^−26 − 6.61 · 10^−23 cm^3s^−1. Future larger telescope arrays, like the upcoming Cherenkov Telescope Array, CTA, will provide better results with the help of this analysis method.