Universität Stuttgart

Permanent URI for this communityhttps://elib.uni-stuttgart.de/handle/11682/1

Browse

Search Results

Now showing 1 - 10 of 324

Open Access
Distributed Deep Reinforcement Learning for Learn-to-optimize
(2023) Mayer, Paul
In the context of increasingly complex applications, e.g., robust performance tuning in Integrated Circuit Design, conventional optimization methods have difficulties in achieving satisfactory results while keeping to a limited time budget. Therefore, learning optimization algorithms becomes more and more interesting, replacing the established way of hand-crafting or tweaking algorithms. Learned algorithms reduce the amount of assumptions and expert knowledge necessary to create state-of-the-art solvers by decreasing the need of hand-crafting heuristics and hyper-parameter tuning. First advancements using Reinforcement Learning have shown great success in outperforming typical zeroth- and first-order optimization algorithms, especially with respect to generalization capabilities. However, training still is very time consuming. Especially challenging is training models on functions with free parameters. Changing these parameters (that could represent, e.g., conditions in a real world example) affects the underlying objective function. Robust solutions therefore depend on thorough sampling, which tends to be the bottleneck considering time consumption. In this thesis we identified the runtime bottleneck of the Reinforcement Learning Algorithm and were able to decrease runtime drastically by distributing data collection. Additionally, we studied the effects of combining sampling strategies in regards to generalization capabilities of the learned algorithm.
Open Access
Scheduling with uncertainty for Time-Sensitive Networking using robust optimization techniques and integer linear programming
(2024) Bauer, Florian
Application services depend on the network to guarantee reliability, which is critical for safety and correct operation. Time-Sensitive Networking is a technology for reliable real-time communication of time-sensitive applications. While many schedulers exist that provide reliability for wired Time-Sensitive Networks (TSN) with the assumption of deterministic packet delays, scheduling for wireless TSN with uncertain packet delays has received significantly less attention. This work leverages the methodology of Robust Optimization (RO) to propose a robust scheduling approach that ensures provable reliability for both wired and wireless TSN. An uncertainty set defines the range of possible values, ensuring that the schedule remains feasible under all possible realizations within this set. As uncertainty sets are a key component in RO, we introduce methods to compute boxed and polytope uncertainty sets containing possible packet delays based on a set of given reliability requirements. A scheduler is deemed robust if it satisfies the given reliability constraints for all possible packet delays within the computed uncertainty set. Although robustness can be achieved through strict isolation and conservative filtering of packets, we demonstrate that several limitations prevent known robust schedulers from fully exploiting arbitrary uncertainty set shapes. As certain problem instances are unsolvable using simple boxed uncertainty sets, we indicate the need for schedulers that can utilize complex shapes of uncertainty sets rather than boxes. In response to this challenge, we introduce Uncertain No-Wait Packet Scheduling (UNWPS), a scheduler capable of computing robust schedules, and prove that UNWPS is robust against arbitrary upper-bounded boxed and polytope uncertainty sets. We assess the influence of uncertainty sets on the quality of the resulting UNWPS schedules, compare their performances to the performance of other robust scheduling approaches across various exemplary TSN networks and message stream configurations and carry out simulations conducted using the DetCom simulation framework to validate the robustness of UNWPS empirically.
Open Access
Learning free-surface flow with physics-informed neural networks
(2021) Hurler, Marcel
This thesis examines the application of physics-informed neural networks to solve free-surface flow problems modeled with the shallow water equations. Physics-informed neural networks allow training of a surrogate model that resembles the latent solution of an underlying partial differential equation, without using any training data sampled from experiments or numerical simulations. The shallow water equations are an approximation of the Navier stokes equations and serve as a model to many environmental flow problems including dam-breaks, floods, and tsunami propagation. The equations form a non-linear system of hyperbolic partial differential equations that describe the evolution of a fluid's depth and momentum through time. Contrary to other models for free-surface flow, where the exact location of the free surface is only given implicitly as an isosurface and needs reconstruction, here, the depth directly yields its location. One characteristic of the shallow water equations is the formation of steep wavefronts and discontinuities. The thesis examines four state-of-the-art techniques to improve accuracy and training speed and discusses their behavior on three initial value problems. These include the famous idealized dam-break and two depth perturbations, one above a flat and one above varying bathymetry. For each of the scenarios, an inspection of suitable network architectures was considered. Additionally, three different formulations of the physics-informed neural network are presented and tested, where one approach implicitly fulfills the mass conservation and thus eliminates one equation of the system. The results show, that it is possible to train a surrogate model with a relative L^2 error of less than 10^(-4) compared to a solution computed by a high-resolution numerical solver in case of a moderate steepening of wavefronts. A relative error close to 10^(-3) can be achieved for the dam break problem, where the initial conditions are discontinuous, and the solution contains shocks that propagate over time. Additionally, it shows that training with bathymetry is possible and the learned depth approximates the varying underground without any noticeable difference.
Open Access
Scalable traffic engineering heuristics for time-triggered communication in real-time networks
(2026) Geppert, Heiko; Rothermel, Kurt (Prof. Dr. rer. nat. Dr. h. c.)
Distributed safety-critical cyber-physical systems require real-time behavior. This means they must respond not just quickly, but in time, to new situations considering both, the task processing and network communication time. From a networking perspective, meticulous, time-driven traffic planning performed at the frame level is necessary to guarantee low end-to-end delay bounds and low latency. This involves carefully planning transmission operations along each time-critical frame's network path are carefully planned, including precise timing, to limit or even eliminate interference from cross-traffic and ensure timely delivery. Since modern real-time systems can consist of hundreds or thousands of devices - for example, large manufacturing plants or continental-sized power grids - the traffic planning must be highly scalable. Although there are many traffic planning approaches in the literature, there is a lack of very fast heuristics that can handle very large stream sets and networks quickly. This thesis investigates traffic planning heuristics and optimization techniques, focusing on different aspects of the traffic planning domain. The traffic planning consists of novel methods for conflict-graph-based scheduling and new heuristics for very large instances of traffic planning problem. The optimizations include multicast partitioning, which combines the benefits of multicast and unicast traffic plans, and load-balanced stream placement, which generates traffic plans that can accommodate additional streams joining the system later. We created prototype implementations and analyzed their performance in solving the traffic planning problem. Our traffic plans yielded a higher accumulated network throughput or admitted more streams while maintaining computation times ranging from sub-seconds to minutes, even for extremely large-scale problem instances. The traffic planning methods and optimization techniques presented in this thesis can be applied to modern real-time networking technologies, such as Time-Sensitive Networking and TTEthernet.
Open Access
Second-order projection-based mapping methods for coupled multi-physics simulations
(2022) Ariguib, Boshra
Data mapping describes the exchange of variables between different, usually non-matching grids for storing data. As different physics require different physical constraints, so do different simulation require different mesh properties. This makes data mapping a crucial part when coupling single-physics simulations into a multi-physics simulation. However, the tradeoff for the available computationally efficient methods is usually low accuracy order. Such a method is the nearest-neighbor mapping method, which relies on a computationally inexpensive mapping algorithm and shows a first-order accuracy, as it is based on a constant interpolation. A second-order projection-based mapping method nearest-neighbor-gradient aims to improve the accuracy order of the nearest-neighbor mapping, while preserving the low computational costs. This is achieved through the extension of the existing method by considering additional gradient data information and applying a Hermite interpolation, in order to balance out both the computational efficiency and the accuracy of the mapping. In this thesis, we implemented this method by extending the coupling library preCICE, which uses state-of-the-art algorithms for coupling partitioned multi-physics simulations in a black-box manner. We confirmed the theoretical observations of the expected second-order accuracy of the method and we found that the method shows the best convergence order by contrast with the existing mapping methods, including radial basis function mappings. It also performs just as well as the existing projection-based method in terms of computational cost and outperforms the radial basis function mapping in respect of runtime costs.
Open Access
A data plane interface for resource-constrained microcontrollers in time-sensitive networking
(2026) Kupka, Bastian
Time-Sensitive Networking (TSN) enables deterministic Ethernet communication through coordinated transmission control. While there are dedicated TSN switches and network interface cards, they are rarely available on resource-constrained microcontrollers. In particular, transmit-time-based scheduling approaches such as the Earliest TxTime First (ETF) queuing discipline are typically not supported on embedded platforms. This thesis investigates the feasibility of implementing ETF-like transmit-time scheduling on a microcontroller using Zephyr RTOS. The implementation targets the NXP i.MX RT1062, which provides a precise PTP hardware clock and compare interrupt mechanism but lacks dedicated hardware support for traffic shaping. The proposed solution integrates transmit-time scheduling into the Zephyr networking stack by combining hardware-triggered interrupts with software-based buffer management inside the network driver. Scheduled frames are prepared in advance and transmitted using a PTP compare event to trigger the transmission routine close to the target time, followed by a short busy-wait phase to improve precision. Best-effort traffic is handled through a driver-level guard band to avoid interference with scheduled transmissions. Experimental evaluation shows that, for periodic traffic with fixed inter-packet gaps, the implementation achieves a bounded transmission window of approximately 3 µs. For variable inter-packet gaps, a larger timing spread is observed. The results demonstrate that ETF-like transmit-time scheduling can be realized on a low-cost microcontroller for certain traffic patterns by leveraging existing PTP hardware features.
Open Access
Dynamic workload balancing for heterogeneous systems
(2020) Strack, Alexander
During the last two decades, GPUs developed into powerful and massively parallel processors. That rose the attention of scientist who started using GPUs for large scale scientific computing, e.g. simulations. However, the architecture of GPUs is different from CPUs. Furthermore, graphic processors have their now fast access memory. Computing in a heterogeneous system consisting of a CPU and multiple GPUs has various challenges. In this work, we focus on how to distribute the load among the different components. We consider an iterative load that can be redistributed after each iteration. The goal of our scheduling methods is to minimise the computation time of the next iteration by estimating the performance of each component. After a short introduction to load balancing, we specify the iterative workload scenario and differentiate it from the typical task-based scenario often found in the literature. Then, we show the basics of GPU programming with the help of NVIDIAs CUDA API. Furthermore, we introduce the different kernels we use for our test and derive multiple schedulers. Our dynamic schedulers use the time each component took to compute its assigned workload in the last iteration as a basis of the performance estimation. After investigating the influence of previous run-time data on the scheduling decisions, we turn our attention towards the properties of the workloads and therefore compare different types of memory management.
Open Access
Enabling multi-tenant scalable IoT platforms
(2020) Ismaiel, Muhammad
Internet of Things (IoT) platforms have multi-layer architectures that facilitate the provisioning, automation of connected devices, and monitoring. These platforms simplify development because they solve a lot of the problems and complexities in building an IoT application. IoT platforms are typically central components with many heterogeneous users, which leads to the effect that these platforms are essential for IoT scenarios and must not become a single point of failure. Furthermore, since many users access these platforms, scalability and multi-tenancy are crucial. The scalability of IoT platforms makes user applications stable and more comfortable to extend. In contrast, the multi-tenancy trait allows multiple tenants to access user applications at the same time. In this thesis, we examine every software stack layer of an IoT platform and explain different methods to make each layer scalable. Enabling multi-tenancy at the application layer and attaining scalability of message brokers, databases, the middleware as a whole, network layer, and device layer are the primary tasks for enabling multi-tenant scalable IoT platforms. To provide a solution to these tasks, we go through the research work already done in the area of scalability and multi-tenancy. We address multiple solutions for each task and also provide the best suitable solution for each task at the corresponding layer of an IoT platform.
Open Access
About the design changes required for enabling ECM systems to exploit cloud technology
(2020) Shao, Gang
Since the late 1980s, Enterprise Content Management Systems (ECM systems) have been used to store, manage, distribute all kinds of documents, media content, and information in enterprises. ECM systems also enable enterprises to integrate their business processes with contents, employing corporate information lifecycle and governance as well as automation of contents processing. The ever-changing business models and increasing demands have pushed ECM systems to evolve into a very active content repository with expectations such as high availability, high scalability, high customizability. These expectations soon became a costly financial burden for enterprises. The on-going hype around cloud computing has raised attention with its claims on improved manageability, less maintenance, and cost-effectiveness. Embracing the cloud might be a good solution for the next high-performance ECM system at an affordable price. To achieve such a goal, the designs of ECM systems must be changed before deployment into the cloud. Thus, this thesis aims to analyze the architecture design of legacy ECM systems, determine its shortcomings, and propose design changes required for embracing cloud technologies. The main proposal to design changes are i) decomposing an ECM system to its constituent components, ii) containerizing those components and create standard images, iii) decoupling the physical link between the data storage device from the applications container by utilizing docker volumes in dedicated persistent data containers instead, iv) utilizing software-defined network infrastructure where possible. These design changes then were tested with a proof-of-concept prototype, where an ECM product was successfully deployed and tested using Docker in a cloud environment backed by OpenStack.
Open Access
Orchestrating data governance workloads as stateful services in cloud environments using Kubernetes Operator Framework
(2022) Wang, Xiaomin
Data is becoming the core corporate asset that will determine the business’s success. As a result, it is critical for governing enterprise data. Previously, the Enterprise Content Management (ECM) system was employed by many companies to manage and process their enterprise data, which is a monolithic data governance application. As the ECM system is typically deployed on bare metal or at most in a virtualized IT infrastructure, it lacks the flexibility to support Continuous Integration (CI) and Continuous Delivery (CD) cost-effectively. Cloud computing has gained popularity as a powerful platform for application deployment, owing to perceived benefits such as elasticity to fluctuating load and reduced operational costs as compared to running in traditional data centers. Therefore, it is promising to migrate the legacy ECM system into the cloud. The goal of this thesis is to orchestrate stateful database workloads in Kubernetes that are typical for ECM systems. For our concept verification, we included a comparison and analysis between traditional and comparable cloud native Relational Database Management System (RDBMS) using IBM DB2, PostgreSQL, CockRoachDB and Google Spanner. We proposed an implementation of the Monitor-Analyze-Plan-Execute (MAPE) concept using the Kubernetes Operator framework. With our prototype implementation, we proved that the Kubernetes operator is able to deploy a cluster for DB2 consisting of a read/write primary and up to three read-only members. Various experiments carried out on the prototype have evidenced its High Availability (HA), Disaster Recovery (DR) features as well as read scalability.

Universität Stuttgart

Browse

Filters

Settings

Sort By

Results per page

Search Results