05 Fakultät Informatik, Elektrotechnik und Informationstechnik
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6
Browse
Search Results
Item Open Access Generic templates for monitoring agents(2018) Weise, MarcThis thesis presents an agent-centric approach for monitoring IT resources, which enables the execution of preprocessing and aggregation steps directly on the target systems in order to limit data transfers to a central server and allow a local event detection and treatment. To keep the agent behavior definition as simple as possible, an extendable template model is introduced which can be used to define Monitoring Pipelines by chaining individual processing steps. Furthermore this work demonstrates how a graphical editor can be implemented which also allows non-experts in the field of monitoring to create and modify Monitoring Templates.Item Open Access Personenbezogene Daten im Data Lake(2018) Ebinger, FelixBig-Data-Analysen bieten Wettbewerbsvorteile, ermöglichen Innovationen und können zu einer höheren Qualität von Produkten oder Serviceleistungen beitragen. Insbesondere die Analyse von Kundendaten und des Kundenverhaltens eröffnet vielfältige Möglichkeiten, um dem Kunden auf ihn zugeschnittene Angebote zu unterbreiten und um so zu höheren Umsätzen und zu einer höheren Kundenzufriedenheit beizutragen. Für die dafür benötigten Daten werden geeignete Speichersysteme benötigt. Ein solches System stellt der Data Lake dar. Neben der gut skalierenden und günstigen Speicherung von Daten ist auch die Auswertung der Daten mittels explorativer Analysen bereits im Design angelegt. Gleichzeitig steht aber auch der Schutz, genauer der fehlende Schutz der Privatsphäre, des Einzelnen bei Big Data Verarbeitungen im Mittelpunkt der öffentlichen Aufmerksamkeit und Kritik. Insbesondere wird vor dem so entstehenden „gläsernen Menschen“ und den daraus resultierenden gesellschaftlichen Folgen gewarnt. Die sich daraus ergebenden Fragen, in welchem Umfang und auf welche Art personenbezogene Daten verarbeitet werden dürfen, bedürfen, neben einer ethisch-moralischen, vor allem einer rechtlichen Antwort. Die europäische Datenschutzgrundverordnung stellt hierzu den rechtlichen Rahmen dar, in dem personenbezogene Daten verarbeitet werden dürfen. In dieser Arbeit werden die gesetzlichen Anforderungen mit dem Konzept des Data Lakes abgeglichen und es wird aufgezeigt, wo Herausforderungen beim Design und bei der Implementierung eines Data Lakes entstehen (z.B. Transparenz, Zweckbindung, Recht auf Löschung). Zudem werden Lösungsansätze für diese Herausforderungen entwickelt und vorgestellt. Aus den einzelnen Lösungsansätzen werden zwei Lösungskonzepte für einige der identifizierten Herausforderungen entwickelt. Eines der Konzepte, ein Metadaten-Modell, wird dabei prototypisch umgesetzt und anhand von Use Cases beispielhaft getestet.Item Open Access Octo-Tiger: Binary star systems with HPX on Nvidia P100(2018) Daiß, GregorStellar mergers between two suns are a significant field of study since they can lead to astrophysical phenomena such as type Ia supernovae. Octo-Tiger simulates merging stars by computing self-gravitating astrophysical fluids. By relying on the high-level library HPX for parallelization and Vc for vectorization, Octo-Tiger combines high performance with ease of development. For accurate simulations, Octo-Tiger requires massive computational resources. To improve hardware utilization, we introduce a stencil-based approach for computing the gravitational field using the fast multipole method. This approach was tailored for machines with wide vector units like Intel's Knights Landing or modern GPUs. Our implementation targets AVX512 enabled processors and is backward compatible with older vector extensions (AVX2, AVX, SSE). We further extended our approach to make use of available NVIDIA GPUs as coprocessors. We developed a tasking system that processes critical compute kernels on the GPU or the processor, depending on their utilization. Using the stencil-based fast multipole method, we gain a consistent speedup on all platforms, over the classical interaction-list-based implementation. On an Intel Xeon Phi 7210, we achieve a speedup of 1.9x. On a heterogeneous node with an Intel Xeon E5-2690 v3, we can obtain a speedup of 1.46x by adding an NVIDIA P100 GPU.Item Open Access Deep reinforcement learning for high-level behavior decision making(2018) Dittrich, FlorianAs the vision of fully autonomous vehicles potentially introduces significant benefits for our society, this work investigates approaches for sequential decision making for high-level actions in highway scenarios. These scenarios are modeled using an markov decision process (MDP) and consider deep reinforcement learning to solve it. Our approach, based on deep Q-networks (DQNs), is able to fully avoid collisions and learns a policy that results in comfortable trajectories compared to baseline policies we developed. One of the main challenges for reinforcement learning are sparse rewards, which we aim to overcome employing reward shaping. Additionally, the necessity of multiple layers of non-liniearities in the DQN algorithm is empirically evaluated using our scenarios. The results support the usage of multiple levels of non-linearities, as a linear variant of the DQN is not capable of learning effective policies in our experiments. Due to a weight initialization with behavioral cloning, an acceleration of the learning procedure is achieved.Item Open Access An energy-aware mobile gateway for Bluetooth low energy-powered Internet of Things devices(2018) Firmansyah, Mochamad DandyThe term of Internet of Things (IoT) has currently become a novelty in the Internet as an innovation to connect things from all around the world where various sensors are connected using gateways. However, it is not a straightforward task to design such gateways owing to several problems. For instance, there typically exist severe energy consumption constraints due to the limited power source. In most cases, a gateway has to spend an amount of energy for processing the collected data in the network. Additionally, there are myriad of different user interface functions for various services, which in turn raises the question about the reliability and scalability of such gateways. To support the IoT vision, many people have recently used smart mobile devices, e.g., smartphones, tablets, PDA, and laptops, as a gateway for data acquisition in IoT so that these IoT devices can be used in a broader scope. This concept of exploiting our smart devices emerges thanks to their ability to connect things to the cloud via the Internet. In fact, there exist a communication gap between the things implemented with limited power sources to sense the environmental data and the cloud services. Fortunately, this gap can be bridged by adopting smartphones for forwarding the collected data using their wireless connection technologies. One of the critical technologies that can be used to bridge this communication gap while also still maintaining low energy consumption is Bluetooth Low Energy (BLE). As leverage from the original Bluetooth technology, BLE or known as Bluetooth Smart was initially designed as a power-friendly wireless technology aimed for some novel applications in many industries. To save energy, BLE can be set in a sleep mode and wake up only to receive or send possible packet periodically. By the usage of BLE in modern smartphones, a mobile gateway system can be made in a way that data from the sensors can be passed to the cloud while also considering the energy efficiency in the mobile gateway itself. In this thesis, we propose a software architecture of energy-aware mobile gateways for IoT applications. The proposed architecture makes continual and efficient data transmission from a set of predefined devices. Moreover, the gateway architecture implements several scheduling algorithms used to efficiently control the sleep mode operations besides handle the simultaneous connection to several BLE sensors. The presented scheduling algorithms comprise Semaphore, Round Robin, Exhaustive Polling and Fair Exhaustive Polling algorithms. To implement the BLE device priority-based approach, several multi-criteria decision making (MCDM) algorithms are also implemented to prioritize the device based on several criteria, such as device power usage, received signal strength indication and the device state. Examples of such MCDM algorithms that have been implemented in this work are the Analytic Hierarchy Process and the Weighted Sum Model. Furthermore, the algorithms implemented are then evaluated based on two quality of service(QoS) metrics, including the power consumption of the mobile gateway and the throughput defined regarding the number of packets received per second. The evaluation results showed that Fair Exhaustive Polling (FEP) consumes the lowest energy consumption compared to all other scheduling algorithms with only 12,79 mW. On the other hand, Exhaustive Polling with Analytical Hierarchy Process (EPAHP) has the worst energy consumption among the examined algorithms with 49,14 mW. Concerning the throughput, the Exhaustive Polling combined with Weighted Sum Model (EPWSM) has the most prominent data throughput compared to all other algorithms with 101.18 packets/s while Fair Exhaustive Polling (FEP) has the lowest throughput value with 50.98 packets/s. To sum up, the proposed mobile gateway architecture is exceptionally efficient for handling data forwarding from multiple BLE sensors to the cloud services with energy awareness.Item Open Access Operator latency in a Complex Event Processing application(2018) Hagenmayer, SimonComplex Event Processing often comes with an enormous amount of event data that needs to be processed. Hence, parallelization plays a significant role in handling high workload situations. The cost of an application however is often defined by the amount of used resources, like in Cloud computing, where the pay-as-you-go model applies. Still, one wants to have a working system that can handle traffic peaks within a given latency bound, so the resources-to-latency-proportion needs to be optimized. Previous work mostly focused on studying complex operator types in specific environments. In this thesis however, we want to get a general view, how parallelization degrees and types influence our CEP system, to be able to estimate what costs could arise. Therefore, a CEP application was created that simulates different system conditions with respect to workload, operator processing time and others, in order to test and analyze the latency properties of a wait operator. This work provides an overview over latency behavior of operators in an example Complex Event Processing application, which can provide a basis for future work in creating an optimized system, that not only keeps a certain latency threshold but also minimizes the costs and resources needed to achieve this goal.Item Open Access Divide-and-conquer scheduling for time-sensitive networks(2018) Bansal, BharatThe advent of the Internet of Things(IoT) and Industry 4.0 coupled with the increase in the deployment of Cyber-physical systems(CPS), which includes industrial automation systems, has increased the urgency to deploy networking technologies with real-time guarantees. The IEEE 802.1Obv standard has recently emerged as a viable alternative for the future of real-time communication over Ethernet that has stringent end-to-end latency and jitter requirements after it has been recently standardized by the Time Sensitive Network Task Group. Scheduling in Time-Sensitive Network(TSN) has not been standardized by the Task Group to carry the scheduled traffic till now. In this paper, we address the scheduling problem for a Time-Sensitive Network(TSN) for large and complex networks. State-of-the-art scheduling algorithms take a centralized approach to compute transmission schedules for time-triggered traffic. Centralized approaches generally compute fairly optimum transmission schedules but the runtimes for the centralized approaches are high (order of days) for large and complex networks. The aim is to generate a scheduling algorithm having lower runtimes as compared to centralized approaches for large networks and having large time-triggered flows even if the computed scheduling has a worse optimum solution than the centralized algorithm. We present a divide-and-conquer approach that might be apter in these scenarios, as it may generate a schedule with lower runtimes as compared to the centralized approaches. In particular, we present an Integer Linear Program(ILP) based formulation and a simple heuristics for the divide-and-conquer approach. Moreover, we parallelize the problem in order to further reduce the runtimes. We evaluate the optimality, utilization, scalability of the different approaches. Furthermore, we evaluate the runtimes for different network configurations and discuss the trade-off between heuristics and ILP based formulation.Item Open Access Vergleich und Analyse geläufiger CEP Systeme(2018) Göggel, JonathanHeutzutage werden komplexe Anfragen in Echtzeit auf großen Datenmengen ausgeführt. Immer mehr Daten fallen an und das Interesse diese in Echtzeit zu analysieren steigt. Die Performance eines Systems ist ein enorm wichtiger Faktor. Momentan setzen besonders große Firmen wie Google, Amazon und Netflix CEP-Systeme ein, um effizient Nutzerdaten zu analysieren und dem Anwender daraufhin Empfehlungen vorzuschlagen. Die aktuell verfügbaren CEP-Frameworks verhalten sich jeweils unterschiedlich und haben unterschiedliche Ziele. Bisherige Auswertungen fokussieren sich nur auf jeweils ein Framework und optimieren dieses. In meiner Arbeit werden verschiedene Frameworks gegenübergestellt und untersucht, wie flexibel sie angesteuert werden können und inwieweit sie zur Laufzeit detaillierte statistische Werte liefern können. Des Weiteren wird eine API entworfen, die ermöglicht verschiedene CEP Frameworks anzusprechen und somit standardisiert den Parallelisierungsgrad und somit die Performance eines CEP-Systems zu verbessern. Durch die Standardisierung ist es auch möglich die Performance bei CEP-Systemen mit mehrere CEP-Frameworks zu regeln. Im ersten Teil der Ausarbeitung werden verschiedene Frameworks verglichen und untersucht inwieweit sich diese für eine zentrale Ansteuerung eignen. Im zweiten Teil wird ein Interface definiert und zum Evaluieren beispielhaft ein Adapter für ein CEP-Framework erstellt.Item Open Access Combining application layer and network layer filtering in Pub/Sub systems(2018) Sakkal, FadiContent-based Publish/Subscribe systems deliver notifications from publishers to subscribers based on the content of the notifications. Low latency and bandwidth efficiency are two major concerns in these systems. Hybrid Publish/Subscribe systems were developed to take advantage of the new trend of Software Defined Networking (SDN). In these systems, notifications can be filtered on two layers, namely the application-layer and the network-layer. Since each one of these two layers has advantages over the other, it was important to have a selection algorithm to decide on which layer a certain notification has to be filtered. When choosing to filter notifications on the application layer, notifications have to be sent to software filters (servers) located in the topology. The placement of these servers is important for the overall performance of the system. In this thesis, we propose three different placement algorithms, each of which considers a different aspect of the system and tries to place the servers in a way that improves this aspect. The K-Center placement algorithm aims at minimizing the maximum distance between the sending node and the destination server. This will give an upper bound on the worst-case scenario regarding the latency. TheK-Median placement algorithm is designed to minimize the average distance that the sent notifications have to pass to reach the server. The Utilitarian placement algorithm focuses on providing faster service to the packets which are intended for many subscribers, giving better overall service to the majority of the users at the expense of a worse service for some others. We have evaluated the proposed placement algorithms and compared them to each other according to different categories using two different topologies and two different distributions for the published notifications. As expected, the K-Center algorithm proved to be better in minimizing the maximum distance required for a notification to reach a server, while the K-Median and the Utilitarian algorithms showed similar results in most of the cases and were better in minimizing the average distance.Item Open Access Automatic indoor modelling using crowd-sensed point clouds(2018) Prabha Sekar, SuriyaThere are many well-established approaches to model outdoor environments but it has been a challenge to come up with an accurate and reliable approach for indoor modelling. Outdoor automatic modelling and mapping are achieved using satellite positioning systems like GPS. However, for indoor modelling, due to localization problems, positioning systems do not help. Therefore, the indoor positioning is done by combining inertial sensors along with area learning. Previously, the 3D models for indoor environments were done either manually which is a time-consuming process or by using range images obtained from laser range scanners which are an expensive approach. For overcoming this drawback, we suggest utilizing crowd sensing in order to obtain the environment’s spatial information. Crowd-sensing is performed by a group of participants using their mobile devices to execute certain assigned tasks. Nowadays modern mobile devices are used for many crowd sensing applications with the help of the ubiquitous presence of such powerful devices. In our approach, the crowd sensing task involves collecting scans of rooms in public buildings using these mobile devices. Thus, we depend on the new wave of powerful devices, e.g. Google Tango, Microsoft Hololens and Apple ARKit which generates optimal scanned data i.e. 3D point clouds which can be crowd sensed and then processed to automatically generate indoor models. These devices provide capabilities like depth perception, area learning and motion tracking, which help to acquire the spatial information of the room and its relative position. Even though there is more energy consumption on the mobile device which is responsible for collecting large dataset, this can be reduced by applying Octree compression which reduces the amount of data of the scanned point clouds. The scanned data is sent to the server where it gets processed and generates the required indoor model. The proposed tool will derive the 3D indoor model of a floor. The tool is capable of extracting all the planes from the point clouds, detecting all the room surfaces (i.e. at least four walls, a ceiling and a floor), classifying wall openings and classifying high-level semantics (i.e. furniture). The basic model for all the rooms in the floor is generated and in order to build a floor model, the ground truth of that floor is utilized. Finally, the accuracy of the basic model is enhanced using the grammar model fitting tool. Using the point cloud scans from one of our University campuses, the final indoor modelling tool was able to derive the complete 3D floor plan in compliance with the ground truth.