Why Is E-VLBI Important

e-VLBI in the U.S.: Planning for the future

NRAO/Haystack Planning group – Draft 4
2007 March 9

What is VLBI and why is it important?

Very-Long-Baseline Interferometry (VLBI) is used by radio astronomers as the most powerful technique: for studying objects in the universe at ultra-high resolution,for measuring earth motions in space with ultra-high accuracy, and for providing precise sky-position navigation of deep-space explorers:

For astronomical studies: VLBI allows images of distant radio sources to be made with angular resolution of tens of microarcseconds, far better than any optical images. As such, the structure and dynamical evolution of some of the most distant objects in the Universe can be studied in unprecedented detail.

For Earth science studies: VLBI provides direct measurements of the Earth’s dynamic orientation in space in the near-inertial reference frame provided by distant quasars, which in turn provides information ranging from plate tectonic motion to mass interactions at the core-mantle boundary, as well as earth-rotation measurements critical to civilian and military navigation.

For deep-space navigation: U.S., European, and Japanese space programs all use VLBI in various ways for high-precision navigation of deep-space craft on the 2-dimensional plane of the sky. Fastturnaround is often required to make critical decisions regarding course-correction maneuvers.

VLBI combines data simultaneously acquired from a global array of20 or more radio telescopes to create a single coherent instrument, as illustrated at right for a simple 2-element VLBI array. Traditionally, VLBI data are collected on magnetic media (magnetic disks)which are shipped to a central site for correlation processing. This laborious and expensive data-collection and transport process, requiring a multi-Petabyte (Petabtye = 1015 bytes)global media pool, now has the possibility of being replaced by modern global high-speed networks (dubbed ‘e-VLBI’), opening important new capabilities,potential scientific returns, and lower costs.

Why is e-VLBI important?

The advantages for scientific productivity and technical operations of e-VLBI over traditional VLBI are:

1. Higher sensitivity: For the majority of VLBI observations, sensitivity increases as the square root of the recorded data rate. Typical recorded VLBI data rates today are 1 gigabit/sec/station (Gbps/station). Short periods of observations at several Gbps are possible but, due to media costs,are uneconomical for extended periods or for extension to many Gbps. The potential to extend e-VLBI to many-Gbps data rates in the future will allow an increase in observation sensitivity well beyond what ispossible today. Furthermore, such high data rates will be sustainable for long periods of time. The only alternatives to higher data rates for increased sensitivity are larger antennas and/or quieter receivers. Larger antennas are, of course, hugely expensive; while many receivers are already very close to theoretical noise limits. For astronomical observations, the benefits of higher sensitivity are obvious. For geodetic observations, the number and distribution of reference sources improves dramatically as sensitivity improves. For spacecraft navigation, which is based on phase referencing with respect to nearby calibrators, systematic errors can often be reduced by being able to select a calibration source near to the spacecraft.

For radio astronomy, the recently formed High-Sensitivity Array (HSA), consisting of the VLBA + phased-VLA + GBT + Arecibo, is among the most sensitive VLBI arrays in the world. (Further information about the component instruments is presented in a later section.) A major increase of data rates for the HSA, to 10Gbps or more, enabled by high-speed networks would open an entirely new class of accessible observations.

1. Targets of opportunity: For a full-time astronomical VLBI instrument, targets of opportunity, which arise primarily from transient astronomical events, present a major opportunity for unique scientific discovery. As the only dedicated VLBI observing instrument in the world, the VLBA (described in a later section) has a strong competitive advantage through a quick, flexible response to targets of opportunity. However, without rapid access to at least preliminary results from such observations, the opportunity to refine or re-direct observing configurations for increased scientific payback is impossible. Using e-VLBI can reduce the time to obtain results from weeks to hours or even, potentially, to minutes, giving scientists rapid feedback. This is especially important for rapidly evolving events, such as extragalactic supernova, gamma-ray-burst events, and other transient phenomena increasingly being observed in the Universe.
2. Lower costs: Ultimately, e-VLBI will eliminate the need for a multi-million-dollar multi-Petabyte global pool of magnetic media. With expeditious planning and implementation of a full-time e-VLBI network, we believe that the benefit-to-cost ratio will be quite high.
3. Quick diagnostics and tests: The verification of the proper operation of VLBI equipment at part-time observing stations is notoriously difficult since correlation must be done before results can be obtained. Unfortunately, this characteristic has too often led to bad or poor observations and the waste of expensive telescope time. With e-VLBI, proper operation of the entire telescope array canbe verified in a matter of minutes to hours. This is particularly important when new equipment is brought on-line, or when equipment that has been repaired needs to be re-verified for service.

Examples ofe-VLBI success

The feasibility of real-time global e-VLBI has now been demonstrated many times using national and international research networks. The work at Haystack Observatory, supported initially by DARPA in 2001-2 and most recently by NSF (ANI-0230759 and ANI-0335266), has been at the forefront of high-speed global eVLBI development. For example, at the Super Computer 2005 meeting in November 2005, data from three telescopes in the U.S. and Europewere streamed at 512Mbps in real-time to theMark 4 VLBI Correlator at Haystack Observatory and the results displayed in real-time on the conference floor. Also demonstrated at the same time were national and international dynamically switched optical pathsin collaboration with the NSF/EIN-supported DRAGON project (ANI-0335266) and the Internet2 HOPI project.

With support from NSF, NRAO has established an 80-km experimental fiber connection to allow the VLBA station at Pie Town, NM, to be used as an extension of the VLA for certain observations.

As a reflection of e-VLBI success and future promise, e-VLBI was recently honored with an Internet2 IDEA award, which recognizes applied advanced networking at its best and hold the promise to increase the impact of next-generation networks around the world. The award was receivedjointly byAlan Whitney of MIT Haystack Observatory, Arpad Szomoru of JIVE, The Netherlands, Yasuhiro Koyama of NICT, Japan, and Hisao Uose of NTT Laboratories, Japan, on behalf of their institutions.

The success of several high-profile demonstrations, plus an increasing level of routine global near-real-time e-VLBI data transfers, provides evidence of both the utility and promise of a fully developed global eVLBI system.

Internationale-VLBI activities

Almost by definition, e-VLBI is a strongly international activity, and development work is international in scope. Japan has been active in e-VLBI development over dedicated networks since the mid-1990s and now participates very actively in global e-VLBI development. The Joint Institute for VLBI in Europe (JIVE) has recently received a 3-yr ~$20M grant from the EC Research Infrastructure Grid Program to connect 16 telescopes in Europe for real-time e-VLBI observations at 1Gbps using a dedicated optical wavelength to each telescope to bring data back to a correlator in Dwingeloo, The Netherlands. In addition, JIVE has also recently received a ~$5M grant from the EC for development of Grid technologies for e-VLBI, including distributed software correlator development. The Australians have recently announced support for connecting all of the country’s major radio telescopes with 10Gbps connections. And the Chinese are aggressively connecting their telescopes to high-speed global networks. Within about three years, most of the major radio telescopes outside the U.S. are expected to be connected to the global grid of high-speed R&E networks at speeds of at least 1 Gbps.

e-VLBI development collaborations with the networking community

By nature, e-VLBI lends itself to collaborative developments with networking research and applications, and such collaborations have already allowed significant development work on the e-VLBI technique. e-VLBI development work at Haystack was first supported in 2001-2 by DARPA, followed byNSF support[1]for development of protocol and transmission techniques specifically tailored for e-VLBI. A prime example of active collaboration with the networking development community is the DRAGON (‘Dynamic Resource Allocation via GMPLS Optical Networks’) project[2], also working with the Internet2 HOPI project, to develop dynamically-switched on-demand dedicated light paths for high-speed data streams.

Broadinternational e-VLBI collaborations have been established between the U.S., Europe, Japan and Australia both for demonstrations and for on-going routine data transfers between connected telescopes and correlators. Many TeraBits/month of data are successfully transferred in this fashion. For telescopes outside the U.S., this activity is increasing dramatically. Within the U.S. it is severely limited by lack of high-speed connectivity to major radio telescopes.

In 2002, Haystack Observatory hosted the first international e-VLBI conference, which was attended by 80 VLBI and network practitioners from around the world, and which formed the basis for an annual e-VLBI conference circulating amongthe U.S., Europe, Japan and Australia. In fall 2006 Haystack Observatory again hosted this international meeting. These meetings have been an important forum for bringing together scientists and network specialists to discuss techniques, applications and tests.

NRAO is currently developing two major interferometer systems, the Atacama Large Millimeter Array (ALMA) in Chile, and the Expanded Very Large Array (EVLA). Although in both cases, the baselines are substantially shorter (~20Km) than what is usually considered “e-VLBI”, the techniques used are similar. The EVLA instrument is already transferring data at 120Gbps from individual antennas to a central correlator. The expertise gained by NRAO in these developmentsis almost all transferable to the development and implementation of high-speed e-VLBI network.

The current e-VLBI situation in the U.S.

The U.S. hosts approximately 15 radio telescopes that are regularly used for VLBI. Ten of these,spread over US territory from St. Croix to Hawaii,form the Very Long Baseline Array (VLBA), operated by the National Radio Astronomy Observatory (NRAO). The VLBA is the only dedicated VLBI observing instrument in the worldand has a rich history of scientific discoveries. A dedicated correlator, located at the NRAO Array Operations Center (AOC) in Socorro, NM, is also part of the VLBA. Plans for replacing t VLBA’s digital data path by new, significantly wider-band equipment, including an e-VLBI-ready correlator, are under development.

Although not dedicated to VLBI, several other U.S. radio telescopes are included regularly: the 300m diameter Arecibo telescope in Puerto Rico; the 100m diameter Green Bank Telescope (GBT) in Green Bank, WV; and the Very Large Array (VLA; a 27-antenna interferometer with effective single-dish diameter 135m, currently being upgraded to the EVLA described above) west of Socorro, NM. Millimeter-wavelength antennas in Hawaii and Arizona have been used for high-frequency VLBI observations, with the expected addition of the CARMA facility in Californiain the near future. The 18m Westford radio telescope at Haystack Observatory in Massachusetts and the 20m telescope at Kokee Park, Hawaii, are used regularly for geodetic VLBI observations supported by NASA. All of these telescopes are good candidates for an e-VLBI network.

Currentlythe Westford telescope and a small experimental telescope at NASA/GSFC in Marylandare the only telescopes connected to the national/international networks at Gbps speeds. These two telescopes currently form the primary e-VLBI testbed in the U.S and are connected at 2Gbps to the global high-speed network. The VLA is connected to the NRAO AOC in Socorro, NM, via an OC-3 link, which is in turn connected to Internet2 via an OC-3 link shared with New Mexico Tech. On Hawaii, the telescopes at the Mauna Kea summit share a 1Gbps link, and Arecibo maintains a connection limited to ~20Mbps, which is only marginally useful for e-VLBI.

What needs to be done to keep the U.S. competitive in the e-VLBI arena?

While Europe, Japan, China and Australia are moving aggressively towards extensive national and global eVLBI capability, the U.S. is nearly stalled. Leadership in e-VLBI is in imminent danger of being ceded by default to the rest of the world. To correct this imbalance, the U.S. must also make a commitment to a healthy and aggressive eVLBI development program if it is to remain a viable partner in this global scientific enterprise.

We suggest the follows steps as guidelines to developing a healthy and robust e-VLBI capability within the U.S.:

Continue e-VLBI development and testing

e-VLBI has already benefited greatly from significant development work funded by NSF, as earlier indicated. However, still more optimization and improved efficiency are needed to keep up with both with higher data rates as well as changing and improving network technology. New capabilities built into network technology are often only accessible by development efforts designed specifically to support them; the dynamic optical switching capabilities developed collaboratively with the NSF-supported DRAGON program are a prime example. Additionally, e-VLBI data occupies a special niche in high-speed networking as it is somewhat loss tolerant and can therefore sometimes use network bandwidth that is less than optimum, as demonstrated by the EGAE development program at Haystack Observatory.

Further software research and development strategies for e-VLBI include integration of other emerging network services such as: 1) national storage depots (REDDNET program) and networking software tools to aid in the identification of inter-domain networking problems, 2) aggressive transport protocols for end-to-end dedicated light paths that take into consideration special VLBI data characteristics, and 3)investigation of multi-cast techniques for possible use with distributed correlation systems of the future.

In order for this research and development to proceed, work inboth laboratory settings and in actual e-VLBI test experiments are necessary. For the most part, networks made up of a small array of working VLBI stations will be sufficient, operating on a non-interfering basis with normal VLBI usage of stations. Of course, collaborative testing on a national and global scale with institutions such as NRAO, NAIC, EVN/JIVE, Japan and Australia is critical to ensure proper behavior across a wide range of network topologies and scales.

Estimated costs: ~0.5 FTE/yr plus ~$30K/yr materials & services

Begin real-time e-VLBI testing using VLA Telescope

The existing connection of the VLA to Internet2 is an OC-3 (~150Mbps) link shared with New Mexico Tech. This connection is sufficient to begin e-VLBI testing using the VLA (or a single antenna of the VLA) and the Westford antenna in Massachusetts for real-time transfer of data to the Mark 4 correlator at Haystack Observatory. The first tests will be at 32Mbps/station and then up to a probable maximum of ~64Mbps/station. These first steps would establish the feasibility of real-time e-VLBI with an NRAO antenna and establish a baseline from which to proceed in the future.

Estimated costs: ~4 person-months plus ~$20K network charges

Investigate and connect VLBA AOCto high-speed network

Establishing a high-speed network connection from the NRAO Array Operations Center (AOC)in Socorro, NM, to anational high-speed R&Econnection point is essential in order to bring data from VLBA stations, as well as potentially from other U.S. and international sites, into the AOCfor processing. The most likely connection point is Albuquerque asInternet2 maintains a 10Gbps connection point in Albuquerque. National Lambda Rail (NLR) also has a presence in Albuquerquewhich might be similarly utilized. Collaboration with both Internet2 and NLR should be initiated to determine the most cost-effectiveway to connect the AOC to the high-speed national network grid.

Ultimately, as the data rate from each of the VLBA stations increases to ~10Gbps or higher, the connection to the AOCcorrelator must be capable of supporting this data-transfer rate from at least the 10 stations.

Estimated costs:
Investigation: ~2 person-months
Actual connection costs: Will be determined by the study

Investigate and connect 6 western continental VLBA sites to high-speed networks

The six western continental VLBA sites (FortDavis, Los Alamos, PieTown, KittPeak, OwensValley, Brewster) are of interest to NASA for near-real-time deep-space navigation and form a good core group of stations for initial high-speed network connection. Cooperation by both NASA and NSF/AST in making these connections could benefit both. In particular, the resources of Internet2 expertise should be tapped to identify the most cost-effective method of connection to the VLBA correlator, which might be located at the AOC, at the VLA site, or elsewhere. The Internet2 organization maintains an extensive database of fiber routes and connections that could prove invaluable in ascertaining the most effective connections. NASA has a division which could interact with the fiber owners once this information is known. Another option would be to involve a subsidiary of Internet2, called WaveCo, which has been actively procuring low-cost long-term access to dark fiber all across the U.S., with the expectation that such access will help to minimize costs for future connections. Of course, there will be inevitable costs of providing ‘last mile’ access to each station, but the one-time ‘last mile’ costs, must be viewed in the perspective of the potential ultimate benefits. Close cooperation between NRAO and NASA should help to minimize costs for both institutions. A carefully conducted costing study involving networking experts such as Internet2 and others should be conducted as a first step. Some preliminary information regarding possible fiber connections at the VLBA sites was gathered by Craig Walker a couple of years ago ( which forms a initial reference, but clearly much further investigation is warranted.