Tan Jerome, Nicholas
PhD thesis, Faculty of Electrical Engineering and Information Technology, Karlsruhe Institute of Technology, 2019.
Exploring large and complex data sets is a crucial factor in a digital library framework. To find a specific data set within a large repository, visualisation can help to validate the content apart from the textual description. However, even with the existing visual tools, the difficulty of large-scale data concerning their size and heterogeneity impedes building visualisation as part of the digital library framework, thus hindering the effectiveness of large-scale data exploration.
The scope of this research focuses on managing Big Data and eventually visualising the core information of the data itself. Specifically, I study three large-scale experiments that feature two Big Data challenges: large data size (Volume) and heterogeneous data (Variety), and provide the final visualisation through the web browser in which the size of the input data has to be reduced while preserving the vital information. Despite the intimidating size, i.e., approximately 30 GB, and the complexity of the data, i.e., about 100 parameters per timestamp, I demonstrated how to provide a comprehensive overview of each data set at an interactive rate where the system response time is less than 1 s—visualising gigabytes of data, and visualising multifaceted data in a single representation. For better data shar- ing, I selected a web-based system which serves as a ubiquitous platform for the domain experts. Being a useful collaborative tool, I also address the shortcomings related to limited bandwidth latency and various client hardware.
In this thesis, I present a design of web-based Big Data visualisation systems based on the data state reference model. Also, I develop frameworks that can process and output multi- dimensional data sets. For any Big Data feature, I propose a standard design guideline that helps domain experts to build their data visualisation. I introduce the use of texture-based images as the primary data object where the images are loaded in the texture memory of the client’s GPU for final visualisation. The visualisation ensures high interactivity since the data resides in the client’s memory. In particular, the interactivity of the system enables domain experts to narrow their search or analysis by using a top-down methodological ap- proach. Also, I provide four use case studies to examine the feasibility of the proposed design concepts: (1) analysing multi-spectral imagery, (2) Doppler wind lidar, (3) ultra- sound computer tomography, and (4) X-ray computer tomography. These case studies show the challenges of dealing with Big Data such as large data size or disperse data sets.
To this end, this dissertation contributes to a better understanding of web-based Big Data visualisation by using the proposed design guideline. I show that domain experts appreciate the WAVE, BORA, and 3D optimal viewpoint finder frameworks as tools to understand and explore their data sets. Mainly, the frameworks help them to build and customise their visualisation system. Although specific customisation is necessary for the different application, the effort is worthwhile, and it helps domain experts to understand their vast amounts of data better. The BORA framework fits perfectly in any time series data repositories where no programming knowledge is required. The WAVE framework serves as a web-based data exploration system. The 3D optimal viewpoint finder framework helps to generate 2D images from 3D data, where the 2D image is based on the 3D scene with optimal view angle. To cope with increasing data rates, a general hierarchical organisation of data is necessary to extract valuable information from data sets.
First assessor: Prof. Dr. M. Weber
Second assessor: Prof. Dr. W. Nahm
Chilingaryan S., Ametova E., Kopmann A., Mirone A.
in Proceedings – 2018 30th International Symposium on Computer Architecture and High Performance Computing, SBAC-PAD 2018 (2019) 158-166, 8645862. DOI:10.1109/CAHPC.2018.8645862
© 2018 IEEE.Synchrotron X-ray imaging is a powerful method to investigate internal structures down to the micro and nanoscopic scale. Fast cameras recording thousands of frames per second allow time-resolved studies with a high temporal resolution. Fast image reconstruction is essential to provide the synchrotron instrumentation with the imaging information required to track and control the process under study. Traditionally Filtered Back Projection algorithm is used for tomographic reconstruction. In this article, we discuss how to implement the algorithm on nowadays GPGPU architectures efficiently. The key is to achieve balanced utilization of available GPU subsystems. We present two highly optimized algorithms to perform back projection on parallel hardware. One is relying on the texture engine to perform reconstruction, while another one utilizes the Core computational units of the GPU. Both methods outperform current state-of-the-art techniques found in the standard reconstructions codes significantly. Finally, we propose a hybrid approach combining both algorithms to better balance load between G PU subsystems. It further boosts the performance by about 30 % on NVIDIA Pascal micro-architecture.
Harbaum T., Balzer M., Weber M., Becker J.
in International System on Chip Conference, 2018-September (2019) 118-123, 8618493. DOI:10.1109/SOCC.2018.8618493
© 2018 IEEE.Modern high-energy physics experiments such as the Compact Muon Solenoid experiment at CERN produce an extraordinary amount of data every 25ns. To handle a data rate of more than 50Tbit/s a multi-level trigger system is required, which reduces the data rate. Due to the increased luminosity after the Phase-II-Upgrade of the LHC, the CMS tracking system has to be redesigned. The current trigger system is unable to handle the resulting amount of data after this upgrade. Because of the latency of a few microseconds the Level 1 Track Trigger has to be implemented in hardware. State-of-the-art pattern recognition filter the incoming data by template matching on ASICs with a content addressable memory architecture. A first implementation on an FPGA, which replaces the content addressable memory of the ASIC, has been developed. This design combines the advantages of a content addressable memory and an efficient utilization of the logics elements of an FPGA. This paper presents an extension of this FPGA design, which is based on the idea of data compression and assemble the stored data to appropriate packages and drastically reduces the required number of write and read cycles. Furthermore, the extended design meets the strong timing constraints, possesses the required properties of the content addressable memory and enabled a compressed storage of an increased amount of data.
Otte F., Farago T., Moosmann J., Hipp A.C., Hammel J.U., Beckmann F.
in AIP Conference Proceedings, 2054 (2019), 060084. DOI:10.1063/1.5084715
© 2019 Author(s). The Helmholtz-Zentrum Geesthacht, Germany, is operating the user experiments for microtomography at the beamlines P05 and P07 using synchrotron radiation produced in the storage ring PETRA III at DESY, Hamburg, Germany. In recent years the software pipeline and sample changing hardware for performing high throughput experiments were developed. To test and optimize the different measurement techniques together with quantification of the quality of different reconstruction algorithms a software framework to simulate experiments was implemented. Results from simulated microtomography experiments using the photon source characteristics of P05 will be shown.
Caselle M., Brundermann E., Dusterer S., Funkner S., Gerth C., Haack D., Kopmann A., Patil M.M., Makowski D., Mielczarek A., Nasse M., Niehues G., Rota L., Steffen B., Wang W., Balzer M.N., Weber M., Muller A.S., Bielawski S.
in Proceedings of SPIE – The International Society for Optical Engineering, 10903 (2019), 1090306. DOI:10.1117/12.2511341
© COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.KALYPSO is a novel detector operating at line rates above 10 Mfps. The detector board holds a silicon or InGaAs linear array sensor with spectral sensitivity ranging from 400 nm to 2600 nm. The sensor is connected to a cutting-edge, custom designed, ASIC readout chip, which is responsible for the remarkable frame rate. The FPGA readout architecture enables continuous data acquisition and processing in real time. This detector is currently employed in many synchrotron facilities for beam diagnostics and for the characterization of self-built Ytterbium-doped fiber laser emitting around 1050 nm with a bandwidth of 40 nm.
Caselle M., Rota L., Kopmann A., Chilingaryan S.A., Mahaveer Patil M., Wang W., Brundermann E., Funkner S., Nasse M., Niehues G., Norbert Balzer M., Weber M., Muller A.S., Bielawski S.
in Proceedings of SPIE – The International Society for Optical Engineering, 10937 (2019), 1093704. DOI:10.1117/12.2508451
© 2019 SPIE.KALYPSO is a novel detector operating at line rates above 10 Mfps. It consists of a detector board connected to FPGA based readout card for real time data processing. The detector board holds a Si or InGaAs linear array sensor, with spectral sensitivity ranging from 400 nm to 2600 nm, which is connected to a custom made front-end ASIC. A FPGA readout framework performs the real time data processing. In this contribution, we present the detector system, the readout electronics and the heterogeneous infrastructure for machine learning processing. The detector is currently in use at several synchrotron facilities for beam diagnostics as well as for single-pulse laser characterizations. Thanks to the shot-to-shot capability over long time scale, new attractive applications are open up for imaging in biological and medical research.
Chilingaryan S., Ametova E., Kopmann A., Mirone A.
in Journal of Real-Time Image Processing (2019). DOI:10.1007/s11554-019-00883-w
© 2019, The Author(s).Back-Projection is the major algorithm in Computed Tomography to reconstruct images from a set of recorded projections. It is used for both fast analytical methods and high-quality iterative techniques. X-ray imaging facilities rely on Back-Projection to reconstruct internal structures in material samples and living organisms with high spatial and temporal resolution. Fast image reconstruction is also essential to track and control processes under study in real-time. In this article, we present efficient implementations of the Back-Projection algorithm for parallel hardware. We survey a range of parallel architectures presented by the major hardware vendors during the last 10 years. Similarities and differences between these architectures are analyzed and we highlight how specific features can be used to enhance the reconstruction performance. In particular, we build a performance model to find hardware hotspots and propose several optimizations to balance the load between texture engine, computational and special function units, as well as different types of memory maximizing the utilization of all GPU subsystems in parallel. We further show that targeting architecture-specific features allows one to boost the performance 2–7 times compared to the current state-of-the-art algorithms used in standard reconstructions codes. The suggested load-balancing approach is not limited to the back-projection but can be used as a general optimization strategy for implementing parallel algorithms.