History & Evolution of the GCN (BACODINE) System

The table below lists the initial genisis and the major improvements made to the system. For a listing of the changes after 1997, please referr to What's New.

Dec 92 The idea of using the real-time data from CGRO-BATSE. (Real-time, because the on-board tape recorders failed, and now all the data HAS to be transmitted to the ground through TDRSS in real-time or else it is lost forever. The basic architecture of using a computer to capture the telemetry stream, scan it for the BATSE portion, look for count-rate increases, calculate a burst direction based on the count-rate ratios in the 8 detectors, and distributing those burst coordinates to a ground-based, rapidly moving, automated, optical telescope to make follow-up observations is formulated in two days. The target instrument was GTOTE (Gamma-ray To Optical Transient Experiemnt). It was called BACODINE (BAtse COordinates DIstribution NEtwork) in these early days.)
May 93 The first cut of the program is running. This basic program just captured and processed the BATSE-portion of the CGRO telemetry stream.
Jun 93 The first GRB is detected in BATSE telemetry stream & the position is calculated. Yea!
Jul 93 The Dedicated Phone distribution method is available.
Aug 93 A version of the program with a single Internet socket connection port is available. The GROCSE instrument connects and is the first robotic instrument on-line.
Dec 93 The e-mail distribution method is added.
Jan 94 A 2-socket-connection-port version of the program is running.
Jan 94 Catch slow-rising bursts; program efficiency: 40 --> 49%. By allowing for an on-the-ground-processing search for a more-than-N-sigma-increase in the count rate beyond the initial BATSE Trigger interval, the efficiency for producing usable burst locations increases.
Feb 94 Multiple sockets version now running -- now as many as 50 socket connections can run simultaneously.
May 94 Larger packet & expanded e-mail format
Jul 94 MAXBC Notices added; program efficiency: 49 --> 60%. This enhancement makes use of a second data product in the BATSE telemetry stream to produce usable burst locations. The MAXBC rates are derived from on-board processing for count-rate increases at the 64-msec sampling interval for up to 10 minutes after the initial trigger.
Jul 94 Created the (long) Pager Notices
Sep 94 The computer system and programs are monitored 24/7 by the people in the PACOR "Bubble". Since the BACODINE system is not in Building 2 (it's in buiilding 23), it helps to have in-room people looking for program crashes and network failures, and to take appropriate action sooner than I can respond from building 2 or from home.
Sep 94 Alpha-numeric pager capability
Feb 95 Trigger Type Identification (identify GRBs from non-GRB triggers in BATSE). A series of ground-processing rules using spectral hardness ratios, detector anisotrpy ratios, and solar proximity allow the trigger to be classified GRB vs non-GRB with greater than 98% confidence.
Mar 95 Daily Reports on the status of the socket connection generated for Socket Sites
Jun 95 The generation of the BACO/IPN Notices is semi-automated; now all but the last human-based sanity check is automated. The human checks to see that the location is valid, and then the Notice is distributed.
Jan 96 Added code so that Original Notices are generated for Burst-OverWrite triggers. This is a another few percent increase inthe efficiency of the system.
Aug 96 Added the "short form" of the Pager Notices for those sites using pager services that do not allow more than about 60 characters int he message.
May 97 Added "importing" & distribution of other sources of GRB locations; the name was change: BACODINE --> GCN. Now that more than just BATSE-based GRB locations are being distributed, the name of the system should reflect this more global aspect.
May 97 Added LOCBURST, RXTE-PCA, & ALEXIS Notices. The LOCBURST Notices are positions derived from near-real-time human processing by people on the BATSE team in Huntsville. The RXTE-PCA-detected burst positions are distributed. And ALEXIS extrame-UV transient positions are distributed.
Jul 97 Added the "Subject-only" Notices (for cell-phones and other pager sevices that restrict the message to about 20 characters).
Jul 97 Sites.cfg files can be re-loaded dynamically and by remote control
Sep 97 BATSE GRB web page table created (r-t updated)
Oct 97 BATSE-Final Notices added; these contain the true peak intensity & fluence of the burst by integrating the lightcurve for up to 30 sec after the intial trigger.
Nov 97 Added COMPTEL Notices: GRBs detected and localized by CGRO-COMPTEL.

Details of Some Selected Improvements:

[For the details of all the improvements, see the What's New.]

The Internet socket method delivery speed has been improved by a factor of two. We now have over 5 site-years of connection experience (as of Feb 95). The typical round-trip travel times for connection-status packets between GCN (BACODINE) and the GROCSE,CA, CLFST,UK, and ETC,AZ instruments are 0.7, 0.9, and 1.2 sec, respectively. These are typical times with less than 0.3% of the packets exceeding twice this time.

The e-mail and pager distribution methods have been added. The e-mail method is still relatively fast (5-30 sec) and suitable for sites with non-automated instruments. The alpha-numeric pagers work well for sites without phone or Internet connections. The pager displays the RA,Dec location, the time and the initial intensity of the GRB. The pager companies accept the GCN (BACODINE) notices via an Internet e-mail address and automatically transmit the message to the designated pager unit. The entire process takes 1 to 3 minutes (depending on time-of-day and geography).

BACODINE now can identify non-GRB BATSE triggers (solar flares, electron precipitation events, SAA entries, etc.) and does not notify those sites that have chosen to have non-GRB triggers filtered out. It is 98% accurate.

The GCN (BACODINE) system has increased its "program efficiency" from 40% to 60%. This increase in the program efficiency means that sites now receive 50% more notifications. There are two contributions to this increase in the program's ability to identify and extract enough count rate information to produce a viable GRB location. The first has to do with GRBs that have slow rise times. The BATSE instrument triggers on LAD count rate increases on 3 different time scales (64, 256, and 1024 msec). The real-time telemetry stream only has the 1024 msec sampled data. By searching for up to 10 sec after the initial onset of the BATSE trigger signal, GCN (BACODINE) can now wait until there is a significant enough rate increase to calculate the burst. The other increase in the program efficiency comes from using a different data product in the telemetry stream. The on-board software scans the 64-msec sampled rates in two energy bands (25-50 and 50-300 keV) for 10 minutes after the initial trigger looking for the maximum values. The 16 rates (the so-called MAXBC rates) are transmitted at T+10min. For those triggers that occurred during a telemetry gap (about 15%), a location can still be calculated from these rates and distributed to the sites, albeit at a much longer time delay. However, even with this 10 minute delay it is still useful since some sites have comparable response times.

Return to GCN/TAN homepage.
This file was last modified on 20-Jan-01.