Future Technology for EVN: A Vision of Advancements in Radio Astronomy

1. Technology and the Future of the EVN Arpad Szomoru

2. Visions of the future • Science vision for the EVN • Including a chapter on technology • 2015 seemed safely remote • At the time... • Exercise is being revisited • Part of the Jumping JIVE project • Presentation by Lindqvist

3. Recommendations in 2007 • Extend the IF bandwidth at the telescopes toat least 1 GHz. Use digital base-band converters (dBBC), with 8-bit sampling (if feasible) and RFI mitigation. Recording systems capable of handling up to 256 Gb/s • Expansion of EVN below 1 GHz, at 6 GHz, 12-15 GHz, 22 GHz, 30 GHz, 43 GHz and 86 GHz • High surface-brightness sensitivity at frequencies below 1 GHz • New broad-band dual-polarization receivers at cm wavelengths, up to 1 GHz bandwidth in C-band and higher. Frequency agility is required between at least a sub-set of important continuum bands, e.g. L, C/X, and K-band • Additional telescopes to improve the EVN2015 sensitivity.

4. Recommendations in 2007 (continued) • Multiple Gb/s connection of the telescopes to the correlator would allow very high-sensitivity and robust real-time operations for extended times, without disk space limitations at the stations. This is especially required for transient observations, and 24-hour coverage by multi-network VLBI observations • Seamless EVN-MERLIN integration as the shorter baseline core of EVN operations can be achieved with a 32 Gbps connection from one EVN telescope into the e-MERLIN system and multiple Gbps connections of e-MERLIN telescopes into the current EVN correlator at JIVE • A new generation EVN correlatorat JIVE with a capability of up to 256 Gb/s per radio telescope for routine network operation with 32 station capacity to accommodate all network related stations. The requirements also accommodate very high spectral resolution as well as time resolution experiments • Further opportunities need to be explored for space science related activities • The development of software tools for new and improved data products, advanced data reduction pipelines, Grid applications, and Virtual Observatory applications, leading to improved astrometry, phase-referencing procedures, and high- and low-frequency calibration procedures

5. Recommendations in 2007 (continued) • EVN to continue to support and enlarge the user community by means of teaching networks, workshops, symposia, and PR activities. Improved data calibration pipelines and VO capabilities will attract more users from outside of VLBI institutes • Maintain a certain fraction of over-subscription in order to assure that the best science is done and a wide range of topics is covered. Improved/simplified procedures may be necessary to deal with an increasing number of triggered and Target of Opportunity projects • More frequent EVN sessions in support of monitoring programs. The EVN should be more flexible toward projects that require coordinated observations with other instruments, and should accommodate rapid response science runs with the possibility of immediate feedback • The EVN needs to develop optimal complementarity for the SKA. The EVN should serve as a technology test bed for SKA • A high priority for spectrum management at all EVN stations and for further development of RFI mitigation techniques is essential • The EVN2015 should ultimately aim for developing a real-time VLBI network on global scales

6. The future in 2007 Telescopes • IF bandwidth of the EVN stations to 1 GHz in the L-band, and to 2-4 GHz in the C-band and higher • Full digital sampling of Ifs (dBBC project) • Higher bit sampling (up to 8 bits) for RFI mitigation • Rapid switching time between frequency bands on timescales of few seconds • Continuous monitoring of telescope performance during the experiments • Rapid data analysis as well as on-the-fly changes in the observing schedule • Transfer of calibration data from the telescopes to the correlator in real-time • Data from co-located WVR radiometers and GPS receivers for the atmospheric and ionospheric phase calibration, respectively • Installation of dual-feed receivers on the smaller dishes. Even better, focal plane arrays • New telescopes in strategic places (e.g. North Africa) • New full EVN members: Latvia (32m), Evpatoria (70m), Simeiz (22m), Miyun (50m), Kunming (40m), Yebes(40m), and the Sardinia telescope (64m).

7. The future in 2007 (continued) Correlator • Correlator capable of processing data from 32 stations in real-time • 16 Gsamples per second at 8-bit sampling. • up to 128 Gbit/s per telescope, or 4 GHz per bandwidth in both polarizations. • Spectral line experiments with16000 channels per correlation product with floating point accuracy • mapping of the full primary beam of a 25-m antenna in standard operations Logistics • e-VLBI observing with ad-hoc global arrays including telescopes with various backend systems will be straightforward Post-processing • Robust and reliable pipeline, providing maps as well as calibrated data. • Data archive with smart selection for data-mining • Virtual Observatory tools to help provide data products • light curves, spectra, pulsar time series, or overlays of images with results from other instruments.

8. And here we are in 2018.... • Previous recommendations were perfectly sensible • And many of them (unfortunately) still are totally relevant • Roadmap was rather ambitious • How to proceed? • Jumping JIVE to the rescue! • Presentation by Lindqvist

9. Current status • New telescopes • Irbene, Yebes, SRT, Kunming fully operational, KVAZAR and KVN networks have joined • Westerbork was lost due to APERTIF • High-bandwidth receiver system under development • BRAND • New backends • DBBC2 in use, early version of DBBC3 available • IAA, ShAO, KVN, JPL, Haystack Obs all have developed backends • Higher bandwidths, increased operational reliability, but compatibility may become an issue • Multi-frequency receivers in KVN (22, 43, 86, 115 GHz) • Partial system at Yebes (22/43 GHz) • New design created, smaller, easier to deploy at other stations?

10. Current status (continued) • Multiple 10G and 100G connections have become standard • At least across Europe • Disk shipping nearly obsolete thanks to e-shipping • Two complete EVN sessions recorded on FlexBuffs • Semi-automatic data transfer • Great simplification of logistics • e-MERLIN has re-joined EVN! • Both recorded and real time • Software correlation has opened up many new features • Flexibility, scalability • Next step, GPUs? Presentation by Phillips • Many improvements in the past years: • Correlation capabilities and data transport • Many new stations • Sensitivity of EVN has lagged • Observing bandwidth limited to 256 MHz, only twice that of 11 years ago...

11. New telescopes, increased sensitivity and fidelity • Newly built or to-be built • FAST 500m (China), QTT 110m (China), NARIT 40m (Thailand), MeerKAT (South Africa) and the proposed 30-40m radio telescope in the United Arab Emirates • Refurbished telecom antennas • Goonhilly 26m (UK), Usuda 64m (Japan), Sao Miguel 32m (Portugal), the Hellenic telescope 30m (Greece), Kuntunse 32m (Ghana), Xi’An 40m (China) and ROT 54m (Armenia) • Geodetic 13.2 m telescopes • NyÅlesund, Wettzell, RAEGE Santa María, RAEGE Gran Canaria • Not straightforward due to limited time availability.

12. What more is needed/wanted? • Enhancement of real-time capacity of the EVN • 4, 8, 32 Gbps (?) • e-MERLIN, KVAZAR for short and intermediate baselines • Sardinia for sensitivity • Correlatorupgrades • More connectivity • More computing power: FPGAs, CPUs, GPUs, all of these? • Sufficient recording capacity to combine real-time and delayed processing: difference between e and recorded vanishing • EVN light? • Has been talked about for many years • Quick response to transient events • Bound to become more important UV coverage EVN + eMerlin at 6cm, 18 stations

13. (Much) wider bandwidths • BRAND now under development • Feed + receiver • 1.5 – 15.5 GHz • DBBC3 • Geodesy: 4 * 1 GHz • Range of 2 – 14 GHz • Investigation into secondary focus options • Will produce implementation document • Including list of suppliers/industrial partners

14. Universal backends • Make VLBI just one of many different observing modes • Switchable within minutes • Separate sampling from rest of processing • As much RF or IF as possible • Maybe after course sampling, store into VDIF packets • Connect to various compute clusters via fast ethernet switch • Several presentations at this workshop

15. RFI • Radio Frequency Interference (RFI): severe global problem • which will worsen in the future • will require action from the EVN to mitigate its effects • major concern for all observatories • may jeopardize the investments to achieve better sensitivity and wider frequency coverage • Continuous monitoring at the EVN stations. Small antennas covering the frequency interval between 1 and 18 GHz and capable of moving in azimuth and elevation would be a necessary investment to identify the source and frequency of the RFI. • The installation of High Temperature Superconductor filters prior to the cryogenic LNAs to avoid the saturation of wide band amplifiers. • An increase of the number of bits to digitize the recorded signal to avoid saturation. Traditionally VLBI has used 2 bits but the presence of strong RFI signals advises the usage of 8 or 10 bits. • Legal measurements to prohibit the usage of the electromagnetic spectrum in the vicinity of the observatories. • A global policy defended by the CRAF. • RFI flagging prior to correlation using advanced algorithms. • Presentation by Baan

16. Software for a global array Scheduling: • Could (should) be far more flexible • Optimiseuse of resources, respond to changing circumstances. • Rapid response to transient events, needs further automatisation • Transparent global scheduling Software: • re-factoring of SCHED • Easierscheduling of different hardware, different versions of firmware • Ongoingworry: Field System • Aging, notunderour control • Depending on the kindness of strangers • VLBI with CASA • Make VLBI data reduction (andscience) more accessibleto new generation of astronomers • Integration into notebooks likeJupyter • Bringcomputeto data • Monitoring of arrays: • Effort at Wettzelltocomeup with “universal” web-based system • Making use of existing monitoring systems

17. Expectations • Five years • Wide-band receivers on at least a sub-set of the EVN. What frequency range will be covered is unclear, and will no doubt partly depend on the performance and price of BRAND, partly on specific science cases. Aiming for maximum sensitivity by using all of the band will of course bring its own set of problems, in terms of availability of media, connectivity, correlation capacity. • At higher frequencies, multi-band receivers like the KVN quasi-optics system on a sub-set of the EVN. • DBBC3 backends (or comparable) at all stations • RFI mitigation at the stations (in-built feature of DBBC3) • Multiple 100G connectivity to JIVE • Inclusion of KVAZAR telescopes in real-time operations • Inclusion of all eMerlin telescopes at high bit rates in both recorded and real-time operations, greatly improving UV coverage on shorter baselines • Inclusion of Ghana, Xi’An, MeerKAT telescopes • Real-time observations using the JIVE UniBoardCorrelator (up to 16 stations at 4 Gbps, or 8 stations at 8 Gbps) • Regular correlation of geodetic observations at JIVE • (semi-) Automated follow-up of transient events by sub-set of EVN stations • EVN – light observations, more regular observing sessions with the EVN stations that are not heavily oversubscribed • Upgraded scheduling and common monitoring tools.

18. And further along • Ten years • “Real” global VLBI, seamless cooperation of different networks • Global VLBI with the SKA • Deployment of “Software Defined Radio” backends, to some degree doing away with the need for different hardware backends for VLBI, pulsar astronomy, etc. Such a system would isolate the digitisation step from the rest of the backend, sampling the RF signal, or IF, and packaging the samples into VDIF packets. These VDIF packets can be stored for later processing or processed on the fly by any suitable piece of equipment, connected via Ethernet. A similar system is currently being considered at the VLBA • True wide-field VLBI, enabling high-resolution blind surveys • Use of solar collectors as radio telescopes • Upgraded software control tools for telescopes and backends.