SKA – The Reference Design Peter Hall SKA International Project Engineer, ISPO Next-Generation Correlators Workshop Groningen, June 28, 2006

PJ Hall, June 2006 ISPO Outline n SKA n SKA Reference Design –Selected antenna technology –Correlator matters (brief) n Project news

PJ Hall, June 2006 ISPO SKA At A Glance n Aperture synthesis radio telescope with 1 km 2 of effective collecting area by 2020 n 1 km 2 ~ 100 x VLA area –Limited gains by reducing receiver noise –Just need more microwave photons! n Frequency range GHz –Large bandwidths (4 GHz), large fields-of-view (50 deg 2 ) n New capabilities: area re-use (“multi-fielding”), RFI mitigation, high dynamic range imaging, …. n Innovative design to reduce cost –€ 1000 per m 2 target is about 0.1 current practice n International funding: ~ € 1 billion n 17-country international consortium n 4 potential sites; ranking in progress Huge data rates & volumes }

2000 Site short-listing ‘1% SKA’ Science ISSC MoAs Science Case published Inter- governmental discussions including site selection First SKA Working Group Initial concept 2000 ‘10% SKA’ Science Feasibility study Full array Build 100% SKA SKA Complete Phase 1 Build 10% SKA Concept exposition Define SKA System SKA Timeline Optimise Design Reference design selected Construct 1% SKA Pathfinders Radio interferometers can be built in stages (Inbuilt risk mitigation)

PJ Hall, June 2006 ISPO SKA: Radio Meets IT (Again) Is SKA a software telescope? Almost!

PJ Hall, June 2006 ISPO Reference Design - Background n Reference Design (RD) –Provides recognizable SKA image –Focuses science and engineering –Forms basis of SKA costing –Is a strong candidate for actual implementation n Result of wide exploration of design space –Original SKA concepts pushed boundaries in key areas: often simultaneously! »Brightness sensitivity, field-of-view, no. FOVs, frequency coverage, … »RD retains major precepts in each frequency range »All SKA concepts had/have much common system design n RD balances innovation and risk –Recognizes need to optimize within selected technology mix –Maps out technology contingencies »Fall-back positions at every decision point

PJ Hall, June 2006 ISPO SKA Reference Design

PJ Hall, June 2006 ISPO SKA Reference Design n A sparse aperture array for GHz ( Low-Band ) –“Era of Recombination” array »Super LOFAR, MWA etc –Multiple independent FOVs, wide FOVs –Low risk n A small dish + “smart feed” for GHz –Radio camera –Dish ~ 10 m diameter –Smart feed  wide response in angle OR frequency »Mid-Band 0.3 – 3 GHz: wide FOV »High-Band 3+ GHz: wide bandwidth –Both low risk + high risk components –Driven in part by need for sensitive, wide FOV telescope a.s.a.p (SKA Phase 1) n An innovation path –Radio “fish eye” lens –Dense aperture array for 0.3 – 1 GHz –Independent FOV capability to 1 GHz –All-sky monitoring capability –High risk (but potentially high return)

PJ Hall, June 2006 ISPO radio “fish-eye lens” Inner core Station Digital radio camera + stations to 3000 km Radio fish-eye lens Reference Design

PJ Hall, June 2006 ISPO SKA – Schematically (About 150 stations) (About 2000 antennas – all correlated)

PJ Hall, June 2006 ISPO Reference Design: Some Technology > 3 GHz: wide-band feed < 0.3 GHz: sparse aperture array 0.3 – 3 GHz: phased array feed Innovation path: dense aperture array Mid-Band High-Band Swinburne/CVA visualization Low-Band

PJ Hall, June 2006 ISPO S mall D ish + P hased A rray F eed Digital beamformer Phased array feed Correlator & further processing Multiple fields 10 m dish cost target: ~ €30k exc. feed FOV expansion factors ~30 may be practical ~  D radian D Terminology: PAF is one type of Focal Plane Array

PJ Hall, June 2006 ISPO PAF Operation Key question: How calibratable are PAFs? D. Hayman, T. Bird, P. Hall

PJ Hall, June 2006 ISPO Reference Design Practicalities n Full frequency range unlikely to be affordable –Possible outcome: (EoR array) + (0.3 – 10 GHz) n Dense AA is least mature technology in cost terms –Balance between AA and SD collecting area will depend on cost and performance demonstration –AA probably has most scientific value as a central collecting area n But SD+PAFs also need rapid demonstration –Cheap dishes and astronomically-capable PAFs are not trivial –Low frequency efficiency is a potential issue n Ultimate contingency if AA, SD+PAF fail: –Super-LOFAR, GHz, plus –“small” dishes (~10 m) + single pixel feed, > 0.3 GHz »7 deg 2 FOV at 0.7 GHz; “small” FOV partially compensated by better A eff /T sys n Large-scale cost – performance estimation begins Q4 ’06 –Closely allied with variational analysis wrt science goals –Strawman design for forthcoming Paris meeting

PJ Hall, June 2006 ISPO Wide Fields n Same FOV  same no. of receiver chains –Concept independent (almost) n Many small antennas  correlator intensive –Small dishes or AA patches n Fewer, larger antennas with FOV expansion  reduced correlator load –Use focal plane beamforming to reduce order of correlation challenge »Bigger dishes + PAF (or larger AA patches) »Big question: does extra PAF calibration cost negate correlator saving? –Bigger antennas »Better low freq performance »More sensitive, easier to calibrate »Better RFI discrimination –Smaller antennas »Probably more attractive production costing »Easier to calibrate? n Low & mid-band wide-FOV operation fits within processing envelope defined by high band SKA spec.

PJ Hall, June 2006 ISPO SKA Correlator – Output Data Rate 4000 stations 4 polarization products 2x16-bit fixed point numbers/complex value  128 MB per visibility set Integration time 0.1 sec 1000 spectral channels 10 station beams  9 TB/sec output data rate Need special purpose hardware for initial stages of post-correlation processing

PJ Hall, June 2006 ISPO SKA Correlator Attributes n Extreme flexibility –Simultaneous low, medium, high band operation –Complete trade-off of parameters (no. inputs, bandwidths, no. FOVs, processing accuracy, ….) –Support for “new” science: »High time resolution imaging, real-time VLBI, …. n Highly scaleable, reliable, maintainable, upgradeable –“Open telescope” ? ; standard data formats/interfaces (accept overheads) ; graceful degradation + hot spares operating model –Minimize NRE over life of telescope  maximum re-useability n Station correlators can be modest –E.g. calibration to optimize station beamforming may not require full or continuous bandwidth coverage n Signal connection and routing will be a major issue n Power is a major issue (remote sites, minimize op cost) n Line between correlator and other DSP will be blurred n Line between DSP engines and computers will be blurred n SKA DSP will likely be a mix of ASIC, FPGA and computers

PJ Hall, June 2006 ISPO SKA – Many Other Challenges n Low-noise, integrated, receivers –E.g. millions-off for mid-band n High speed data transport –Looking for 100 Gb/s trans-continental and trans-oceanic n Signal processing – beyond just correlation (IM, tied array modes, …) n Post-processing – computing capacity will limit initial science but cannot dominate system design –Archive and sharing of data will be a major challenge n Pathfinders and demonstrators are pivotal –Allen Telescope Array, LOFAR, xNTD, Karoo Array Telescope, DSNA, APERTIF, EMBRACE, 2-PAD …. –€200M committed so far; €80M explicitly for SKA; additional €40M expected in China shortly

PJ Hall, June 2006 ISPO SKA Engineering Philosophy n Strong emphasis on technology demonstration –Retire risk as early as possible n Focus on: –Aggressive cost reduction strategies (e.g. SKADS) –International collaboration & deliverables –Industry engagement »Pre-competitive R&D »Paradigm shift to deliver SKA on required timescales e.g. SD+PAF Demonstrator - NTD

PJ Hall, June 2006 ISPO Current SKA Happenings n Site assessment –RFI and other studies complete; list of “acceptable” or “qualified” sites soon n Funding agencies and SKA –Formed Inter-agency Working Group; continuing engagement n Funding opportunities (e.g. ESFRI) n Forthcoming Engineering – Science meeting –Paris, 4-8 Sept –Emphasis on Reference Design, project costing n International engineering review Q4/07 – Q1/08 –Reference Design, specifications, … n Continued science and engineering exposition n More outreach –New animations, telescope model, …. n More industry engagement –Major structural, governance implications n More inter-region collaboration –Easier as technology concepts coalesce

PJ Hall, June 2006 ISPO Summary n Site selection in progress n Reference Design identified n RD technologies being developed via regional pathfinders –Rapidly increasing inter-region collaboration n Initial SKA system design in progress –Incl. cost and performance modelling –Preliminary engineering reviews n Industry interaction increasing n SKA Phase 1 - start 2011