realfast@VLA
Author/Presenters
Event Type
Workshop
Data Analytics
Data management
SIGHPC Workshop
Visualization
TimeSunday, November 12th11:30am - 11:35am
Location605
DescriptionWe describe a system being deployed at the National Radio Astronomy
Observatory's Karl G. Jansky Very Large Array (VLA) to commensally identify
and record millisecond timescale astrophysical transient events in real time.
This system distributes a high time resolution data stream to a dedicated fast
transient detection system while allowing processing of a primary observation
to continue with the typical (lower) time resolution data. This form of dual
time resolution, commensal observing is enabled by the vys protocol,
implemented with existing VLA computing infrastructure. The fast transient
detection system performs real-time analysis in situ to detect events
of interest and record relatively short duration data "cut-outs" of those
events. By selectively recording high time resolution data, provided by
vys at rates of up to 1.4 GB/s, realfast will reduce the recorded data volume by an estimated factor
of up to 1000. This makes it possible to search for transients commensally in
a high data rate stream over the thousands of hours needed to find the rarest
events.
Observatory's Karl G. Jansky Very Large Array (VLA) to commensally identify
and record millisecond timescale astrophysical transient events in real time.
This system distributes a high time resolution data stream to a dedicated fast
transient detection system while allowing processing of a primary observation
to continue with the typical (lower) time resolution data. This form of dual
time resolution, commensal observing is enabled by the vys protocol,
implemented with existing VLA computing infrastructure. The fast transient
detection system performs real-time analysis in situ to detect events
of interest and record relatively short duration data "cut-outs" of those
events. By selectively recording high time resolution data, provided by
vys at rates of up to 1.4 GB/s, realfast will reduce the recorded data volume by an estimated factor
of up to 1000. This makes it possible to search for transients commensally in
a high data rate stream over the thousands of hours needed to find the rarest
events.
Author/Presenters
