Refine
Has Fulltext
- yes (165)
Is part of the Bibliography
- yes (165)
Year of publication
Document Type
- Doctoral Thesis (165) (remove)
Keywords
- Leistungsbewertung (29)
- Netzwerk (10)
- Robotik (10)
- Quality of Experience (9)
- Modellierung (8)
- Simulation (8)
- Autonomer Roboter (7)
- Mobiler Roboter (7)
- Optimierung (7)
- Performance Evaluation (7)
Institute
- Institut für Informatik (165) (remove)
Schriftenreihe
Sonstige beteiligte Institutionen
- Deutsches Zentrum für Luft- und Raumfahrt (DLR), Institut für Raumfahrtsysteme (2)
- Zentrum für Telematik e.V. (2)
- Airbus Defence and Space GmbH (1)
- Beuth Hochschule für Technik Berlin (1)
- DLR (1)
- Deutsches Zentrum für Luft- und Raumfahrt e.V. (1)
- European Space Agency (1)
- Hochschule Wismar (1)
- Technische Hochschule Nürnberg Georg Simon Ohm (1)
- University of Applied Sciences and Arts Western Switzerland, Fribourg (1)
The importance of proactive and timely prediction of critical events is steadily increasing, whether in the manufacturing industry or in private life. In the past, machines in the manufacturing industry were often maintained based on a regular schedule or threshold violations, which is no longer competitive as it causes unnecessary costs and downtime. In contrast, the predictions of critical events in everyday life are often much more concealed and hardly noticeable to the private individual, unless the critical event occurs. For instance, our electricity provider has to ensure that we, as end users, are always supplied with sufficient electricity, or our favorite streaming service has to guarantee that we can watch our favorite series without interruptions. For this purpose, they have to constantly analyze what the current situation is, how it will develop in the near future, and how they have to react in order to cope with future conditions without causing power outages or video stalling.
In order to analyze the performance of a system, monitoring mechanisms are often integrated to observe characteristics that describe the workload and the state of the system and its environment. Reactive systems typically employ thresholds, utility functions, or models to determine the current state of the system. However, such reactive systems cannot proactively estimate future events, but only as they occur. In the case of critical events, reactive determination of the current system state is futile, whereas a proactive system could have predicted this event in advance and enabled timely countermeasures. To achieve proactivity, the system requires estimates of future system states. Given the gap between design time and runtime, it is typically not possible to use expert knowledge to a priori model all situations a system might encounter at runtime. Therefore, prediction methods must be integrated into the system. Depending on the available monitoring data and the complexity of the prediction task, either time series forecasting in combination with thresholding or more sophisticated machine and deep learning models have to be trained.
Although numerous forecasting methods have been proposed in the literature, these methods have their advantages and disadvantages depending on the characteristics of the time series under consideration. Therefore, expert knowledge is required to decide which forecasting method to choose. However, since the time series observed at runtime cannot be known at design time, such expert knowledge cannot be implemented in the system. In addition to selecting an appropriate forecasting method, several time series preprocessing steps are required to achieve satisfactory forecasting accuracy. In the literature, this preprocessing is often done manually, which is not practical for autonomous computing systems, such as Self-Aware Computing Systems. Several approaches have also been presented in the literature for predicting critical events based on multivariate monitoring data using machine and deep learning. However, these approaches are typically highly domain-specific, such as financial failures, bearing failures, or product failures. Therefore, they require in-depth expert knowledge. For this reason, these approaches cannot be fully automated and are not transferable to other use cases. Thus, the literature lacks generalizable end-to-end workflows for modeling, detecting, and predicting failures that require only little expert knowledge.
To overcome these shortcomings, this thesis presents a system model for meta-self-aware prediction of critical events based on the LRA-M loop of Self-Aware Computing Systems. Building upon this system model, this thesis provides six further contributions to critical event prediction. While the first two contributions address critical event prediction based on univariate data via time series forecasting, the three subsequent contributions address critical event prediction for multivariate monitoring data using machine and deep learning algorithms. Finally, the last contribution addresses the update procedure of the system model. Specifically, the seven main contributions of this thesis can be summarized as follows:
First, we present a system model for meta self-aware prediction of critical events. To handle both univariate and multivariate monitoring data, it offers univariate time series forecasting for use cases where a single observed variable is representative of the state of the system, and machine learning algorithms combined with various preprocessing techniques for use cases where a large number of variables are observed to characterize the system’s state. However, the two different modeling alternatives are not disjoint, as univariate time series forecasts can also be included to estimate future monitoring data as additional input to the machine learning models. Finally, a feedback loop is incorporated to monitor the achieved prediction quality and trigger model updates.
We propose a novel hybrid time series forecasting method for univariate, seasonal time series, called Telescope. To this end, Telescope automatically preprocesses the time series, performs a kind of divide-and-conquer technique to split the time series into multiple components, and derives additional categorical information. It then forecasts the components and categorical information separately using a specific state-of-the-art method for each component. Finally, Telescope recombines the individual predictions. As Telescope performs both preprocessing and forecasting automatically, it represents a complete end-to-end approach to univariate seasonal time series forecasting. Experimental results show that Telescope achieves enhanced forecast accuracy, more reliable forecasts, and a substantial speedup. Furthermore, we apply Telescope to the scenario of predicting critical events for virtual machine auto-scaling. Here, results show that Telescope considerably reduces the average response time and significantly reduces the number of service level objective violations.
For the automatic selection of a suitable forecasting method, we introduce two frameworks for recommending forecasting methods. The first framework extracts various time series characteristics to learn the relationship between them and forecast accuracy. In contrast, the other framework divides the historical observations into internal training and validation parts to estimate the most appropriate forecasting method. Moreover, this framework also includes time series preprocessing steps. Comparisons between the proposed forecasting method recommendation frameworks and the individual state-of-the-art forecasting methods and the state-of-the-art forecasting method recommendation approach show that the proposed frameworks considerably improve the forecast accuracy.
With regard to multivariate monitoring data, we first present an end-to-end workflow to detect critical events in technical systems in the form of anomalous machine states. The end-to-end design includes raw data processing, phase segmentation, data resampling, feature extraction, and machine tool anomaly detection. In addition, the workflow does not rely on profound domain knowledge or specific monitoring variables, but merely assumes standard machine monitoring data. We evaluate the end-to-end workflow using data from a real CNC machine. The results indicate that conventional frequency analysis does not detect the critical machine conditions well, while our workflow detects the critical events very well with an F1-score of almost 91%.
To predict critical events rather than merely detecting them, we compare different modeling alternatives for critical event prediction in the use case of time-to-failure prediction of hard disk drives. Given that failure records are typically significantly less frequent than instances representing the normal state, we employ different oversampling strategies. Next, we compare the prediction quality of binary class modeling with downscaled multi-class modeling. Furthermore, we integrate univariate time series forecasting into the feature generation process to estimate future monitoring data. Finally, we model the time-to-failure using not only classification models but also regression models. The results suggest that multi-class modeling provides the overall best prediction quality with respect to practical requirements. In addition, we prove that forecasting the features of the prediction model significantly improves the critical event prediction quality.
We propose an end-to-end workflow for predicting critical events of industrial machines. Again, this approach does not rely on expert knowledge except for the definition of monitoring data, and therefore represents a generalizable workflow for predicting critical events of industrial machines. The workflow includes feature extraction, feature handling, target class mapping, and model learning with integrated hyperparameter tuning via a grid-search technique. Drawing on the result of the previous contribution, the workflow models the time-to-failure prediction in terms of multiple classes, where we compare different labeling strategies for multi-class classification. The evaluation using real-world production data of an industrial press demonstrates that the workflow is capable of predicting six different time-to-failure windows with a macro F1-score of 90%. When scaling the time-to-failure classes down to a binary prediction of critical events, the F1-score increases to above 98%.
Finally, we present four update triggers to assess when critical event prediction models should be re-trained during on-line application. Such re-training is required, for instance, due to concept drift. The update triggers introduced in this thesis take into account the elapsed time since the last update, the prediction quality achieved on the current test data, and the prediction quality achieved on the preceding test data. We compare the different update strategies with each other and with the static baseline model. The results demonstrate the necessity of model updates during on-line application and suggest that the update triggers that consider both the prediction quality of the current and preceding test data achieve the best trade-off between prediction quality and number of updates required.
We are convinced that the contributions of this thesis constitute significant impulses for the academic research community as well as for practitioners. First of all, to the best of our knowledge, we are the first to propose a fully automated, end-to-end, hybrid, component-based forecasting method for seasonal time series that also includes time series preprocessing. Due to the combination of reliably high forecast accuracy and reliably low time-to-result, it offers many new opportunities in applications requiring accurate forecasts within a fixed time period in order to take timely countermeasures. In addition, the promising results of the forecasting method recommendation systems provide new opportunities to enhance forecasting performance for all types of time series, not just seasonal ones. Furthermore, we are the first to expose the deficiencies of the prior state-of-the-art forecasting method recommendation system.
Concerning the contributions to critical event prediction based on multivariate monitoring data, we have already collaborated closely with industrial partners, which supports the practical relevance of the contributions of this thesis. The automated end-to-end design of the proposed workflows that do not demand profound domain or expert knowledge represents a milestone in bridging the gap between academic theory and industrial application. Finally, the workflow for predicting critical events in industrial machines is currently being operationalized in a real production system, underscoring the practical impact of this thesis.
Internet applications are becoming more and more flexible to support diverge user demands and network conditions. This is reflected by technical concepts, which provide new adaptation mechanisms to allow fine grained adjustment of the application quality and the corresponding bandwidth requirements. For the case of video streaming, the scalable video codec H.264/SVC allows the flexible adaptation of frame rate, video resolution and image quality with respect to the available network resources. In order to guarantee a good user-perceived quality (Quality of Experience, QoE) it is necessary to adjust and optimize the video quality accurately. But not only have the applications of the current Internet changed. Within network and transport, new technologies evolved during the last years providing a more flexible and efficient usage of data transport and network resources. One of the most promising technologies is Network Virtualization (NV) which is seen as an enabler to overcome the ossification of the Internet stack. It provides means to simultaneously operate multiple logical networks which allow for example application-specific addressing, naming and routing, or their individual resource management. New transport mechanisms like multipath transmission on the network and transport layer aim at an efficient usage of available transport resources. However, the simultaneous transmission of data via heterogeneous transport paths and communication technologies inevitably introduces packet reordering. Additional mechanisms and buffers are required to restore the correct packet order and thus to prevent a disturbance of the data transport. A proper buffer dimensioning as well as the classification of the impact of varying path characteristics like bandwidth and delay require appropriate evaluation methods. Additionally, for a path selection mechanism real time evaluation mechanisms are needed. A better application-network interaction and the corresponding exchange of information enable an efficient adaptation of the application to the network conditions and vice versa. This PhD thesis analyzes a video streaming architecture utilizing multipath transmission and scalable video coding and develops the following optimization possibilities and results: Analysis and dimensioning methods for multipath transmission, quantification of the adaptation possibilities to the current network conditions with respect to the QoE for H.264/SVC, and evaluation and optimization of a future video streaming architecture, which allows a better interaction of application and network.
Graphs provide a key means to model relationships between entities.
They consist of vertices representing the entities,
and edges representing relationships between pairs of entities.
To make people conceive the structure of a graph,
it is almost inevitable to visualize the graph.
We call such a visualization a graph drawing.
Moreover, we have a straight-line graph drawing
if each vertex is represented as a point
(or a small geometric object, e.g., a rectangle)
and each edge is represented as a line segment between its two vertices.
A polyline is a very simple straight-line graph drawing,
where the vertices form a sequence according to which the vertices are connected by edges.
An example of a polyline in practice is a GPS trajectory.
The underlying road network, in turn, can be modeled as a graph.
This book addresses problems that arise
when working with straight-line graph drawings and polylines.
In particular, we study algorithms
for recognizing certain graphs representable with line segments,
for generating straight-line graph drawings,
and for abstracting polylines.
In the first part, we first examine,
how and in which time we can decide
whether a given graph is a stick graph,
that is, whether its vertices can be represented as
vertical and horizontal line segments on a diagonal line,
which intersect if and only if there is an edge between them.
We then consider the visual complexity of graphs.
Specifically, we investigate, for certain classes of graphs,
how many line segments are necessary for any straight-line graph drawing,
and whether three (or more) different slopes of the line segments
are sufficient to draw all edges.
Last, we study the question,
how to assign (ordered) colors to the vertices of a graph
with both directed and undirected edges
such that no neighboring vertices get the same color
and colors are ascending along directed edges.
Here, the special property of the considered graph is
that the vertices can be represented as intervals
that overlap if and only if there is an edge between them.
The latter problem is motivated by an application
in automated drawing of cable plans with vertical and horizontal line segments,
which we cover in the second part.
We describe an algorithm that
gets the abstract description of a cable plan as input,
and generates a drawing that takes into account
the special properties of these cable plans,
like plugs and groups of wires.
We then experimentally evaluate the quality of the resulting drawings.
In the third part, we study the problem of abstracting (or simplifying)
a single polyline and a bundle of polylines.
In this problem, the objective is to remove as many vertices as possible from the given polyline(s)
while keeping each resulting polyline sufficiently similar to its original course
(according to a given similarity measure).
This thesis is devoted to Bernoulli Stochastics, which was initiated by Jakob Bernoulli more than 300 years ago by his master piece 'Ars conjectandi', which can be translated as 'Science of Prediction'. Thus, Jakob Bernoulli's Stochastics focus on prediction in contrast to the later emerging disciplines probability theory, statistics and mathematical statistics. Only recently Jakob Bernoulli's focus was taken up von Collani, who developed a unified theory of uncertainty aiming at making reliable and accurate predictions. In this thesis, teaching material as well as a virtual classroom are developed for fostering ideas and techniques initiated by Jakob Bernoulli and elaborated by Elart von Collani. The thesis is part of an extensively construed project called 'Stochastikon' aiming at introducing Bernoulli Stochastics as a unified science of prediction and measurement under uncertainty. This ambitious aim shall be reached by the development of an internet-based comprehensive system offering the science of Bernoulli Stochastics on any level of application. So far it is planned that the 'Stochastikon' system (http://www.stochastikon.com/) will consist of five subsystems. Two of them are developed and introduced in this thesis. The first one is the e-learning programme 'Stochastikon Magister' and the second one 'Stochastikon Graphics' that provides the entire Stochastikon system with graphical illustrations. E-learning is the outcome of merging education and internet techniques. E-learning is characterized by the facts that teaching and learning are independent of place and time and of the availability of specially trained teachers. Knowledge offering as well as knowledge transferring are realized by using modern information technologies. Nowadays more and more e-learning environments are based on the internet as the primary tool for communication and presentation. E-learning presentation tools are for instance text-files, pictures, graphics, audio and videos, which can be networked with each other. There could be no limit as to the access to teaching contents. Moreover, the students can adapt the speed of learning to their individual abilities. E-learning is particularly appropriate for newly arising scientific and technical disciplines, which generally cannot be presented by traditional learning methods sufficiently well, because neither trained teachers nor textbooks are available. The first part of this dissertation introduces the state of the art of e-learning in statistics, since statistics and Bernoulli Stochastics are both based on probability theory and exhibit many similar features. Since Stochastikon Magister is the first e-learning programme for Bernoulli Stochastics, the educational statistics systems is selected for the purpose of comparison and evaluation. This makes sense as both disciplines are an attempt to handle uncertainty and use methods that often can be directly compared. The second part of this dissertation is devoted to Bernoulli Stochastics. This part aims at outlining the content of two courses, which have been developed for the anticipated e-learning programme Stochastikon Magister in order to show the difficulties in teaching, understanding and applying Bernoulli Stochastics. The third part discusses the realization of the e-learning programme Stochastikon Magister, its design and implementation, which aims at offering a systematic learning of principles and techniques developed in Bernoulli Stochastics. The resulting e-learning programme differs from the commonly developed e-learning programmes as it is an attempt to provide a virtual classroom that simulates all the functions of real classroom teaching. This is in general not necessary, since most of the e-learning programmes aim at supporting existing classroom teaching. The forth part presents two empirical evaluations of Stochastikon Magister. The evaluations are performed by means of comparisons between traditional classroom learning in statistics and e-learning of Bernoulli Stochastics. The aim is to assess the usability and learnability of Stochastikon Magister. Finally, the fifth part of this dissertation is added as an appendix. It refers to Stochastikon Graphics, the fifth component of the entire Stochastikon system. Stochastikon Graphics provides the other components with graphical representations of concepts, procedures and results obtained or used in the framework of Bernoulli Stochastics. The primary aim of this thesis is the development of an appropriate software for the anticipated e-learning environment meant for Bernoulli Stochastics, while the preparation of the necessary teaching material constitutes only a secondary aim used for demonstrating the functionality of the e-learning platform and the scientific novelty of Bernoulli Stochastics. To this end, a first version of two teaching courses are developed, implemented and offered on-line in order to collect practical experiences. The two courses, which were developed as part of this projects are submitted as a supplement to this dissertation. For the time being the first experience with the e-learning programme Stochastikon Magister has been made. Students of different faculties of the University of Würzburg, as well as researchers and engineers, who are involved in the Stochastikon project have obtained access to Stochastikon Magister via internet. They have registered for Stochastikon Magister and participated in the course programme. This thesis reports on two assessments of these first experiences and the results will lead to further improvements with respect to content and organization of Stochastikon Magister.
This work is composed of three main parts: remote control of mobile systems via Internet, ad-hoc networks of mobile robots, and remote control of mobile robots via 3G telecommunication technologies. The first part gives a detailed state of the art and a discussion of the problems to be solved in order to teleoperate mobile robots via the Internet. The focus of the application to be realized is set on a distributed tele-laboratory with remote experiments on mobile robots which can be accessed world-wide via the Internet. Therefore, analyses of the communication link are used in order to realize a robust system. The developed and implemented architecture of this distributed tele-laboratory allows for a smooth access also with a variable or low link quality. The second part covers the application of ad-hoc networks for mobile robots. The networking of mobile robots via mobile ad-hoc networks is a very promising approach to realize integrated telematic systems without relying on preexisting communication infrastructure. Relevant civilian application scenarios are for example in the area of search and rescue operations where first responders are supported by multi-robot systems. Here, mobile robots, humans, and also existing stationary sensors can be connected very fast and efficient. Therefore, this work investigates and analyses the performance of different ad-hoc routing protocols for IEEE 802.11 based wireless networks in relevant scenarios. The analysis of the different protocols allows for an optimization of the parameter settings in order to use these ad-hoc routing protocols for mobile robot teleoperation. Also guidelines for the realization of such telematics systems are given. Also traffic shaping mechanisms of application layer are presented which allow for a more efficient use of the communication link. An additional application scenario, the integration of a small size helicopter into an IP based ad-hoc network, is presented. The teleoperation of mobile robots via 3G telecommunication technologies is addressed in the third part of this work. The high availability, high mobility, and the high bandwidth provide a very interesting opportunity to realize scenarios for the teleoperation of mobile robots or industrial remote maintenance. This work analyses important parameters of the UMTS communication link and investigates also the characteristics for different data streams. These analyses are used to give guidelines which are necessary for the realization of or industrial remote maintenance or mobile robot teleoperation scenarios. All the results and guidelines for the design of telematic systems in this work were derived from analyses and experiments with real hardware.
This dissertation presents controller design methodologies for a formation of cooperative mobile robots to perform trajectory tracking and convoy protection tasks. Two major problems related to multi-agent formation control are addressed, namely the time-delay and optimality problems. For the task of trajectory tracking, a leader-follower based system structure is adopted for the controller design, where the selection criteria for controller parameters are derived through analyses of characteristic polynomials. The resulting parameters ensure the stability of the system and overcome the steady-state error as well as the oscillation behavior under time-delay effect. In the convoy protection scenario, a decentralized coordination strategy for balanced deployment of mobile robots is first proposed. Based on this coordination scheme, optimal controller parameters are generated in both centralized and decentralized fashion to achieve dynamic convoy protection in a unified framework, where distributed optimization technique is applied in the decentralized strategy. This unified framework takes into account the motion of the target to be protected, and the desired system performance, for instance, minimal energy to spend, equal inter-vehicle distance to keep, etc.
Both trajectory tracking and convoy protection tasks are demonstrated through simulations and real-world hardware experiments based on the robotic equipment at Department of Computer Science VII, University of Würzburg.
Gegenstand der Arbeit stellt eine erstmalig unternommene, architekturübergreifende Studie über feldprogrammierbare Logikbausteine zur Implementierung synchroner Schaltkreise dar. Zunächst wird ein Modell für allgemeine feldprogrammiebare Architekturen basierend auf periodischen Graphen definiert. Schließlich werden Bewertungsmaße für Architekturen und Schaltkreislayouts angegeben zur Charakterisierung struktureller Eigenschaften hinsichtlich des Verhaltens in Chipflächenverbrauch und Signalverzögerung. Ferner wird ein generisches Layout-Werkzeug entwickelt, das für beliebige Architekturen und Schaltkreise Implementierungen berechnen und bewerten kann. Abschließend werden neun ressourcenminimalistische Architekturen mit Maschen- und mit Inselstruktur einander gegenübergestellt.
The field of genetics faces a lot of challenges and opportunities in both research and diagnostics due to the rise of next generation sequencing (NGS), a technology that allows to sequence DNA increasingly fast and cheap.
NGS is not only used to analyze DNA, but also RNA, which is a very similar molecule also present in the cell, in both cases producing large amounts of data.
The big amount of data raises both infrastructure and usability problems, as powerful computing infrastructures are required and there are many manual steps in the data analysis which are complicated to execute.
Both of those problems limit the use of NGS in the clinic and research, by producing a bottleneck both computationally and in terms of manpower, as for many analyses geneticists lack the required computing skills.
Over the course of this thesis we investigated how computer science can help to improve this situation to reduce the complexity of this type of analysis.
We looked at how to make the analysis more accessible to increase the number of people that can perform OMICS data analysis (OMICS groups various genomics data-sources).
To approach this problem, we developed a graphical NGS data analysis pipeline aimed at a diagnostics environment while still being useful in research in close collaboration with the Human Genetics Department at the University of Würzburg.
The pipeline has been used in various research papers on covering subjects, including works with direct author participation in genomics, transcriptomics as well as epigenomics.
To further validate the graphical pipeline, a user survey was carried out which confirmed that it lowers the complexity of OMICS data analysis.
We also studied how the data analysis can be improved in terms of computing infrastructure by improving the performance of certain analysis steps.
We did this both in terms of speed improvements on a single computer (with notably variant calling being faster by up to 18 times), as well as with distributed computing to better use an existing infrastructure.
The improvements were integrated into the previously described graphical pipeline, which itself also was focused on low resource usage.
As a major contribution and to help with future development of parallel and distributed applications, for the usage in genetics or otherwise, we also looked at how to make it easier to develop such applications.
Based on the parallel object programming model (POP), we created a Java language extension called POP-Java, which allows for easy and transparent distribution of objects.
Through this development, we brought the POP model to the cloud, Hadoop clusters and present a new collaborative distributed computing model called FriendComputing.
The advances made in the different domains of this thesis have been published in various works specified in this document.
Der Betrieb von Satelliten wird sich in Zukunft gravierend ändern. Die bisher ausgeübte konventionelle Vorgehensweise, bei der die Planung der vom Satelliten auszuführenden Aktivitäten sowie die Kontrolle hierüber ausschließlich vom Boden aus erfolgen, stößt bei heutigen Anwendungen an ihre Grenzen. Im schlimmsten Fall verhindert dieser Umstand sogar die Erschließung bisher ungenutzter Möglichkeiten. Der Gewinn eines Satelliten, sei es in Form wissenschaftlicher Daten oder der Vermarktung satellitengestützter Dienste, wird daher nicht optimal ausgeschöpft.
Die Ursache für dieses Problem lässt sich im Grunde auf eine ausschlaggebende Tatsache zurückführen: Konventionelle Satelliten können ihr Verhalten, d.h. die Folge ihrer Tätigkeiten, nicht eigenständig anpassen. Stattdessen erstellt das Bedienpersonal am Boden - vor allem die Operatoren - mit Hilfe von Planungssoftware feste Ablaufpläne, die dann in Form von Kommandosequenzen von den Bodenstationen aus an die jeweiligen Satelliten hochgeladen werden. Dort werden die Befehle lediglich überprüft, interpretiert und strikt ausgeführt. Die Abarbeitung erfolgt linear. Situationsbedingte Änderungen, wie sie vergleichsweise bei der Codeausführung von Softwareprogrammen durch Kontrollkonstrukte, zum Beispiel Schleifen und Verzweigungen, üblich sind, sind typischerweise nicht vorgesehen. Der Operator ist daher die einzige Instanz, die das Verhalten des Satelliten mittels Kommandierung, per Upload, beeinflussen kann, und auch nur dann, wenn ein direkter Funkkontakt zwischen Satellit und Bodenstation besteht. Die dadurch möglichen Reaktionszeiten des Satelliten liegen bestenfalls bei einigen Sekunden, falls er sich im Wirkungsbereich der Bodenstation befindet. Außerhalb des Kontaktfensters kann sich die Zeitschranke, gegeben durch den Orbit und die aktuelle Position des Satelliten, von einigen Minuten bis hin zu einigen Stunden erstrecken. Die Signallaufzeiten der Funkübertragung verlängern die Reaktionszeiten um weitere Sekunden im erdnahen Bereich. Im interplanetaren Raum erstrecken sich die Zeitspannen aufgrund der immensen Entfernungen sogar auf mehrere Minuten. Dadurch bedingt liegt die derzeit technologisch mögliche, bodengestützte, Reaktionszeit von Satelliten bestenfalls im Bereich von einigen Sekunden.
Diese Einschränkung stellt ein schweres Hindernis für neuartige Satellitenmissionen, bei denen insbesondere nichtdeterministische und kurzzeitige Phänomene (z.B. Blitze und Meteoreintritte in die Erdatmosphäre) Gegenstand der Beobachtungen sind, dar. Die langen Reaktionszeiten des konventionellen Satellitenbetriebs verhindern die Realisierung solcher Missionen, da die verzögerte Reaktion erst erfolgt, nachdem das zu beobachtende Ereignis bereits abgeschlossen ist.
Die vorliegende Dissertation zeigt eine Möglichkeit, das durch die langen Reaktionszeiten entstandene Problem zu lösen, auf. Im Zentrum des Lösungsansatzes steht dabei die Autonomie. Im Wesentlichen geht es dabei darum, den Satelliten mit der Fähigkeit auszustatten, sein Verhalten, d.h. die Folge seiner Tätigkeiten, eigenständig zu bestimmen bzw. zu ändern. Dadurch wird die direkte Abhängigkeit des Satelliten vom Operator bei Reaktionen aufgehoben. Im Grunde wird der Satellit in die Lage versetzt, sich selbst zu kommandieren.
Die Idee der Autonomie wurde im Rahmen der zugrunde liegenden Forschungsarbeiten umgesetzt. Das Ergebnis ist ein autonomes Planungssystem. Dabei handelt es sich um ein Softwaresystem, mit dem sich autonomes Verhalten im Satelliten realisieren lässt. Es kann an unterschiedliche Satellitenmissionen angepasst werden. Ferner deckt es verschiedene Aspekte des autonomen Satellitenbetriebs, angefangen bei der generellen Entscheidungsfindung der Tätigkeiten, über die zeitliche Ablaufplanung unter Einbeziehung von Randbedingungen (z.B. Ressourcen) bis hin zur eigentlichen Ausführung, d.h. Kommandierung, ab. Das Planungssystem kommt als Anwendung in ASAP, einer autonomen Sensorplattform, zum Einsatz. Es ist ein optisches System und dient der Detektion von kurzzeitigen Phänomenen und Ereignissen in der Erdatmosphäre.
Die Forschungsarbeiten an dem autonomen Planungssystem, an ASAP sowie an anderen zu diesen in Bezug stehenden Systemen wurden an der Professur für Raumfahrttechnik des Lehrstuhls Informatik VIII der Julius-Maximilians-Universität Würzburg durchgeführt.
The first part of this thesis deals with the approximability of the traveling salesman problem. This problem is defined on a complete graph with edge weights, and the task is to find a Hamiltonian cycle of minimum weight that visits each vertex exactly once. We study the most important multiobjective variants of this problem. In the multiobjective case, the edge weights are vectors of natural numbers with one component for each objective, and since weight vectors are typically incomparable, the optimal Hamiltonian cycle does not exist. Instead we consider the Pareto set, which consists of those Hamiltonian cycles that are not dominated by some other, strictly better Hamiltonian cycles. The central goal in multiobjective optimization and in the first part of this thesis in particular is the approximation of such Pareto sets.
We first develop improved approximation algorithms for the two-objective metric traveling salesman problem on multigraphs and for related Hamiltonian path problems that are inspired by the single-objective Christofides' heuristic. We further show arguments indicating that our algorithms are difficult to improve. Furthermore we consider multiobjective maximization versions of the traveling salesman problem, where the task is to find Hamiltonian cycles with high weight in each objective. We generalize single-objective techniques to the multiobjective case, where we first compute a cycle cover with high weight and then remove an edge with low weight in each cycle. Since weight vectors are often incomparable, the choice of the edges of low weight is non-trivial. We develop a general lemma that solves this problem and enables us to generalize the single-objective maximization algorithms to the multiobjective case. We obtain improved, randomized approximation algorithms for the multiobjective maximization variants of the traveling salesman problem. We conclude the first part by developing deterministic algorithms for these problems.
The second part of this thesis deals with redundancy properties of complete sets. We call a set autoreducible if for every input instance x we can efficiently compute some y that is different from x but that has the same membership to the set. If the set can be split into two equivalent parts, then it is called weakly mitotic, and if the splitting is obtained by an efficiently decidable separator set, then it is called mitotic. For different reducibility notions and complexity classes, we analyze how redundant its complete sets are.
Previous research in this field concentrates on polynomial-time computable reducibility notions. The main contribution of this part of the thesis is a systematic study of the redundancy properties of complete sets for typical complexity classes and reducibility notions that are computable in logarithmic space. We use different techniques to show autoreducibility and mitoticity that depend on the size of the complexity class and the strength of the reducibility notion considered. For small complexity classes such as NL and P we use self-reducible, complete sets to show that all complete sets are autoreducible. For large complexity classes such as PSPACE and EXP we apply diagonalization methods to show that all complete sets are even mitotic. For intermediate complexity classes such as NP and the remaining levels of the polynomial-time hierarchy we establish autoreducibility of complete sets by locally checking computational transcripts. In many cases we can show autoreducibility of complete sets, while mitoticity is not known to hold. We conclude the second part by showing that in some cases, autoreducibility of complete sets at least implies weak mitoticity.
Network planning has come to great importance during the past decades. Today's telecommunication, traffic systems, and logistics would not have been evolved to the current state without careful analysis of the underlying network problems and precise implementation of the results obtained from those examinations. Graphs with node and arc attributes are a very useful tool to model realistic applications, while on the other hand they are well understood in theory. We investigate network design problems which are motivated particularly from applications in communication networks and logistics. Those problems include the search for homogeneous subgraphs in edge labeled graphs where either the total number of labels or the reload cost are subject to optimize. Further, we investigate some variants of the dial a ride problem. On the other hand, we use node and edge upgrade models to deal with the fact that in many cases one prefers to change existing networks rather than implementing a newly computed solution from scratch. We investigate the construction of bottleneck constrained forests under a node upgrade model, as well as several flow cost problems under a edge based upgrade model. All problems are examined within a framework of multi-criteria optimization. Many of the problems can be shown to be NP-hard, with the consequence that, under the widely accepted assumption that P is not equal to NP, there cannot exist efficient algorithms for solving the problems. This motivates the development of approximation algorithms which compute near-optimal solutions with provable performance guarantee in polynomial time.
Software frameworks for Realtime Interactive Systems (RIS), e.g., in the areas of Virtual, Augmented, and Mixed Reality (VR, AR, and MR) or computer games, facilitate a multitude of functionalities by coupling diverse software modules. In this context, no uniform methodology for coupling these modules does exist; instead various purpose-built solutions have been proposed. As a consequence, important software qualities, such as maintainability, reusability, and adaptability, are impeded.
Many modern systems provide additional support for the integration of Artificial Intelligence (AI) methods to create so called intelligent virtual environments. These methods exacerbate the above-mentioned problem of coupling software modules in the thus created Intelligent Realtime Interactive Systems (IRIS) even more. This, on the one hand, is due to the commonly applied specialized data structures and asynchronous execution schemes, and the requirement for high consistency regarding content-wise coupled but functionally decoupled forms of data representation on the other.
This work proposes an approach to decoupling software modules in IRIS, which is based on the abstraction of architecture elements using a semantic Knowledge Representation Layer (KRL). The layer facilitates decoupling the required modules, provides a means for ensuring interface compatibility and consistency, and in the end constitutes an interface for symbolic AI methods.
In recent years, great progress has been made in the area of Artificial Intelligence (AI) due to the possibilities of Deep Learning which steadily yielded new state-of-the-art results especially in many image recognition tasks.
Currently, in some areas, human performance is achieved or already exceeded.
This great development already had an impact on the area of Optical Music Recognition (OMR) as several novel methods relying on Deep Learning succeeded in specific tasks.
Musicologists are interested in large-scale musical analysis and in publishing digital transcriptions in a collection enabling to develop tools for searching and data retrieving.
The application of OMR promises to simplify and thus speed-up the transcription process by either providing fully-automatic or semi-automatic approaches.
This thesis focuses on the automatic transcription of Medieval music with a focus on square notation which poses a challenging task due to complex layouts, highly varying handwritten notations, and degradation.
However, since handwritten music notations are quite complex to read, even for an experienced musicologist, it is to be expected that even with new techniques of OMR manual corrections are required to obtain the transcriptions.
This thesis presents several new approaches and open source software solutions for layout analysis and Automatic Text Recognition (ATR) for early documents and for OMR of Medieval manuscripts providing state-of-the-art technology.
Fully Convolutional Networks (FCN) are applied for the segmentation of historical manuscripts and early printed books, to detect staff lines, and to recognize neume notations.
The ATR engine Calamari is presented which allows for ATR of early prints and also the recognition of lyrics.
Configurable CNN/LSTM-network architectures which are trained with the segmentation-free CTC-loss are applied to the sequential recognition of text but also monophonic music.
Finally, a syllable-to-neume assignment algorithm is presented which represents the final step to obtain a complete transcription of the music.
The evaluations show that the performances of any algorithm is highly depending on the material at hand and the number of training instances.
The presented staff line detection correctly identifies staff lines and staves with an $F_1$-score of above $99.5\%$.
The symbol recognition yields a diplomatic Symbol Accuracy Rate (dSAR) of above $90\%$ by counting the number of correct predictions in the symbols sequence normalized by its length.
The ATR of lyrics achieved a Character Error Rate (CAR) (equivalently the number of correct predictions normalized by the sentence length) of above $93\%$ trained on 771 lyric lines of Medieval manuscripts and of 99.89\% when training on around 3.5 million lines of contemporary printed fonts.
The assignment of syllables and their corresponding neumes reached $F_1$-scores of up to $99.2\%$.
A direct comparison to previously published performances is difficult due to different materials and metrics.
However, estimations show that the reported values of this thesis exceed the state-of-the-art in the area of square notation.
A further goal of this thesis is to enable musicologists without technical background to apply the developed algorithms in a complete workflow by providing a user-friendly and comfortable Graphical User Interface (GUI) encapsulating the technical details.
For this purpose, this thesis presents the web-application OMMR4all.
Its fully-functional workflow includes the proposed state-of-the-art machine-learning algorithms and optionally allows for a manual intervention at any stage to correct the output preventing error propagation.
To simplify the manual (post-) correction, OMMR4all provides an overlay-editor that superimposes the annotations with a scan of the original manuscripts so that errors can easily be spotted.
The workflow is designed to be iteratively improvable by training better models as soon as new Ground Truth (GT) is available.
Performance Assessment of Resource Management Strategies for Cellular and Wireless Mesh Networks
(2015)
The rapid growth in the field of communication networks has been truly amazing in the last decades. We are currently experiencing a continuation thereof with an increase in traffic and the emergence of new fields of application. In particular, the latter is interesting since due to advances in the networks and new devices, such as smartphones, tablet PCs, and all kinds of Internet-connected devices, new additional applications arise from different areas. What applies for all these services is that they come from very different directions and belong to different user groups. This results in a very heterogeneous application mix with different requirements and needs on the access networks.
The applications within these networks typically use the network technology as a matter of course, and expect that it works in all situations and for all sorts of purposes without any further intervention. Mobile TV, for example, assumes that the cellular networks support the streaming of video data. Likewise, mobile-connected electricity meters rely on the timely transmission of accounting data for electricity billing. From the perspective of the communication networks, this requires not only the technical realization for the individual case, but a broad consideration of all circumstances and all requirements of special devices and applications of the users.
Such a comprehensive consideration of all eventualities can only be achieved by a dynamic, customized, and intelligent management of the transmission resources. This management requires to exploit the theoretical capacity as much as possible while also taking system and network architecture as well as user and application demands into account. Hence, for a high level of customer satisfaction, all requirements of the customers and the applications need to be considered, which requires a multi-faceted resource management.
The prerequisite for supporting all devices and applications is consequently a holistic resource management at different levels. At the physical level, the technical possibilities provided by different access technologies, e.g., more transmission antennas, modulation and coding of data, possible cooperation between network elements, etc., need to be exploited on the one hand. On the other hand, interference and changing network conditions have to be counteracted at physical level. On the application and user level, the focus should be on the customer demands due to the currently increasing amount of different devices and diverse applications (medical, hobby, entertainment, business, civil protection, etc.).
The intention of this thesis is the development, investigation, and evaluation of a holistic resource management with respect to new application use cases and requirements for the networks. Therefore, different communication layers are investigated and corresponding approaches are developed using simulative methods as well as practical emulation in testbeds. The new approaches are designed with respect to different complexity and implementation levels in order to cover the design space of resource management in a systematic way. Since the approaches cannot be evaluated generally for all types of access networks, network-specific use cases and evaluations are finally carried out in addition to the conceptual design and the modeling of the scenario.
The first part is concerned with management of resources at physical layer. We study distributed resource allocation approaches under different settings. Due to the ambiguous performance objectives, a high spectrum reuse is conducted in current cellular networks. This results in possible interference between cells that transmit on the same frequencies. The focus is on the identification of approaches that are able to mitigate such interference.
Due to the heterogeneity of the applications in the networks, increasingly different application-specific requirements are experienced by the networks. Consequently, the focus is shifted in the second part from optimization of network parameters to consideration and integration of the application and user needs by adjusting network parameters. Therefore, application-aware resource management is introduced to enable efficient and customized access networks.
As indicated before, approaches cannot be evaluated generally for all types of access networks. Consequently, the third contribution is the definition and realization of the application-aware paradigm in different access networks. First, we address multi-hop wireless mesh networks. Finally, we focus with the fourth contribution on cellular networks. Application-aware resource management is applied here to the air interface between user device and the base station. Especially in cellular networks, the intensive cost-driven competition among the different operators facilitates the usage of such a resource management to provide cost-efficient and customized networks with respect to the running applications.
Automation in Software Performance Engineering Based on a Declarative Specification of Concerns
(2019)
Software performance is of particular relevance to software system design, operation, and evolution because it has a significant impact on key business indicators. During the life-cycle of a software system, its implementation, configuration, and deployment are subject to multiple changes that may affect the end-to-end performance characteristics. Consequently, performance analysts continually need to provide answers to and act based on performance-relevant concerns. To ensure a desired level of performance, software performance engineering provides a plethora of methods, techniques, and tools for measuring, modeling, and evaluating performance properties of software systems. However, the answering of performance concerns is subject to a significant semantic gap between the level on which performance concerns are formulated and the technical level on which performance evaluations are actually conducted. Performance evaluation approaches come with different strengths and limitations concerning, for example, accuracy, time-to-result, or system overhead. For the involved stakeholders, it can be an elaborate process to reasonably select, parameterize and correctly apply performance evaluation approaches, and to filter and interpret the obtained results. An additional challenge is that available performance evaluation artifacts may change over time, which requires to switch between different measurement-based and model-based performance evaluation approaches during the system evolution. At model-based analysis, the effort involved in creating performance models can also outweigh their benefits.
To overcome the deficiencies and enable an automatic and holistic evaluation of performance throughout the software engineering life-cycle requires an approach that: (i) integrates multiple types of performance concerns and evaluation approaches, (ii) automates performance model creation, and (iii) automatically selects an evaluation methodology tailored to a specific scenario. This thesis presents a declarative approach —called Declarative Performance Engineering (DPE)— to automate performance evaluation based on a humanreadable specification of performance-related concerns. To this end, we separate the definition of performance concerns from their solution. The primary scientific contributions presented in this thesis are:
A declarative language to express performance-related concerns and a corresponding processing framework:
We provide a language to specify performance concerns independent of a concrete performance evaluation approach. Besides the specification of functional aspects, the language allows to include non-functional tradeoffs optionally. To answer these concerns, we provide a framework architecture and a corresponding reference implementation to process performance concerns automatically. It allows to integrate arbitrary performance evaluation approaches and is accompanied by reference implementations for model-based and measurement-based performance evaluation.
Automated creation of architectural performance models from execution traces:
The creation of performance models can be subject to significant efforts outweighing the benefits of model-based performance evaluation. We provide a model extraction framework that creates architectural performance models based on execution traces, provided by monitoring tools.The framework separates the derivation of generic information from model creation routines. To derive generic information, the framework combines state-of-the-art extraction and estimation techniques. We isolate object creation routines specified in a generic model builder interface based on concepts present in multiple performance-annotated architectural modeling formalisms. To create model extraction for a novel performance modeling formalism, developers only need to write object creation routines instead of creating model extraction software from scratch when reusing the generic framework.
Automated and extensible decision support for performance evaluation approaches:
We present a methodology and tooling for the automated selection of a performance evaluation approach tailored to the user concerns and application scenario. To this end, we propose to decouple the complexity of selecting a performance evaluation approach for a given scenario by providing solution approach capability models and a generic decision engine. The proposed capability meta-model enables to describe functional and non-functional capabilities of performance evaluation approaches and tools at different granularities. In contrast to existing tree-based decision support mechanisms, the decoupling approach allows to easily update characteristics of solution approaches as well as appending new rating criteria and thereby stay abreast of evolution in performance evaluation tooling and system technologies.
Time-to-result estimation for model-based performance prediction:
The time required to execute a model-based analysis plays an important role in different decision processes. For example, evaluation scenarios might require the prediction results to be available in a limited period of time such that the system can be adapted in time to ensure the desired quality of service. We propose a method to estimate the time-to-result for modelbased performance prediction based on model characteristics and analysis parametrization. We learn a prediction model using performancerelevant features thatwe determined using statistical tests. We implement the approach and demonstrate its practicability by applying it to analyze a simulation-based multi-step performance evaluation approach for a representative architectural performance modeling formalism.
We validate each of the contributions based on representative case studies. The evaluation of automatic performance model extraction for two case study systems shows that the resulting models can accurately predict the performance behavior. Prediction accuracy errors are below 3% for resource utilization and mostly less than 20% for service response time. The separate evaluation of the reusability shows that the presented approach lowers the implementation efforts for automated model extraction tools by up to 91%. Based on two case studies applying measurement-based and model-based performance evaluation techniques, we demonstrate the suitability of the declarative performance engineering framework to answer multiple kinds of performance concerns customized to non-functional goals. Subsequently, we discuss reduced efforts in applying performance analyses using the integrated and automated declarative approach. Also, the evaluation of the declarative framework reviews benefits and savings integrating performance evaluation approaches into the declarative performance engineering framework. We demonstrate the applicability of the decision framework for performance evaluation approaches by applying it to depict existing decision trees. Then, we show how we can quickly adapt to the evolution of performance evaluation methods which is challenging for static tree-based decision support systems. At this, we show how to cope with the evolution of functional and non-functional capabilities of performance evaluation software and explain how to integrate new approaches. Finally, we evaluate the accuracy of the time-to-result estimation for a set of machinelearning algorithms and different training datasets. The predictions exhibit a mean percentage error below 20%, which can be further improved by including performance evaluations of the considered model into the training data. The presented contributions represent a significant step towards an integrated performance engineering process that combines the strengths of model-based and measurement-based performance evaluation. The proposed performance concern language in conjunction with the processing framework significantly reduces the complexity of applying performance evaluations for all stakeholders. Thereby it enables performance awareness throughout the software engineering life-cycle. The proposed performance concern language removes the semantic gap between the level on which performance concerns are formulated and the technical level on which performance evaluations are actually conducted by the user.
In produzierenden Unternehmen werden verschiedene Vorgehensweisen zur Planung, Überwachung und Steuerung von Produktionsabläufen eingesetzt. Einer dieser Methoden wird als Vorgangsknotennetzplantechnik bezeichnet. Die einzelnen Produktionsschritte werden als Knoten definiert und durch Pfeile miteinander verbunden. Die Pfeile stellen die Beziehungen der jeweiligen Vorgänge zueinander und damit den Produktionsablauf dar. Diese Technik erlaubt den Anwendern einen umfassenden Überblick über die einzelnen Prozessrelationen. Zusätzlich können mit ihr Vorgangszeiten und Produktfertigstellungszeiten ermittelt werden, wodurch eine ausführliche Planung der Produktion ermöglicht wird. Ein Nachteil dieser Technik begründet sich in der alleinigen Darstellung einer ausführbaren Prozessabfolge. Im Falle eines Störungseintritts mit der Folge eines nicht durchführbaren Vorgangs muss von dem originären Prozess abgewichen werden. Aufgrund dessen wird eine Neuplanung erforderlich. Es werden Alternativen für den gestörten Vorgang benötigt, um eine Fortführung des Prozesses ungeachtet der Störung zu erreichen. Innerhalb dieser Arbeit wird daher eine Erweiterung der Vorgangsknotennetzplantechnik beschrieben, die es erlaubt, ergänzend zu dem geplanten Soll-Prozess Alternativvorgänge für einzelne Vorgänge darzulegen. Diese Methode wird als Maximalnetzplan bezeichnet. Die Alternativen werden im Falle eines Störungseintritts automatisch evaluiert und dem Anwender in priorisierter Reihenfolge präsentiert. Durch die Verwendung des Maximalnetzplans kann eine aufwendige Neuplanung vermieden werden. Als Anwendungsbeispiel dient ein Montageprozess, mithilfe dessen die Verwendbarkeit der Methode dargelegt wird. Weiterführend zeigt eine zeitliche Analyse zufallsbedingter Maximalnetzpläne eine Begründung zur Durchführung von Alternativen und damit den Nutzen des Maximalnetzplans auf. Zusätzlich sei angemerkt, dass innerhalb dieser Arbeit verwendete Begrifflichkeiten wie Anwender, Werker oder Mitarbeiter in maskuliner Schreibweise niedergeschrieben werden. Dieses ist ausschließlich der Einfachheit geschuldet und nicht dem Zweck der Diskriminierung anderer Geschlechter dienlich. Die verwendete Schreibweise soll alle Geschlechter ansprechen, ob männlich, weiblich oder divers.
Energy efficiency of computing systems has become an increasingly important issue over the last decades. In 2015, data centers were responsible for 2% of the world's greenhouse gas emissions, which is roughly the same as the amount produced by air travel.
In addition to these environmental concerns, power consumption of servers in data centers results in significant operating costs, which increase by at least 10% each year.
To address this challenge, the U.S. EPA and other government agencies are considering the use of novel measurement methods in order to label the energy efficiency of servers.
The energy efficiency and power consumption of a server is subject to a great number of factors, including, but not limited to, hardware, software stack, workload, and load level.
This huge number of influencing factors makes measuring and rating of energy efficiency challenging. It also makes it difficult to find an energy-efficient server for a specific use-case. Among others, server provisioners, operators, and regulators would profit from information on the servers in question and on the factors that affect those servers' power consumption and efficiency. However, we see a lack of measurement methods and metrics for energy efficiency of the systems under consideration.
Even assuming that a measurement methodology existed, making decisions based on its results would be challenging. Power prediction methods that make use of these results would aid in decision making. They would enable potential server customers to make better purchasing decisions and help operators predict the effects of potential reconfigurations.
Existing energy efficiency benchmarks cannot fully address these challenges, as they only measure single applications at limited sets of load levels. In addition, existing efficiency metrics are not helpful in this context, as they are usually a variation of the simple performance per power ratio, which is only applicable to single workloads at a single load level. Existing data center efficiency metrics, on the other hand, express the efficiency of the data center space and power infrastructure, not focusing on the efficiency of the servers themselves. Power prediction methods for not-yet-available systems that could make use of the results provided by a comprehensive power rating methodology are also lacking. Existing power prediction models for hardware designers have a very fine level of granularity and detail that would not be useful for data center operators.
This thesis presents a measurement and rating methodology for energy efficiency of servers and an energy efficiency metric to be applied to the results of this methodology. We also design workloads, load intensity and distribution models, and mechanisms that can be used for energy efficiency testing. Based on this, we present power prediction mechanisms and models that utilize our measurement methodology and its results for power prediction.
Specifically, the six major contributions of this thesis are:
We present a measurement methodology and metrics for energy efficiency rating of servers that use multiple, specifically chosen workloads at different load levels for a full system characterization.
We evaluate the methodology and metric with regard to their reproducibility, fairness, and relevance. We investigate the power and performance variations of test results and show fairness of the metric through a mathematical proof and a correlation analysis on a set of 385 servers. We evaluate the metric's relevance by showing the relationships that can be established between metric results and third-party applications.
We create models and extraction mechanisms for load profiles that vary over time, as well as load distribution mechanisms and policies. The models are designed to be used to define arbitrary dynamic load intensity profiles that can be leveraged for benchmarking purposes. The load distribution mechanisms place workloads on computing resources in a hierarchical manner.
Our load intensity models can be extracted in less than 0.2 seconds and our resulting models feature a median modeling error of 12.7% on average. In addition, our new load distribution strategy can save up to 10.7% of power consumption on a single server node.
We introduce an approach to create small-scale workloads that emulate the power consumption-relevant behavior of large-scale workloads by approximating their CPU performance counter profile, and we introduce TeaStore, a distributed, micro-service-based reference application. TeaStore can be used to evaluate power and performance model accuracy, elasticity of cloud auto-scalers, and the effectiveness of power saving mechanisms for distributed systems.
We show that we are capable of emulating the power consumption behavior of realistic workloads with a mean deviation less than 10% and down to 0.2 watts (1%). We demonstrate the use of TeaStore in the context of performance model extraction and cloud auto-scaling also showing that it may generate workloads with different effects on the power consumption of the system under consideration.
We present a method for automated selection of interpolation strategies for performance and power characterization. We also introduce a configuration approach for polynomial interpolation functions of varying degrees that improves prediction accuracy for system power consumption for a given system utilization.
We show that, in comparison to regression, our automated interpolation method selection and configuration approach improves modeling accuracy by 43.6% if additional reference data is available and by 31.4% if it is not.
We present an approach for explicit modeling of the impact a virtualized environment has on power consumption and a method to predict the power consumption of a software application. Both methods use results produced by our measurement methodology to predict the respective power consumption for servers that are otherwise not available to the person making the prediction.
Our methods are able to predict power consumption reliably for multiple hypervisor configurations and for the target application workloads. Application workload power prediction features a mean average absolute percentage error of 9.5%.
Finally, we propose an end-to-end modeling approach for predicting the power consumption of component placements at run-time. The model can also be used to predict the power consumption at load levels that have not yet been observed on the running system.
We show that we can predict the power consumption of two different distributed web applications with a mean absolute percentage error of 2.2%. In addition, we can predict the power consumption of a system at a previously unobserved load level and component distribution with an error of 1.2%.
The contributions of this thesis already show a significant impact in science and industry. The presented efficiency rating methodology, including its metric, have been adopted by the U.S. EPA in the latest version of the ENERGY STAR Computer Server program. They are also being considered by additional regulatory agencies, including the EU Commission and the China National Institute of Standardization. In addition, the methodology's implementation and the underlying methodology itself have already found use in several research publications.
Regarding future work, we see a need for new workloads targeting specialized server hardware. At the moment, we are witnessing a shift in execution hardware to specialized machine learning chips, general purpose GPU computing, FPGAs being embedded into compute servers, etc. To ensure that our measurement methodology remains relevant, workloads covering these areas are required. Similarly, power prediction models must be extended to cover these new scenarios.
In this work, a novel method for estimating the relative pose of a known object is presented, which relies on an application-specific data fusion process. A PMD-sensor in conjunction with a CCD-sensor is used to perform the pose estimation. Furthermore, the work provides a method for extending the measurement range of the PMD sensor along with the necessary calibration methodology. Finally, extensive measurements on a very accurate Rendezvous and Docking testbed are made to evaluate the performance, what includes a detailed discussion of lighting conditions.
This thesis is devoted to the study of computational complexity theory, a branch of theoretical computer science. Computational complexity theory investigates the inherent difficulty in designing efficient algorithms for computational problems. By doing so, it analyses the scalability of computational problems and algorithms and places practical limits on what computers can actually accomplish. Computational problems are categorised into complexity classes. Among the most important complexity classes are the class NP and the subclass of NP-complete problems, which comprises many important optimisation problems in the field of operations research. Moreover, with the P-NP-problem, the class NP represents the most important unsolved question in computer science. The first part of this thesis is devoted to the study of NP-complete-, and more generally, NP-hard problems. It aims at improving our understanding of this important complexity class by systematically studying how altering NP-hard sets affects their NP-hardness. This research is related to longstanding open questions concerning the complexity of unions of disjoint NP-complete sets, and the existence of sparse NP-hard sets. The second part of the thesis is also dedicated to complexity classes but takes a different perspective: In a sense, after investigating the interior of complexity classes in the first part, the focus shifts to the description of complexity classes and thereby to the exterior in the second part. It deals with the description of complexity classes through leaf languages, a uniform framework which allows us to characterise a great variety of important complexity classes. The known concepts are complemented by a new leaf-language model. To a certain extent, this new approach combines the advantages of the known models. The presented results give evidence that the connection between the theory of formal languages and computational complexity theory might be closer than formerly known.
Parametric weighted finite automata (PWFA) are a multi-dimensional generalization of weighted finite automata. The expressiveness of PWFA contains the expressiveness of weighted finite automata as well as the expressiveness of affine iterated function system. The thesis discusses theory and applications of PWFA. The properties of PWFA definable sets are studied and it is shown that some fractal generator systems can be simulated using PWFA and that various real and complex functions can be represented by PWFA. Furthermore, the decoding of PWFA and the interpretation of PWFA definable sets is discussed.