KIT, Eggenstein-Leopoldshafen, Germany
Integrating control and high-speed data processing is a fundamental requirement to operate a beam line efficiently and improve user's beam time experience. Implementing such control environments for data intensive applications at synchrotrons has been difficult because of vendor-specific device access protocols and distributed components. Although TANGO addresses the distributed nature of experiment instrumentation, standardized APIs that provide uniform device access, process control and data analysis are still missing. Concert is a Python-based framework for device control and messaging. It implements these programming interfaces and provides a simple but powerful user interface. Our system exploits the asynchronous nature of device accesses and performs low-latency on-line data analysis using GPU-based data processing. We will use Concert to conduct experiments to adjust experimental conditions using on-line data analysis, e.g. during radiographic and tomographic experiments. Concert's process control mechanisms and the UFO processing framework* will allow us to control the process under study and the measuring procedure depending on image dynamics.
* Vogelgesang, Chilingaryan, Rolo, Kopmann: “UFO: A Scalable GPU-based Image Processing Framework for On-line Monitoring”
KIT, Eggenstein-Leopoldshafen, Germany
The ultrafast tomography system built up at the ANKA Synchrotron Light Source at KIT makes possible the study of moving biological objects with high temporal and spatial resolution. The resulting amounts of data are challenging in terms of reconstruction algorithm, automatic processing software and computing. The standard operated reconstruction method yields limited quality of reconstruction images due to much fewer projections obtained from the ultrafast tomography. Thus an algebraic reconstruction technique based on a more precise forward transform model and compressive sampling theory is investigated. It results in high quality images, but is computationally very intensive. For near real–time reconstruction, an automatic workflow is started after data ingest, processing a full volume data in parallel using the Hadoop cluster at the Large Scale Data Facility (LSDF) to reduce the computing time greatly. It will not only provide better reconstruction results but also higher data analysis efficiency to users. This study contributes to the construction of the fast tomography system at ANKA and will enhance its application in the fields of chemistry, biology and new materials.
