Huib%20Jan%20van%20Langevelde

1
e-VLBIa real-time telescope larger than Europe

• Huib Jan van Langevelde
• Joint Institute for VLBI in Europe, Dwingeloo NL

2
Team effort/many collaborations

• EVN European VLBI Network
• Consortium of (European) Telescopes
• Arecibo, Puerto Rico, Cambridge (UK), Effelsberg
  (D) Jodrell Bank (UK), Medicina (I), Metsahovi
  (FI), Noto (I), Onsala (S), Shanghai (CN), Torun
  (PL), Urumqi (CN), Westerbork (NL), Yebes (ES)
• Joint Institute for VLBI in Europe
• Institute established in the Netherlands
• Funded by NWO, ASTRON, STFC (UK), INAF (I),
  ICN-IG (ES), OSO (S), CAS (CN), CNRS (F), MPG
  (D)
• EXPReS EXpress PRoduction e-VLBI Service
• EC project funded by DG-INFSOC _at_ 3.9 M
• Partners DANTE, SURFNET, various NRENs and most
  radio-telescopes

3
On behalf of EXPReS partners
And especially all the people working for the
demo
4
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

5
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

6
Radio-astronomy

• Radio waves with ? of 0.7mm to 90cm
• Compared to optical light 400 700 nm
• Can be detected and amplified with antenna

• radio emission from hot gas between the stars
• Super bright emission from vicinity of black holes

• The Galaxy at 320 MHz

7
Relatively young science

• Reber (1937)

• Jansky (1931)

8
Some principles
Lovell Telescope D 76 m, Manchester

• Larger telescopes detect weaker signals
• Sensitivity ? D2
• Larger telescopes resolve more details
• Resolution ? ?/D
• ? wavelength
• D Diameter

• Arecibo D 305 m, Puerto Rico

9

• Radio emission from astrophysical plasmas can be
  detected against sky noise with telescopes larger
  than few meters
• But interesting resolution at cm wavelengths
  requires big telescopes
• Interferometer measures Fourier components of the
  sky brightness

connected Radio-Interferometry e.g. Westerbork
JIVE ASTRON
Long baselines can observe fantastic detail in
radio source
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
Principle of VLBI

• Record same frequency band simultaneously at N
  telescopes
• Use best possible local clocks and frequency
  standard
• Sample and digitize and record (on magnetic
  medium)
• Find all ½N(N-1) correlation coefficients at
  correlator
• Compute back image from thousands of these
  measurements

weak radio source
correlator
recorder
maser clock
14
Results in extraordinary high resolution (for
bright objects)
High enough to see things move at cosmological
distance
15
(No Transcript)
16
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

17
VLBI digital processing

• To reach the faint end of the universe
• Need many big telescopes
• And as much frequency space as you can get
• Bandwidth!
• Typical VLBI uses 64 MHz bandwidth in 2 pols
• Nyquist sampling
• 2 x 2pol x 32 MHz 128 M samples/s
• Noisy data 2 bits/sample recover 86 SNR
• 256 Mb/s, more is preferred
• Bits are not sacred
• Some losses are tolerable

radio sources in the Hubble deep field require
several days of integration (Garrett et al., 2000)
18
Based on Mk5 recording system

• Recording on Linux PC platform
• Harddisk recorder based system
• Parallel writing on 8 disk system
• Ship disk-packs around the world

19
Now turn to e-VLBI!

• PC based recording
• Also allows Internet transmission
• Upgrade EVN to e-EVN
• Started with a pilot in 2004
• And was boosted with EXPReS
• Retrofit correlator to work real-time
• Help solve last mile problem at telescopes
• Work closely with NRENs on robust connectivity
• Push to 1024 Mb/s limit
• Bring in the big telescopes
• And start the revolution in radio-astronomy
  culture

20
Includes academic networks
Across Europe GÉANT2
Established lightpaths to most telescopes,
providing undisturbed connectivity
The Netherlands SURFnet6
21
EXPReS network upgrade
22
Remarkable progress

• 6 telescopes regularly on line, interesting for
  science
• Correlator operations optimal for direct results
  and feedback
• Connectivity reached impressive reliability
• Started with TCP, but obviously not the optimal
  protocol for e-VLBI
• Now dedicated light-paths to most telescopes
• UDP protocols implemented for optimal streaming

Size of balloon set by number of telescopes
participating, height by station sustained
bit-rate
23
Connectivity progress Effelsberg

• On-line on 1/4/08
• Through GÉANT2 connection
• Thanks to local loop funded by MPG
• Reached almost 1 Gb/s in first test
• Also tried exercise Mk5B
• One of the few testbeds for this
• And got Mk5B-Mk5A fringes on noise
• Extremely important for science impact
• Boosts the sensitivity of the e-VLBI array

24
Connectivity progress towards 1Gbps

• Gbps connectivity
• Progress with data dropping
• First fringes on 27/12/07
• Channel selection coming soon
• Also longest user experiments so far
• 6 x 512 Mbps
• 2 user experiments
• gt 12h uninterrupted

25
Tools for feedback and streamlining

• Public Status
• Rapid feedback
• Streamlined processing
• Adaptive observing

26
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

27
(No Transcript)
28
EXPReS connectivity progress

• Long-haul high-bandwidth data transport
• TCP on old linux kernels clearly inadequate
• Circuit TCP perfoms well (TCP with UDP-like
  behaviour, without congestion control)
• UDP better, but needs modification of Mark5A
  control code
• And can be hostile to other users on open network
• Stability of original code a serious issue
• Has led to complete re-write of subset of Mark5
  control code

29
and to China
TEIN2
Miyun
Urumqi
Seshan
Kunmin
30
Link to China, Australia
31
Link to China, Australia
32
Todays demo

• To connect telescopes in 4 continents
• Europe Westerbork (NL), Effelsberg (DE), Onsala
  (SE), Medicina (IT)
• North America Arecibo, Puerto Rico
• Centennial, AMPATH, AtlanticWave, NGIX,
  Internet2, StarLight, GEANT2, Surfnet
• Africa Hartebeesthoek, South Africa
• TBD
• TIGO, Chile
• Transportable Integrated Geodetic Observatory
• Reuna, RedCLARA, GEANT2, Surfnet

33
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

34
More telescopes detailed images
35
e-VLBI, an operational facility

• e-VLBI offered for Target-of-Opportunities
• From start of project in 2006 for brave
  astronomers

The X-ray binary CygX-3 was observed after a
major outburst in May 2006 using e-VLBI.
36
Reduces time to publication

• Data processing took 1-2 weeks with first images
  within 48 hours
• Publication took less than 2 months

37
Why e-VLBI is exciting for astronomy?

• Rapid response for rapid variability
• Fast response to requests
• Immediate analysis of data, adapt observing
  parameters
• Coordination with current and future
  observatories
• Immediate feedback
• More robust data
• Fewer consumables, logistics
• Constantly available VLBI network
• Monitoring for example astrometry
• Spacecraft tracking
• Growth path for more bandwidth
• More sensitive astronomy

38
More spacecraft tracking

• LAPLACE and TANDEM
• accepted by ESA for study for 2015-2025
• Earlier projects may include Bepi-Colombo

• LAPLACE a mission to Jupiter and Europa
• VLBI experiments with Europa landers/orbiter
• Radio astronomy experiments Jovian orbiter

• TANDEM Titan and Enceladus mission
• VLBI experiments with Titan probes/balloons
• Radio astronomy exps Enceladus orbiter

39
(No Transcript)
40
(No Transcript)
41
e-VLBI science

• Is now booming in 2008!
• Had 6 e-VLBI runs this year
• Including 3 epoch ToO observation
• Can say it was on Cyg X-3
• Dutch PhD students involved in several
  experiments
• Also open for spectral line observations (maser
  flares, astrometry)

42
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

43
Doing RD for future e-VLBI

• Current data acquisition limited to 1 Gbps
• Could make use of 10 Gbps connections
• Or even 64 Gbps
• Development of data acquisition system
• Based on so-called IBoB
• FPGA based boards with standard connections
• Test to interface to UK eMERLIN system
• to Onsala (SE) telescope
• And Metsahovi (FI)
• Will also need samplers
• And wider IF systems

44
Next Generation Correlator

• Developed a science case in EVN2015
• Multiple data streams of 10 - 100 Gbps
• 32 stations, or at least more than 16
• Multiple bit representation
• High bandwidth at higher frequency
• Multiple beam correlation
• Next-generation EVN correlator will be enormous
• 100 times larger as current one
• Based on custom VSI chips
• As FX software correlator, 150 200 Tflops
• Current MarkIV correlator (implemented as FX) 2
  Tflops
• Similar requirements as some (other) radio
  astronomy projects

The current EVN Mk4 correlator is Based on 32x32
custom chips
45
Distributed correlation

• Solution could be distributed correlator
• Complies with distributed nature EVN
• Pilot project on Grid
• Looking for P-ops regime

46
Grid enabled correlator

• Workflow developed in Poznan
• Follows workflow of virtual lab
• Complexity associated with VLBI schedule
• And data formats
• Real challenge time critical
• Grid broker
• Must keep up with data rate
• Resource balancing
• Allocate rather than batch
• Benchmarking on NL
• Starplane cluster
• Collaboration UvA

47
Software correlator core

• Implemented on standard CPUs
• Using MPI for parallel processing
• Portable for Grid deployments
• Benchmarking promising
• 4 station correlation at 256 Mbits/s
• 30-40 speed of real-time
• on 30 quad-core machines on DAS-3
• Comparison with hardware correlator
• Better SNR by floating point accuracy
• Towards first image

Comparison with hardware correlator
48
Software correlator in action

• Now operational for pre-session telescope
  checking
• Web site where telescopes can check progress

49
Outline

• What is radio-astronomy? What is VLBI?
• Progress with e-VLBI, role of Networks
• And global connectivity, introducing todays
  demo.
• Science results, transient phenomena
• RD efforts distributed correlation on the Grid
• Developments in radio-astronomy, future of e-VLBI

50
Roadmap of future telescopes
LOFAR
SKA and SKA pathfinders
51
Roadmap to the future

• e-VLBI sets the path for future
• Recognized SKA pathfinder
• Path to more bandwidth
• More sensitivity
• Unique science case for VLBI
• In the Northern hemisphere
• On a global scale
• And can use a lot of the same technology as SKA
• Looking at the next generation correlator
• FPGA based 100 times more powerful
• Power consumption issues
• Interesting option to add more telescopes
• Capable at high frequencies

52
What we achieved learned

• e-VLBI is an operational, robust facility in
  Europe
• Enabling unique science from the start
• Reaching competitive level with all telescopes
  and 1 Gb/s for any science application
• Requires light-paths for optimal performance
• Had to overcome some network quirks for special
  applications
• But when it runs it really flies on Geant!
• Can do even intercontinental VLBI
• Were able to overcome problems with delays
• e-VLBI is the grow path for more sensitivity
• And global VLBI!

53
Wish list...

• Light paths, probably dynamically allocated
• To accommodate distributed correlation
• And around the globe in some uniform manner
• Must continue close collaboration with
  NRENs/GEANT
• Will use 10Gb/s to make fellow astronomers forget
  conventional VLBI...
• but not quite ready for that
• Considering some improvements in VLBI
  architecture
• Buffering data at telescopes and correlator for
  robustness
• Maybe involve more supercomputing/GRID in the
  future?
• Must address a number of things at astronomy side
• Decide on correlator architecture for next
  generation
• Keep a focus on Global e-VLBI, incl NRAO antennas
  in the US
• Develop common ground with E-LOFAR
• Continue to explore technological synergy with SKA

54
(No Transcript)