Title: Next generation telescopes
1Next generation telescopes
- Conceptual Design for
- the Square Kilometre Array
- R. D. Ekers
- CSIRO, Australia Telescope National Facility
- ADASS Baltimore
- Oct 14 2002
2Radio Telescope Sensitivity
- Exponential increase in sensitivity x 105 since
1940 ! - 3 year doubling time for sensitivity
3Exponential Growth
- Livingstone Curve
- Blewett, Brookhaven 1950
- Fermi 1954
- Livingstone 1962
- De Solla Price 1963
- exponential growth
- The exponential is the envelope of many
technologies - Little science, big Science
- Moores Law (1965)
- computing power doubles every 18months
4The Original Moores Law Plot
- In 1965 Gordon Moore (co-founder of Intel) noted
that the transistor density of semiconductor
chips doubled roughly every 18 months.
Extrapolation accurate for another 35 years!
5Microprocessor performance
6Radio Telescope Sensitivitythe Future
- Exponential increase in sensitivity x 105 since
1940 ! - 3 year doubling time for sensitivity
- No technical reason not to continue this growth
Allen Telescope Array SETI Institute UC
Berkeley Path to the Future
Future Radio Telescopes
7Astronomy Before the Light
8Square km telescope the concept
- Frequency range 0.03 - 20GHz
- Sensitivity 100 x VLA
- Resolution 0.1 0.001
- Multibeam (at lower frequencies)
- Need innovative design to reduce cost
- International funding unlikely to exceed 1000m
- 106 sq metre gt 1000 / sq metre
- cf VLA 10,000 / sq metre (50GHz)
- GMRT 1,000 / sq metre (1GHz)
- ATA 2,000/sq metre (11GHz)
9Square km telescope the concept
- Frequency range 0.03 - 20GHz
- Sensitivity 100 x VLA
- Resolution 0.1 0.001
- Multibeam (at lower frequencies)
- Need innovative design to reduce cost
- International funding unlikely to exceed 1000m
- 106 sq metre gt 1000 / sq metre
- cf VLA 10,000 / sq metre (50GHz)
- GMRT 1,000 / sq metre (1GHz)
- ATA 2,000/sq metre (11GHz)
10Square Kilometer Array
- Current large telescope technology dates from
1960/70s - Era of facilities upgrades approaching its end
- Large increase in sensitivity is needed (100x)
- Match developments in other wavebands
- epoch of first stars and galaxies
- New challenges
- cost
- frequency coverage
- man-made interference outside protected bands
- Technology shift will be required...
11Radio Telescopes of the Future
- HEMT receivers
- wide band, cheap, small and reliable
- Can build low noise systems with many elements
- Focal plane arrays
- Field of view
- Interference rejection
- adaptive nulling can work in single dishes and
arrays - More computing capacity
- computing power doubles every 18months (Moores
Law) - Software time scales are much longer
- it becomes a capital cost !
12Terrestrial Interference
FORTÉ satellite 131 MHz
13Digital
- Using Moores law
- Wide bandwidth signal processing
- 2-20GHz
- Software radios
- 20GHz in next decade
- consumer market driven
- Interference mitigation
- Access to entire spectrum
- Smart Antennas
- Image processing
- Can make arrays easy to use
14Mass-produced parabolas The Allen Telescope
Array
- SETI Institute UC Berkeley
- 100m equivalent
- 350 x 6.1 m parabolas
- 0.5-11 GHz (simultaneously)
- 2.5o FOV at 1.4 GHz
- 4 simultaneous beams
- Complete 2005
15Big Dishes or Arrays
16Phased Array
Alternate idea replace mechanical pointing, beam
forming by electronic means
17THEA
18SKA options..
19How to Build the SKA?
- We have the technology to build the SKA now
- Diffraction limited interferometry
- Self calibration (Adaptive Optics)
- We must find the most cost effective solution
- Arrays of small dishes
- Planar Phased array
- Single adaptive reflector
- KARST formations (multiple Arecibos)
- Array of Luneburg lenses
20Sensitivity
21Future Sensitivity
HST
22Computing Demand Sensitivity
- Bits per sample
- At least a few (4-8) in the Fourier plane
- Reliability
- Monitoring, debugging, maintenance
- Interference rejection
- To reach sensitivity - 40db
- Dynamic range
- 106 required
23Interference excision
- 8 narrow-band elements
- MVDR algorithm minimises S/(IN)
NFRA 1998
24Removal of GlonassAdaptive canceller
- Example using Ellingson data
25ATA Can Null Out Interfering Satellites
- 350 antennas imply 700 independent degrees of
freedom - Nulls can be created anywhere on the sky
- Nulls can be arbitrarily deep in intensity or
broad in frequency - Nulls can track satellite orbits in real-time
Our own brand of skywriting!
26Dynamic range
- Calibrate a synthesized time-variable primary
beam - AIPS calibration and imaging formalism allows
correction for image-plane calibration effects - Self calibration (adaptive optics)
- A large N array (N 100 10000)
- Algorithm development
- AIPS should help increase this pace
- Very powerful tool-based infrastructure
- Community for sharing software, algorithms
27Large-N Arrays
The radio interferometry group at MIT/Haystack
is studying the design of an array made up of a
large number of telescopes.
- What is meant by large-N?
- Number of antennas is large compared to most
current SKA concepts - Take N to be several thousand, implying millions
of baselines, and antenna diameters of several
meters - Key-characteristics of large-N configurations
- Extremely dense u-v coverage (configuration
dependant) - Small antennas and large primary beams
- Large number of Rx, cables, and massive
correlation - From these characteristics spring a startling
number of benefits
28VLA - 27 Elements
1000 Element Array
Configuration
Configuration
Snapshot
Snapshot
Tapered Beam
Natural Beam
UV Coverage
UV Coverage
Magnification of Beam
Tapered Beam
UV tapering of the form cosn(?k(u2 v2)) was
used for the simulations, with n 2 for VlA and
n 6 for SKA J. Hewitt and A.
Cohen, MIT
29Computing demand Resolution
- Maximum baseline
- For full FOV 300km
- For full resolution in subfields up to 5000km
- Signal distribution
- Transport GHz signals over 1000km to hundreds of
stations - this will limit ultimate performance
30One example of an SKA configuration
Not a single 1 km square aperture !
a wide range of baselines
31Field of View
32SKAs 1o field-of-view
- for surveys and transient events in 106 galaxies !
SKA 20 cm
15 Mpc at z 2
ALMA
33Computing demand FOV
- Wide field synthesis corrections
- Non planar effects
- Chromatic aberration
- Image size
- 105 x 105 pixels (400x VLA)
- Use hierarchical beamforming?
- Correlator size
- 3000 stations 107 baselines
- 1000 frequency channels
- 1000 samples/image
34Computing demand spectral
- Very wideband signal paths
- gt1 GHz
- 1000 frequency channels
35Correlator
36Multiple Beams
37SKA Poster
38Multi beams
Element antenna pattern
Station antenna patterns
Synthesized beams
16
12
- Observing teams with their own beams
- like particle accelerator, but can have all beams
simultaneously - Baseband buffer
- Observe before trigger !
8
4
NFRA 1998
39Computing demand multiple beams
- Electronic beam steering
- Adaptive nulling
- Diverse backend configurations
- Spectral line imaging
- Pulsar timing
- GRB detector
- SETI processor
- Parallel simultaneous observations
- Multiple backend configurations
- Multiple users
40InterstellarScintillation
- Frail Kulkarni
- VLA 8GHz
- Scintillates if
- ? lt 10 ?as
- Calibrate of field SNRs
- Only 1 GRB strong enough in 4 years
- Many days integration
41Computing demandVirtual observing
- Store undetected signals
- Eg 106 elements at 100MHz for 10sec 1015
samples - Trigger machinery
- Electronic beamformers
42International
43Achieving the vision - International
Collaboration?
- To build facilities which no single nation can
afford - Coordination
- Avoiding wasteful competition
- Broader knowledge base, cross fertilisation
- Wealth creation
44Computing demand International
- Joint software development
- Remote observing
- Archival access
45Computing demand Simulations
- Must simulate real problems,
- Not just what you can do
- Computing requirements are always
state-of-the-art in hardware and software - simulation computing calibration computing
- For SKA, need to simulate imaging in the presence
of time-variable primary beams - Large computing needs parallel computing is vital
46THEA