USNO Prototype Software Correlator: Status Report - CIRA

cira.ivec.org

USNO Prototype Software Correlator: Status Report - CIRA

USNO Prototype Software Correlator: Status Report

Outline

USNO/WACO Background

• UPSC

• Recent Developments

• Future Plans

• Outstanding Issues

• e-VLBI with VLBA

DiFX Workshop, Socorro, NM

Dave Boboltz, Nicole Geiger, Alan Fey,

Ralph Gaume, Kerry Kingham & Dave Hall


USNO’s Role in the Geodetic Community

USNO is the International Earth Rotation and Reference

Systems Service (IERS) Rapid Service/Prediction Center for

Earth Orientation.

USNO provides daily, weekly, and long-term data products.

– Weekly EOP updates (Bulletin A) are issued on Thursdays.

USNO has several functions within the International VLBI

Service for Astrometry and Geodesy (IVS).

Correlator Operations Center

– VLBI Data Analysis Center.

DiFX Workshop, Socorro, NM


DiFX Workshop, Socorro, NM

Geodetic VLBI and the IVS

• Geodetic VLBI is coordinated through the IVS.

• http://ivscc.gsfc.nasa.gov/


DiFX Workshop, Socorro, NM

WAshington COrrelator (WACO)

• Current MarkIV hardware

correlator was designed

and constructed by MIT

Haystack Observatory

and is operated in

cooperation with NASA.

• 1 of 4 IVS correlators

– WACO, Haystack,

Bonn, Japan.

• All correlators are heavily

subscribed.


• Daily “Intensive” experiments

DiFX Workshop, Socorro, NM

WACO Workload

– Used to measure variable rotation rate of the earth (UT1-UTC).

– 4 days/week: single-baseline, 1-hour experiment.

– 1 day/week: 3 station, 1-hour experiment.

• Weekly R4 experiments

– Used to measure EOP (polar motion, nutation, precession).

– 1 per week: 5-8 stations, 24-hr experiment.

• Other experiments

– 1 per month: 2-4 stations, 24-hr CRF experiments.

– 4 to 6 per year: 24-hr “R1” experiments backing up Bonn.

– Occasionally: other 24-hr experiments (APSG, CONT, etc).


USNO Proto-type Software Correlator (UPSC)

• Heterogeneous cluster.

– 1 PC with 8 CPUs.

– 4 PCs with 2 CPUs each

• 2 TB hard drive storage.

• 1 Gb/s ethernet switch.

• Uses “Rocks” for cluster

management.

• 64-bit Red Hat Enterprise

Linux operating system.

• DiFX 1.5.4 installed.

• Successful correlation

using disk files.

DiFX Workshop, Socorro, NM


DiFX Workshop, Socorro, NM

Recent Developments

• Acquired two Mark5 B+ units for use with the UPSC.

– Want to be able to correlate geodetic data from the units.

• Difficulties:

– 32-bit Mark5 units and 64-bit UPSC Rocks cluster

compatibility issues.

– Stripped-down Debian operating system on Mark5 units.

– Communications with NRAO temporarily halted.

• Solutions:

– Pirated a new 24-processor Dell PowerEdge T610 system

intended for geodetic VLBI data reduction (UPSC2).

– Installed 32-bit Red Hat Enterprise Linux on UPSC2.

– Wiped Debian from both Mark5 units and installed Red Hat

Enterprise Linux.


• 1 Gb/s switch

between Mark5 units

and UPSC2.

• SDK8.2 installed on

Mark5 units with Red

Hat Linux.

• DiFX 1.5.4 installed on

UPSC2.

• Mark5 units use NFS

mounted DiFX

directory on UPSC2.

DiFX Workshop, Socorro, NM

Current UPSC2 Set-up


DiFX Workshop, Socorro, NM

Testing: Geodetic Intensive Experiments

• Correlated single baseline

intensive off of Mark5 units.

– Intensive i10231 (KK-WZ).

– Data from both stations in

Mark5A (MkIV) format.

• Fringes detected.

• Tried three station intensive

e10266 (KK-NY-SV) and ran

into some trouble.

– SV recorded Mark5B format.

– Vex file format (freqs. and

tracks incorrect).

Plot file version 1 created 12-NOV-2010 13:04:26

1803+784 I10231 BOTH.UVDATA.1

Freq = 2.2130 GHz, Bw = 4.000 MH No calibration applied and no bandpass applied

200

100

0

-100

3.0

2.5

2.0

1.5

1.0

0.5

KK - WZ 1 - 2

IF 1(RR) IF 2(RR) IF 3(RR) IF 4(RR) IF 5(RR) IF 6(RR) IF 7(RR) IF 8(RR) IF 9(RR) IF 10(RR) IF 11(RR) IF 12(RR) IF 13(RR) IF 14(RR) IF 15(RR) IF 16(RR)

0.0

0 5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 150

5 15

Channels

Lower frame: Milli Ampl Jy Top frame: Phas deg

Vector averaged cross-power spectrum Baseline: KK (01) - WZ (02)

Timerange: 00/18:30:01 to 00/18:30:39


DiFX Workshop, Socorro, NM

Testing: 5 Station R4 Experiment

• 1 hour worth of typical 24-hr R4 geodesy experiment.

• Chosen because of mixed Mark5A/Mark5B formatted data.

• 5 stations (BADARY, KOKEE, MATERA, YEBES40, ZELENCHK).

– 2 stations recorded in Mark5A format (KK, MA).

– 3 stations recorded in Mark5B format (BD, YS, ZC).

• Data from disk files and two Mark5 units.

– 3 stations from disk files.

– 2 stations streamed from Mark5 units.

• Fringes on all baselines.

– after determining the track and frequency set-ups from hardware

correlator info.


0.8

0.4

0.0

DiFX Workshop, Socorro, NM

Testing: 5 Station R4 Experiment

Plot file version 1 created 28-OCT-2010 10:11:51

1357+769 R4450MIX.UVDATA.3

Freq = 2.2260 GHz, Bw = 8.000 MH No calibration applied and no bandpass applied

200

1.20

BD - KK 1 - 2

1.0

0.0

200

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

100

-100

2.0

BD - YS 1 - 4

400

200

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

-200 KK - MA 2 - 3

200

0

1.2 BD - MA 1 - 3

0.8

0.4

0.0

1.0

0.0

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

0

-200 BD - ZC 1 - 5

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

200

800 0

KK - YS 2 - 4

IF 1(RR) IF 2(RR) IF 3(RR) IF 4(RR) IF 5(RR) IF 6(RR) IF 7(RR) IF 8(RR) IF 9(RR) IF 10(RR) IF 11(RR) IF 12(RR) IF 13(RR) IF 14(RR) IF 15(RR) IF 16(RR) IF 1(RR) IF 2(RR) IF 3(RR) IF 4(RR) IF 5(RR) IF 6(RR) IF 7(RR) IF 8(RR) IF 9(RR) IF 10(RR) IF 11(RR) IF 12(RR) IF 13(RR) IF 14(RR) IF 15(RR) IF 16(RR)

0

0

0 150

150

150

150

150

150

150

150

150

150

150

150

150

150

150

15

0 150

150

150

150

150

150

150

150

150

150

150

150

150

150

150

15

Channels

Channels

Lower frame: Milli Ampl Jy Top frame: Phas deg

Vector averaged cross-power spectrum Several baselines displayed

Timerange: 00/21:57:30 to 00/22:04:50

400


DiFX Workshop, Socorro, NM

Testing: 5 Station R4 Experiment

100

-100

0.8

0.4

0.0

100

-100

1.0

0.0

Plot file version 2 created 28-OCT-2010 10:11:51

1357+769 R4450MIX.UVDATA.3

Freq = 2.2260 GHz, Bw = 8.000 MH No calibration applied and no bandpass applied

KK - ZC 2 - 5

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

MA - ZC 3 - 5

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

Lower frame: Milli Ampl Jy Top frame: Phas deg

Vector averaged cross-power spectrum Several baselines displayed

Timerange: 00/21:57:30 to 00/22:04:50

0

-200

800 MA - YS 3 - 4

400

0

IF 1(RR) IF 2(RR) IF 5(RR) IF 8(RR) IF 11(RR) IF 15(RR)

200

0

2.0 YS - ZC 4 - 5

1.0

IF 1(RR) IF 2(RR) IF 3(RR) IF 4(RR) IF 5(RR) IF 6(RR) IF 7(RR) IF 8(RR) IF 9(RR) IF 10(RR) IF 11(RR) IF 12(RR) IF 13(RR) IF 14(RR) IF 15(RR) IF 16(RR)

0.0

0 150

150

150

150

150

150

150

150

150

150

150

150

150

150

150

15

Channels


DiFX Workshop, Socorro, NM

Future Plans

• Near-term (3-6 months) plans:

– Move to DiFX 2.0 and continue testing.

– Move to more traditional geodetic post-processing path

(difx2mark4).

– Streamline pre-processing path.

– Finalize requirements for software correlator and consider

hardware options.

• Longer-term (3-12 months) plans:

– Settle on final correlator configuration and purchase equipment.

– Develop “operator friendly” GUI for correlator.

– Begin side-by-side comparisons with hardware correlator.


DiFX Workshop, Socorro, NM

Outstanding Issues

• Project currently under-manned.

– All those involved have operational duties.

– Kingham and Hall: hardware correlator operations and e-VLBI.

– Fey, Boboltz and Geiger: VLBI data analysis, EOP, CRF.

• Recent departure of R. Ojha to Fermi NASA-GSFC.

– A major blow to effort.

– Looking for a replacement (spread the word).

Correlator load projected to increase.

– VLBI 2010 antennas coming online.

– Daily e-VLBI intensives from VLBA starting in 2012

– Dual correlation on WACO and UPSC


DiFX Workshop, Socorro, NM

Why Use e-VLBI for UT1-UTC?

• Reducing data latency from 2.25 days to 6 hours results in:

– Factor of 5 reduction in UT1-UTC prediction uncertainty.

– 40% reduction UT1-UTC prediction errors 7 days out.


DiFX Workshop, Socorro, NM

e-VLBI at the VLBA

• Goal: Stream VLBI data over high-speed internet connections

to reduce latency due to recording and shipping disk-packs

• Two important applications of near-real-time VLBI

– Variable rotation rate of the Earth (UT1-UTC)

• UT1-UTC observations performed 1-hr each day

• Important component for GPS accuracy

– Spacecraft navigation

• Phase-referencing of spacecraft to nearby quasar

• Requires CRF

USNO interested in e-VLBI at VLBA.

– US-controlled assets.

– Continuity of Operations (COOP) requirements.


DiFX Workshop, Socorro, NM

UT1-UTC Testing With the VLBA

• With in conjunction with NRAO,

USNO began a series of

“Pseudo-intensive” experiments

to measure UT1-UTC.

• TC015

– 5 stations (HN, LA, MK, PT, SC)

– 13 experiments

– Optimized for MK-SC baseline

• TB014

– 3 stations (MK, LA, PT)

– 5 experiments

– Optimized for MK-PT baseline


Results from Pseudo-Intensive Experiments

• Despite shorter baselines,

VLBA measurements meet

future requirements for UT1-

UTC.

• MK-PT baseline selected for

e-VLBI.

– Fiber to PT antenna already

in place (VLA-PT config.).

– MK gives long east-west

baseline for UT1-UTC.

DiFX Workshop, Socorro, NM


DiFX Workshop, Socorro, NM

Status of e-VLBI using the VLBA

• Contracts in place to connect the Pie Town and Mauna Kea

VLBA stations with optical fiber in 2011.

– $380K contract with AUI/NRAO.

• Light the fiber to PT.

• MK/PT equipment.

– $376K contract with Univ. of Hawaii.

• MKOCN to VLBA site.

USNO-NSF-NRAO MOU signed.

USNO to contribute $1M/yr beginning in 2012.

– Access to PT - MK baseline for 1.5 hr/day.

• Back-up antenna (without e-VLBI) if needed.

– Data to be streamed back to USNO via Internet2/DREN.

More magazines by this user
Similar magazines