1 Kashima Space Research Center
Communications Research Laboratory
893-1 Hirai, Kashima, Ibaraki 314-0012, Japan
2Communications Research Laboratory
4-2-1 Nukui-kita, Koganei, Tokyo 184-8795, Japan
3Kety Co.Ltd.
4CosmoResearch Co. Ltd.
1. Introduction
Key Stone Project(KSP) real-time VLBI observations started in June of 1997
by using Asynchronous data Transfer Mode(ATM) network. Real-time VLBI
correlation processing software "RKATS" has developed for fully automatic
real-time VLBI data processing. Overview of RKATS was reviewed in TDC News No.9
p.15. In this report,
"Dynamic Clock Adjustment" and "Automatic Fringe Search"
functions are introduced as key functions for RKATS.
2. Differences between Tape-based VLBI and Real-time VLBI
The main differences of correlation processing between tape-based VLBI and
real-time VLBI are listed in Table 1.
Table 1. Differences between tape-base VLBI and real-time VLBI.
.
TAPE-BASE
REAL-TIME
Data transportation and replay
Human operation is inevitable at work of Magnetic tape transportation and
tape mount on recorders
Real-time data transfer by ATM enables automatic operation.
Log information
Observation information and Clock parameter is provided by log file.
Information is collected by computer network just before the observation.
Trouble recovery
Trouble at data processing can be recovered by re-processing of
recorded data.
Stop of data processing due to troubles leads to data loss directly.
Rapidity of output
Analysis result comes out 1-2 days after observation.
Analysis result comes out just 10-20 minutes after observation.
Real-time VLBI is advanced from the standpoint of automatic operation and
rapidity of output, but interruption of correlation processing due to any
problems leads to direct loss of data. Therefore, non-stop operability is
strongly requested of real-time correlation processing software.
Except for hardware errors and software hang ups, most probable cause of data
loss would be wrong clock parameters. Wrong clock parameters will move fringes
out of the lag window of the correlator, and such data cannot be used for
baseline analysis.
Also, fringe monitoring and adjustment of clock parameters by an operator is
difficult, because VLBI experiments may start at midnight and the operators are
not supposed to be familiar with VLBI. Therefore we implemented "Dynamic Clock
Adjustment(DCA)" and "Automatic Fringe Search(AFS)" functions in RKATS. By
using these functions, fully automatic real-time daily VLBI observation has
realized.
3. Automatic Fringe Search and Dynamic Clock Adjustment
Figure 1. Performance test of AFS(left) and DCA(right) function. Clock
parameter of Kashima is fixed as reference. For the performance test, clock
parameters of Koganei, Miura, and Tateyama are artificially biased by
10 usec,
-10 usec, and -5 usec respectively. At first, correlation has
started in "Normal Mode" but fringes were not detected. Then RKATS changed the
mode to "Fringe Search Mode" and detected fringes automatically. After that
clock parameter was adjusted, the mode was returned back to "Normal Mode". In
the "Normal Mode" operation, DCA function adjusts the clock parameter and keeps
the fringes at the center of lag window.
Figure 1 shows the clock parameters are adjusted by AFS and they are kept almost
within +/-0.1 usec by DCA function. AFS and DCA works as follows:
AFS: In the case that the station clock information is wrong and fringes
are shifted outside of the lag window of the correlator (+/-1 usec
at 16Mbps/ch,
+/-4 usec at 4Mbps/ch), correlation mode is changed from "Normal Mode"
to "Fringe Search Mode". Then "Fringe Search Mode" arranges all lags of 16
channels sequentially and makes the lag window wider (+/-16 usec at
16Mbps/ch, +/-64 usec at 4Mbps/ch) to get fringes. After fringes are
detected, the station clock parameter is adjusted and a return is made to
"Normal Mode".
DCA: Position of fringes within the lag window is monitored by the
software. Clock parameters of each station are adjusted so that fringes
are kept at the center of the lag window.
Even though the clock offset is changed by about a few hundred nanoseconds,
no differences in the analysis results were observed.
Data processed in "Fringe Search Mode (FSM)" is unusable for baseline analysis.
The frequency of transition to FSM must be as small as possible.
For stabilization of AFS and DCA, RKATS has several parameters.
They are listed in Table 2.
By optimizing these parameters empirically, RKATS can be tuned to operate as
reliably as possible.
Table 2. Parameter for AFS and DCA.
Parameter name
Meanings
Fringe Search Mode transition threshold
If ratio of fringe detection (#detected/#total) become lower than this
value, correlation mode is changed to FSM.
Averaging Number
Averaging number for fringe detection ratio.
SNR threshold
SNR threshold to judge fringe detection.
Source list for DCA
radio source list, on which fringe is expected
to be detected certainly.
4. Summary
AFS and DCA are introduced as key functions of RKATS. Under the combination of
RKATS and automated KSP system, baseline vectors among four stations are being
measured automatically daily.
