Interview Gert Kruithof

– Research and Development

Gert Kruithof

“Many results have been achieved in the past two years and we have new challenges ahead!”

Accelerate science
The Research & Development department develops and builds innovative instruments for all of our telescopes. Besides that, we elaborate new technologies for the observatory facilities of the future. With our instruments, we help to push the boundaries of science. Our starting point every time again is the dialogue with the astronomers. We listen to their questions, problems and wishes and try to come up with relevant answers and solutions. The background of the 60 men and women of our R&D department is quite diverse. The team includes among others designers of mechanics and electronics, engineers, information experts and mathematicians. Their expertise is very valuable. The products we manufacture are always innovative and groundbreaking. That’s something I am very proud of!

Cooperation
Our R&D department has been divided in six competence and support groups that work closely together in various national and international projects:

  • Computing Group
  • Digital & Embedded Signal Processing Group
  • Mechanical Group
  • Radio Group
  • System Design & Integration Group
  • Technical Support Group

Cooperation is key in our relations with astronomy institutes around the world and companies like IBM. International partnerships are essential to realize innovative projects nowadays. Therefor we work closely together with several trading companies, both regionally and nationally. Our partners indicate, that we challenge them to innovate again and again, with the additional advantage that they increase their competitiveness.

Projects

APERture Tile In Focus (APERTIF)
The APERTIF project aims to place new ultra-sensitive receivers in the current Westerbork telescopes. These new receivers have a much larger field of view than the old receivers and can observe the sky approximately 20 times faster. This enables the Westerbork telescope to carry out large-scale surveys and to detect and map up to one million galaxies in the years ahead. The design of the new APERTIF receivers was finished in 2014. We are very satisfied that we have succeeded to meet all requirements for sensitivity and system temperature, in spite of an interfering area. In October 2014 the complete APERTIF design was reviewed by a panel of international experts. After this Critical Design Review (CDR) the upgrade of the Westerbork telescopes could start. First, we have shut down most of the telescope dishes in 2015. We expect the new receivers to produce their first results in 2016.

AARTFAAC

AARTFAAC stands for Amsterdam-ASTRON Radio Transients Facility and Analysis Centre. It is an extension of the LOFAR telescope, meant to detect radio flashes from space (transients). For this purpose 12 LOFAR stations have been provided with additional hardware (in particular UniBoards) to tap raw data of all antennas for a limited bandwith. All these antenna signals are correlated with each other by GPU machines (graphics processing unit) in Groningen (576 dual-pol receivers, 10 GPU’s in one machine). In 2015 the last six stations were expanded with AARTFAAC hardware and an extra GPU machine was was purchased to increase the correlator capacity. We also purchased equipment for the transient pipeline, to be able to detect transients in real-time.

DOME
The DOME project, supported by grants from the Dutch Ministry of Economic Affairs and the Province of Drenthe, focuses on three main areas: Green Computing, Data & Streaming and Nano-photonics. It’s a close cooperation between ASTRON and the IBM Center for Exascale Technology. Data centers all over the world are confronted with an ever increasing demand of web and cloud services, like data analysis and search engines. To meet these challenges and to reduce the energy consumption at the same time, we have developed an extreme energy efficient and high density micro server within the DOME project. The micro server has the size of a cell phone and acts like a full-fledged server. The new micro server concept including a prototype were officially presented to Dutch SME companies, during the Micro Server Event in Dwingeloo, on April 14th, 2014. The reduction of energy consumption in large computer systems is crucial for the future Square Kilometre Array (SKA) radio telescope and is also very important for data centers in general.

In 2015 we presented the water-cooled micro server at various conferences and events (IEEE SC, ISSCC and CEBIT). Three of our DOME partners (Philips Drachten, TU Eindhoven and Sensor City Assen) tested the application of the micro-server in 2015, in order to assess the competitiveness of the computing platform. Within Dome we have additionally developed a tool that studies the estimated power consumption and processing requirements of algorithms and computer architecture. This tool is being used in multiple SKA consortia now. All areas of research within DOME show promising results. They can lead to further reduction of the required computing power. Besides the power issues we have developed a new method to make pictures of the sky very efficiently. As a next step, the algorithm will be applied and verified on our LOFAR data. To let SME companies benefit from the new DOME technologies, we organized a Dome Users Platform Match Making Event in Dwingeloo, on October 29th, 2014.

LOFAR
Although the LOFAR-telescope has been operational since 2013, it is still quite a challenge to process the generated data into high-quality maps of the sky. The R&D Computing Group has worked out a strongly improved method for calibration and imaging the data. They successfully completed the first Calibration & Imaging Tiger Team. The development and implementation of the so called StefCal method has made the callibration of data in some cases up to 50 times faster. For imaging, modern methods like Multiscale Multifrequency Clean were adapted and we have started the investigation and implementation of the Image Domain Gridder, that can accelerate the slowest part of making maps.

In 2014 we completed the LOFAR station in Hamburg Germany and signed the contracts for building three LOFAR stations in Poland. The three stations were completed in 2015. Finally, we have talked with Ireland about the possibilities to build a LOFAR station there, too.

201400047_Lofar_poster-1.indd

SKA

SKA, the Square Kilometre Array, has entered a new phase. About 15 years of research have been completed and the project proceeded into a pre-production development phase. This phase should eventually lead to qualified designs, on the basis of which the production and the deployment can begin, scheduled for launch in 2018.

To guide the various developments in the right direction, the SKA Office has divided the work into 11 subprojects. The work in each subproject is carried out by international consortia. ASTRON participates in seven of these consortia and is in charge in two of them (Aperture Array consortia LFAA en MFAA).

In the Low Frequency Aperture Array (LFAA), ASTRON works together with teams from Italy, the UK and Australia, at the development of antenna arrays for the SKA1 Low telescope. In many aspects the SKA1 Low is a ‘bigger version of LOFAR’. Much of the expertise that ASTRON has gained in LOFAR is relevant for this project. This also goes for the Mid Frequency Aperture Array (MFAA). Besides both mentioned AA consortia, ASTRON has made a significant contribution to the consortium for signal processing (CSP) and data processing (SDP). The aim is, that a large part of the consortia will complete the preliminary design review in 2015 and will start to work at the detailed design review, to be finished in 2017.

SKA Subprojects

SKA – LFAA / AAVS1
The Low-Frequency Aperture Array (LFAA) consists of a large number of antennas, each equipped with a low-noise amplifier and advanced radio-over-fiber connection. The antennas are placed in the field, in a specific configuration, combined into stations. Each station consists of 256 antenna elements. At station level, the signal is digitalized and beamforming takes place. For the control of all elements in the system, monitoring and control software is used. In 2015 the consortium has continued to work on the planned roll-out (in 2017) of the Aperture Array Verification System 1 (AAVS1), the first LFAA demonstration system. AAVS1 will consist of 400 antennas, distributed over four stations. With this system we can verify our requirements and start up production tests. We will also verify our cost models.

AAVS1 test station Cambridge

SKA – CSP
In 2015 we started a collaboration with the Commonwealth Scientific and Industrial Research Organisation (CSIRO), to work on a joint draft for the correlator and the beamformer for the SKA-low array, as part of the Central Signal Processor consortium. In the correlator, signals of all LFAA stations come together to be correlated, as the name suggests. For the pulsar applications, the signals are combined in a beamformer. ASTRON’s activities take place across the full spectrum, including system engineering, project management, hardware design and firmware design tasks. One of the interests of ASTRON is to bring in the technology-independent firmware design methodology and expand it along with CSIRO. In 2015 we mainly worked on the system design, the definition of the work and the preparation of the delta PDR (Preliminary Design Review). 

SKA – SDP
The R&D department makes a major contribution to the SKA Science Data Processor (SDP) consortium. This consortium designs the calculation platform that converts the detected signals into data products, suitable for scientific use, for the filing system and for the system that makes data available to other parties. The main result of 2015 was, that we established the first version of the SDP architecture.

SKA – MFAA
ASTRON is in charge of the consortium that develops the Mid-Frequency Aperture Arrays (MFAA) for the second phase of the SKA. Millions of small antennas are combined to one enormous telescope. ASTRON develops the system design, the antennas and the analogue electronics. We do this in close cooperation with international partners, the industry and the DOME-project. Due to the vast number of antennas, a sustainable design is essential, with low power consumption and low manufacturing and maintenance costs. To achieve this, new technologies such as 3D-MID and bio based materials, get a lot of attention. This will result in a new modular antenna tile in 2016.

RF over Fiber (RFoF)
In 2014, our R&D department has worked at the development of optical analogue signal technology for the transport of low-frequency radio signals (RFoF), generated by ASTRON’s photonics technology programme. Across multiple DOME workstreams we have worked on the development of technology for both the SKA central processor systems as for the SKA/LFAA and SKA/MFAA receiver systems.

The RFoF technology enables the SKA/LFAA consortium to apply a low-cost system architecture, in which the LFAA antenna signals are sent straight to the central LFAA signal processor, without significant signal processing. To obtain the desired low cost level for LFAA, ASTRON develops ultra-low cost RFoF transmitter and receiver modules. We do this in close collaboration with Dutch companies.

Besides the short-term RFoF technology development for LFAA, DOME/MEMPHIS also contributes to the medium term technology development for SKA/MFAA. In the first place, we are working on the development of middle-frequency, integrated electronic/photonic technology for signal transmission and processing. Secondly, we work on systems with integrated photonic components (multi-wavelength lasers and photonic beamformers). We do this in close collaboration with Dutch institutes and companies and with IBM Research in Zurich.

Uniboard 2
UniBoard 2 is the sequel of the UniBoard project. It aims to develop a generic data processing platform, that can be used for future radio astronomy applications. We use the next generation technology in the board. The total capacity of data that can be processed through the board is immense (approximately equal to the amount of internet data that enters the Netherlands via the Amsterdam Internet Exchange). In 2015 we produced and tested the first prototype of the board and wrote the firmware to test all interfaces.

Together with our engineers of the Radio Observatory, we took the new correlator COBALT into production in 2014. Cooperation across the boundaries of R&D is considered a core value of our department. Many results have been achieved in the past two years and we have new challenges ahead! I am convinced that we will be able to meet all those challenges, together with our team of dedicated and motivated staff, the wealth of knowledge and experience that we have and all of our partners.