Compression of time-vectors in W7-X archived measurements

Wendelstein 7-X is, as a stellerator, in principle capable of very long operation. The Control and Data-acquisition (codac) systems around w7x have, therefore, been designed for continuous operation. One aspect of this design is the recording of the absolute time of every measurement. Whereby for ev...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Fusion engineering and design 2018-04, Vol.129, p.326-329
Hauptverfasser: Krom, J.G., Daher, R., Bluhm, T., Dumke, S., Grahl, M., Grün, M., Hennig, C., Holtz, A., Laqua, H., Lewerentz, M., Riemann, H., Spring, A., Werner, A.
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Wendelstein 7-X is, as a stellerator, in principle capable of very long operation. The Control and Data-acquisition (codac) systems around w7x have, therefore, been designed for continuous operation. One aspect of this design is the recording of the absolute time of every measurement. Whereby for every relevant event, such as every sample-clock tick of every adc, the time is recorded as a 64 bit integer value. On the one hand, this approach has several advantages. On the other hand, it generates much more data to process and archive. In extreme cases it can lead to a multiplication of the data amounts. Compression of these time-stamps data can be expected to save storage and network resources. However, even though most of these time-stamps have a high degree of predictability, the normal, de-facto standard, compression approaches (zip, flac, etc.) perform rather poorly on this type of data. Thus, inspired by well-known audio compression approaches, we developed a specialised loss-less compression algorithm, aimed at fairly constantly incrementing 64 bit long-integers. Experiments on data collected in the past demonstrate very large compression factors, whilst retaining all details in the timing data.
ISSN:0920-3796
1873-7196
DOI:10.1016/j.fusengdes.2018.01.026