DS3-T1 Network Infrastructure

1
DS3-T1Network Infrastructure data
transmissionSimon GarringtonUniversity of
Manchester/Jodrell Bank Observatory, UK

• Feasibility study of network for SKA
• Data link from dishes/stations
• Monitor Control links to dishes/station
• Clock distribution across network
• Partners
• Manchester/Jodrell Bank e-MERLIN
• Spencer, Garrington, Anderson, McCool, Bentley,
  Davis
• ASTRON LOFAR
• Maat, Gloudemans
• IRA/INAF BEST
• Montebugnoli, Boschi

2
DS3-T1SKADS Deliverables

• Detailed cost models for data/MC links
• COTS technology
• Standard protocols commercial equipment
• Component based technology
• Protocol-free mass market components
• Short-distance analogue links
• Prototype phase transfer links

3
DS3-T1 Data transfer
SKA Reference Design From AF Talk

• Scale of problem
• 14-72 Gb/s per beam per dish/station
• x 4 for parallel operation?
• 10Gb/s 300 - 10000 depending on distance
• 3k per beam per dish/station
• Nb x Ns 105 ? 300M
• Diggingcable 15/m
• 5km 150km arrays -gt 15M 15M/1000km of
  long links

4
Initial Cost Model

• Memo 75 Ralph Spencer - see Ralphs talk
• Inner 5km array 2000 dishes
• lt3GHz 40 beam PAF 300 MHz b/w 8 bits
• gt3GHz 1 beam WBF 2.5 GHz b/w 6 bits
• 300 10G Tx-Rx pair, 10/m trenchcable
• 160M
• Outer array cost comparable for single beam,
  otherwise dominates

5
DS3-T1 Approach

• Cannot wait for DS3-T2/3 to deliver array
  configuration, processing topologies then design
  network
• Must provide refine unit costs for various
  useful links and build generic cost model tool
• Transmission equipment
• Multiplexing CWDM, DWDM, TDM
• Amplification (gt100km) Regeneration (gt400 km)
• Connectors fibre management
• Installation commissioning costs
• Multiplexing trade-offs
• Clarify cost break-points in terms of data rates,
  distances and aggregation
• Iterate within DS3

6
DS3-T1Initial Progress

• Initial technology survey for COTS links (ASTRON)
• Based on LOFAR experience 1,10 GigE
• Looking at switches, DWDM,CDM
• Anticipate move 1/10 G ?
  10/40/100G
• Generic WAN modelling tool started (ASTRON)
• Use for COTS data links and low bandwidth MC
  network

LOFAR Arms 45 stations, 2Gb/s, 50km Core 32
stations, 2 Gb/s, 5km
7
DS3-T1Data transfer component based

• Technology survey for component-based links
  (Jodrell/Manchester)
• Based on e-MERLIN ALMA experience 30 -120 Gb/s
  custom DWDM on single fibre from antenna
  developed by NRAO/JBO
• e-MERLIN
• 600km dark fibre network leased/installed
• Looking at Fibre-to-the-Home for cheap short
  links
• VCSELs, Plastic Optical Fibre
• mechanical splices, cheap connectors, easier
  fibre management
• Industrial links being sought

8
Issues emerging

• No. of bits to transport
• No of beams on outer stations
• Cheapest short (0.1 5 km) links
• New Technology (VCSELs, POF)
• Topology
• Fibre management

9
Phase Transfer

• Need 1 ps timing control for 1 loss at 20 GHz
• Options
• Independent clocks (masers) expensive (200k)
• Transmit signal steer oscillator at remote end
  track round trip delay
• Fibre
• EVLA (2 fibres), ALMA (optical LO distribution)
• Temperature control non-reciprocity
  polarisation effects
• Multi-hop links
• Radio
• MERLIN (pulsed 1486.3 MHz)
• RFI scale-up?
• 2-way satellite (TWSTT)
• USNO clock transfer lt100 ps/day over gt1000 km
• Weinreb et al few ps
• Cost to rent transponder channels
• In practice, may use mixture, depending on
  distance

10
Phase Transfer

• Isolating return signal
• Circulators
• Simple system tested at JBO
• Single freq 1 GHz modulation VVM phase
  measurement
• Good circulator 30dB -gt 2dg
• Dual freq schemes
• Time multiplex
• MERLIN L-band link
• 1486.3 MHz pulsed at 88 Hz
• Pulse length discrimination
• Works to few ps over gt200km
• Tests planned
• e-MERLIN fibre spans (above below ground,
  300km)
• Implement MERLIN 1484.3 MHz TDM on fibre

Spencer, Sureshkumar,Strong,STG
Two-way 2(one-way)
11
Phase Transfervia GPS

• GPS Common View being tested for time transfer
  (USNO, NIST, BIPM,) using carrier phase
• Better than TWSTT for 100s lt t lt 1d
• Used dual-freq geodetic antennas/receivers,
• final GPS ephemerides
• Short timescale limitations
• Noise GPS multipath (rms 50cm)
• Want beamforming antenna
• SKADS Aperture Array (1227,1575 MHz)
• High gain improve SNR
• High degree of multipath rejection
• Strategies to use this for phase transfer
• Next gen. GPS, Galileo

Plumb Larson 2005
12
(No Transcript)
13
Allan variance plot
GPS stability floor 2 10-13 ? -0.44
14
GPS
IGS Product Orbit Clock Latency Update
Broadcast 200cm 7ns Real-time Daily
Predicted 10cm 5ns Real-time 6h
Ultra-rapid lt5cm 0.2ns 3h 6h
Final lt5cm 0.2ns 14days