Transcript SuperGPS through optical networks

SuperGPS through
optical networks
Jeroen Koelemeij
LaserLaB VU Amsterdam
[email protected]
Partners
Funding
Objectives
• Optical methods to back up GNSS timing via fiber
• Next-generation timing via live optical networks,
T&F encoded in data streams
• Next-generation positioning (optical/radio)
Vision of SuperGPS
Activities at LaserLaB VU Amsterdam
• Frequency transfer through 2×317 km DWDM link
(VU, SURFnet, KVI Groningen)
• Optical data, time and frequency transfer
– 1 Gbps White Rabbit Ethernet
(VU, VSL, NIKHEF/WR project team)
– 10 Gbps data streams and amplified bidir fiber links
(VU, TU Eindhoven, SURFnet)
2×317 km DWDM link
2×317 km DWDM link
• Link part of SURFnet DWDM network
Frequency
comb
• Length 317 km, round trip 635 km
• Single l-channel (1559.79 nm)
• Fiber carrying live data traffic
• Not (yet) bidirectional; dark fiber also
available
2×317 km DWDM link stability
Results: stability measurements
Comparison Rb/GPS clock VU vs. Rb/GPS
clock KVI through 317 km fiber link
Specified stability of
Rb/GPS clocks used
at VU/KVI (@24h)
Clock laser 1 after roundtrip
through 2×317 km fiber link
Stability of state-ofthe-art GPS-CV time
transfer (@24h)
Clock laser 1 vs. clock laser 2
(‘noise floor’)
Stability of state-ofthe-art fiber links
(<1×10-18@104 s) [3,4]
Data suggest that one-way T&F transfer can back up GPS at least during short outages
(Cf. earlier data by SP/STUPI, CESNET/IPE :
1-way TT at GPS level on timescales months – year)
Optical TFT through White Rabbit
Optical TFT through White Rabbit
• Collaboration VSL and VU with technical support from NIKHEF
Amsterdam and WR project team
• Aim 1: use WR equipment for dissemination of UTC(VSL)
through installed telecom fiber link (u < 1 ns)
• Aim 2: demonstrate long-haul, scalable time transfer using
quasi-bidirectional amplifiers (u < 1 ns)
l1
OA
l2
WDM
OA
l1
WDM
l2
WDM
WDM
WR
node
Work in progress…
l1
WR
node
l2
Amplifier design: Amemiya et al., IEEE FCSE (2005)
Optical TFT through 10G data streams
Optical TFT through 10G data streams
•
•
•
•
Metropolitan area networks moving towards 10G Ethernet
10G more compatible than 1G but larger impact of dispersion
10G modulation offers better timing/frequency stability
Collaboration VU, TU Eindhoven: laboratory test setup
l1
A
Rb
clock
10 MHz
CH1 CH2 CH3
Oscilloscope
+ BER
Crosscorrelation
l2
O/E
E/O
l2
Determine delays using
Different wavelengths for
correlation of PRN codes
uplink and downlink
(cf. GPS, and recent work
Chromatic dispersion leads
at PTBtoand Paris (Rost et

Offline file transfer
& data analysis
l2
SOA
l1
WDM
O/E
SOA
25 – 50 km
WDM
C
WDM
E/O
l1
25 -37 km
WDM
10 Gbps
pattern
generator
delay
(tA - tC )/2
= tA -asymmetry
tB ?
al., Lopez et al. 2012)
B
Dealing with dispersion: method 1
Direct determination of dispersion: F. Devaux et al., J. Lightw. Tech. 11, 1937 (1993)
• Phase modulate optical carrier to produce sidebands at distance
±f from carrier
• Send through L = 50 km of fiber under test
• Detect modulation signal power while sweeping f by 0-20 GHz
Certain f: sidebands out of phase
due to dispersion
 signal disappears
Find dispersion from L and f :
D = -16.5(1) ps/nm km
End-end measurement, but can
be done from a single location!
Dealing with dispersion: method 2
• Concept suggested by Henk Peek: use a third wavelength
(uplink: l1, downlink: l1, l2)
• Measure two round-trip delays t12 and t12
• Taylor expansion of theoretical pulse propagation delay
 find OWD as function of li and t1i
𝐿
delay(𝜆) = 𝑛 𝜆 − 𝜆 𝑛′(𝜆)
𝑐
𝜆1 𝜏12 − 𝜏13 + 𝜆2 𝜏13 − 𝜆3 𝜏12
 OWD 𝜆1 ≈
2 𝜆2 − 𝜆3
• l1, l2 , l2 C-band, spaced by 200 GHz / 1.6 nm
For 50 km link, Dl = 1.6 nm :
• Typical delay asymmetry due to dispersion  1.3 ns
• Higher-order terms in Taylor expansion contribute ~ 4 ps
10G pseudo-random bit sequence
• Symbol length 100 ps
• Word length 223-1 is longer than ~0.5 ms round-trip time
• First test: sample 10 G PRBS at 6.25 GS/s (to speed up correlation)
Preliminary results (last week)
One way delay (OWD)
(50 km)
Round-trip delay (RTD)
(100 km)
240 ps
160 ps
Bit rate: 10 Gbps
Scope sampling rate: 6.25 GS/s
• Instrument delay asymmetry (~10 ns) measured seperately
with few-ps resolution (simply remove fiber spools)
• Observed delay asymmetry drifts over ~1 hr by < ±20 ps
Preliminary results (last week)
Round-trip delay (RTD)
(100 km)
One way delay (OWD)
(50 km)
160 ps
240 ps
Improved sampling rate to 12.5 GS/s
(80 ps resolution)
l1 = 1550.92 nm
l2 = 1549.32 nm
D [ps]
Disp. asymm. [ps]
Instr. asymm. [ps]
OWD (from RTD) 246 211.32(9)
0(90)
-1 309(7)
14 591(40)
Sellmeier
(53±50)×103
50 km link Delay [ns]
OWD (direct)
246 211.32(8)
246 222(50)
Time transfer results 50 &75 km links
OWD (from round trip) – OWD (direct measurement)
Using dispersion estimate
Time deviation [ps]
Using two-l formula
St. dev.
27 ps
50 km
l1, l2
50 km
l1, l3
75 km
l1, l2
75km
l1, l3
50 km
l2+ l3
75 km
l2+ l3
Comparison with state of the art
Method
Distance
Accuracy
Ref.
GNSS
>1000 km
3 – 50 ns
TWSTFT
>1000 km
1 ns
T2L2
> 1000 km
200 ps expected
Fridelance et al.,
Exp. Astr. (1997)
Optical fiber
(20 Mbps PRBS)
540 km
250 ps
Lopez et al., Appl.
Opt. (2012)
This work
(10 Gbps PRBS)
75 km
90 ps
Optical fiber
(20 Mbps PRBS)
73 km
74 ps
Rost et al.,
Metrologia (2012)
Dedicated optical fiber
(10 MHz + 1pps)
69 km
(480 km
preliminary)
7 ps
Krehlik et al., IEEE
Trans. Instr. Meas.
(2012)
Data transfer results
Simultaneous sub-100 ps time transfer and…
error-free 10 Gbps data transfer!
To do list
• Measure frequency stability @ 10G
 Towards 10G time, frequency and data transfer
• Implement lightweight correlation algorithm:
correlate bit vectors (00110101010110…) rather than
oscilloscope traces (FPGA?)
• Implement in installed optical fiber link
• Study long-term stability (days-months-year)
Outlook: TWTT + one-way FT
• Assume 100 ps timing accuracy (via 10G or WR)
• Assume one-way frequency dissemination
• Typical stability of one-way fiber frequency transfer
at 10G:
– ADEV(1s) = 10-13

TDEV = 60 fs
– ADEV(2hrs) = 2×10-14 
TDEV = 80 ps
 every 2 hrs new synchronization required to
maintain 100 ps timing accuracy
• Possible technology for long-haul, sub-ns level
timing + data link for OPERA experiment CERN-LNGS
• c × 100 ps = 3 cm
 ‘SuperGPS’ positioning!
Thanks!
• Tjeerd Pinkert, Matthijs Jansen, Wim Ubachs, Kjeld Eikema
(VU)
• Roeland Nuijts, Richa Malhotra, N.O.C. (SURFnet)
• Oliver Böll, Lorenz Willmann, Elwin Dijck, Bart Groeneveld,
Klaus Jungmann (KVI/RU Groningen)
• Huug de Waardt, Chigo Okonkwo (TU Eindhoven)
• Erik Dierikx, Marc Pieksma (VSL)
• Henk Peek, Peter Jansweijer (NIKHEF), White Rabbit Team