An alternative way of WAM system time synchronization. Presented by Vojtěch Stejskal ATM Madrid 2015

Transcription

1 An alternative way of WAM system time synchronization Presented by Vojtěch Stejskal ATM Madrid 2015

2 Presentation Overview WAM around the world Page 2 Introduction Synchronization techniques GNSS vulnerability System behavior Over-determination Basic assumptions Principle Real Data Results WAM system selection Results System fusion advantages Conclusion

3 Synchronization Techniques WAM Synchronization Central/Common Time Systems Distributed Time Systems Transponder/beacon based synchronization GNSS based synchronization Standalone GNSS Common View Time normal Oscil RXS Oscil RXS Time normal Page 3 Dedicated line RXS Oscil SNTP CPS UTC Common Clock Time mark Central Time synchronization: Very robust to outer influence No computing part on the receiver side Required dedicated communication lines between RXS and CPS (microwave, optical link) with constant propagation delay Need to be calibrated Applicable to small and mid size WAM systems

4 Synchronization Techniques WAM Synchronization Central/Common Time Systems Distributed Time Systems Transponder/beacon based synchronization GNSS based synchronization Standalone GNSS Common View Time normal Oscil Oscil Time normal RXS RXS Reference Transponder RXS SNTP CPS UTC Time mark Transponder based synchronization: No dedicated lines required Limited by LoS Applicable to system with short baselines Influenced by atmospheric behavior Wider system may consist of multiple Reference Transponders Page 4 Oscil Time normal

5 Synchronization Techniques WAM Synchronization Central/Common Time Systems Distributed Time Systems Transponder/beacon based synchronization GNSS based synchronization Standalone GNSS Common View Time normal Oscil Oscil Time normal Page 5 Time mark GNSS RXS RXS GNSS UTC UTC Time mark Each RXS sends to CPS: Time marks from GPS UTC time and additional GPS info Mode A/C/S codes with TOA Content of mode A/C/S codes SNTP CPS Common View CPS: Time mark GNSS RXS Oscil UTC Time normal Provides corrections to TOA Calculates TDOA GNSS based synchronization: The most frequent type of synchronization Provides corrections No dedicated lines required More GNSS Types (GPS, GLONASS, ) Applicable up to large WAM systems Dependent on GNSS availability Jamming, spoofing, space weather

6 GNSS vulnerability Intentional Jamming and Spoofing Continuous wave, wide-band, narrow-band radio frequency signals Personal Privacy Devices (PPD), special purpose jamming devices Noted by GNSS interference monitoring or after service outage Unintentional - Interferences In-, near-, and out-of-band interference, spurious emissions or harmonic interference from other systems (communication systems (Lightsquared 4G), repeaters, pseudolites) Unintentional Space Weather Recent NOAA warning for 7 th January :22-15:00 when Kpindex should exceed 7 Intermittent satellite navigation (GPS) problems, including loss-of-lock and increased range error may occur K=7 29 th Jun 2013, 9 th Mar 2012, 1 st Oct 2012, 9 th Sep 2011, 24 th Oct 2011 K=8 26st Sep 2011, 5 th Aug 2011 Page 6

7 Free running reference clock stability Local time reference can keep WAM system sufficiently synchronous for certain amount of time to overlap GNSS outages Deviation [ns] Page 7 Time [min]

8 Problem formulation Recent State Nowadays DT WAM systems are capable to reflect short-term GNSS outages High GNSS dependency will force aircraft to land during long term outages anyway Requirement Need to find way how to face longer GNSS outage How to safely get aircrafts to the airports Need to find solution without requirement for HW changes fitting already deployed systems Page 8

9 Basic Assumptions T-1 Existing Time Reference Targets are tracked in real 3D space, i.e. TOA measurements belonging to target have dimension greater than 3. T=0 Loss of Time Reference Synchronization inspired by calibration techniques The system can be synchronized/calibrated only from real traffic TDOA measurements with targets of unknown position. Proposed approaches: Synchronization from overdetermined TDOA measurements with known height (req. 4 stations and more) Synchronization from overdetermined TDOA measurements of targets with unknown position (req. 5 stations and more) Page 9

10 Synchronization Principle Step I. - Primary measurements Any TDOA measurements with dimensions equal or higher than 4. Step II. - Conversion from hyperbolic to Cartesian 3D space Approximate target s position in 3D computed via analytical conversion or updated from an existing tracked target corrected by over-det. measurements. Difference between the most probable target position in 3D (3D->Hyp.) from previous steps and entering primary TDOA measurements -> inferior clock deviations. Step III. - Clock deviation estimate Coalescence of ca. 1 minute consequent differences between measured and the most probable target s position in hyperbolic space -> estimated correction of clocks additional delays. Step IV. - Correction of additional delays Kalman filter allows to model an additional delays error trend (error of clock synchronization). The 3D position is corrected by all TDOA measurements ->over-determined recalculation. Page 10

11 Verification on WAM systems Method evaluated on real data recorded simultaneously from two WAM systems Brno and Ostrava. Each WAM system consists of 7 stations and partly overlap in coverage. Evaluation started on targets with known position and time, when synchronization was lost. Reference clocks on each of stations are mutually deviated by random number generator and diverge without any synchronization. Output: Estimated clock deviations and accuracy from over-determined measurement on station 2-7 are referenced to clock at station 1. The difference between deviations on station 1 and estimated deviations on the rest of stations produced by synchronization measurements corresponds to systematic errors affecting all targets. Page 11

12 Results WAM Brno Systematic errors of over-determined TDOA measurements on each of WAM Brno stations Traffic during the worst time period for synchronization by over-determination method Page 12 The weak synchronization during night traffic from a worst time period for synchronization by the over-determination method is between 600th and 660th minute (from 1 AM until 2AM). SLIDE 12

13 Results WAM Ostrava Systematic errors of over-determined TDOA measurements on each of WAM Ostrava stations Recorded traffic time period for synchronization by over-determination method Page 13 SLIDE 13

14 Virtually Merged Systems Decreasing synchronization error by fusing data from multiple WAM systems (stations). Combination of synchronized and unsynchronized system. Required target s common view, i.e. coverage overlap. Tested on related systems Brno and Ostrava with sufficient overlap. H RMS WAM Ostrava alt >4575m H RMS WAM Brno alt >4800m Page 14

15 Merged WAM Morava Systematic errors of over-determined TDOA measurements on fused WAM Brno and Ostrava Overnight traffic density 99% of sync errors doesn t exceed 2ns Page 15 SLIDE 15

16 Conclusions Proved Wide Area System synchronization by over-determination when any kind of clock reference is lost. 99% of all sync errors didn t exceed 10ns. Required sparse sensor distribution and amount of at lest 5 sensors. The method is moreover valid also for weak traffic. Negative influence on synchronization can have atmosphere in case when its actual state will diverge from expected model. 99% percentile margin can be reduced to even less then 2 ns by increasing amount of stations. Page 16 Protected by U.S. Patent Application Ser. No. 61/474,350

17 Page 17