NPL Time & Frequency NPL Time and Frequency Section: NPL S CONTRIBUTION TO TUGGS J A Davis, P W Stacey, R Hlavac, and P B Whibberley. Date: 21st April 2004
THALES UK-BASED GNSS GROUND SEGMENT (TUGGS) Aim is to demonstrate a UK capability to generate a GNSS ground segment. Useful for future UK participation in Galileo. Use GPS satellite signals. Do not use the navigation message. Instead the aim of the project is to produce an independent navigation message. Project funded by BNSC.
NPL S PARTICIPATION NPL to provide the timing reference. Provide real time access to a stable time scale UTC(NPL). Provide UTC UTC(NPL) offsets. Develop timescale infrastructure at NPL for use in future GNSS systems.
WORK OF THALES Key Thales Staff: Andy Vooght, Andrew Batchelor Key jobs: Orbit determination, satellite force modelling. Software development and evaluation.
DESCRIPTION GPS satellite position and velocity over 24 hours Estimated by integrating satellite forces over period of 24 hours Case 1: Gravity model Case 2: Gravity model + third body forces Case 3: Gravity model + third body forces + solar radiation Performance compared to true position and velocity obtained from IGS precise orbit file
Case 1 results - position 4000 3000 2000 Position error (m) 1000 0-1000 x1 error x2 error x3 error -2000-3000 -4000 0 7200 14400 21600 28800 36000 43200 50400 57600 64800 72000 79200 86400 Time (s)
Case 2 results - velocity 0.03 0.02 0.01 Velocity error (m/s) 0-0.01 v1 error v2 error v3 error -0.02-0.03 0 7200 14400 21600 28800 36000 43200 50400 57600 64800 72000 79200 86400 Time (s)
Software Implementation & Testing (1) u Software structure RTODS Configuration and Monitor (RTOD GUI) Quasi-Static Data Configuration & Monitor Data Tropospheric Delay Data RTODP Data Server GRIB Files IERS Bulletins A & B GPS Observations Common Pre-Processing Batch Orbit Real Time Orbit Clock Data Formatter SP3 Files Real Time Correction Data Status Alarms
NPL S CONTRIBUTION TO TUGGS WP 1.1 Phase and pulse logging system. WP 1.2 Time scale algorithm. WP 1.3 UTC Predictions.
WORK PACKAGE 1.1 PHASE AND PULSE LOGGING SYSTEM Phase comparator data logging software is complete. Data processing software to generate phase-smoothed clock time offsets is complete. NPL HARDWARE Three active hydrogen masers Two Phase Comparators Two 1PPS logging systems
WP1.1 DATA LOGGING AND PROCESSING SYSTEM HM1 HM2 Clock HM3 Room New Building FDA FDA FDA PG PG PG HM FDA PG CT PDA Legend Hydrogen Maser Frequency Distribution Amp. Pulse Generator Counter Timer Pulse Distribution Amplifier 10Mhz 1PPS GPIB Network PComp1 PComp2 Europa PC Office PDA PDA PDA CT CT Ananke PC Chaldene PC
COMBINING CODE AND PHASE MEASUREMENTS
PLOTS OF LOG 10 (σ y ) AGAINST LOG 10 (τ)
KEY FEATURES OF CODE PHASE DATA COMBINING SOFTWARE Uses a simple Kalman filter to combine the code and phase data sets. Checks and removes outliers and steps in the data. Automatically switches onto reserve hardware if phase comparator fails. Will cope with missing data to maintain continuity of output.
WORK PACKAGE 1.2 TIME SCALE ALGORITHM Developed a modern, Kalman filter based clock algorithm. Used C Greenhall s covariance reduction techniques to defeat the problem of partly observable state vectors. Better approach than the pseudo-measurements used in the GPS composite clock. Refined the algorithm to operate in the presence of flicker frequency modulation. Evaluated using simulated data.
TESTING ALGORITHM USING SIMULATED DATA CLOCKS 1 AND 3 WFM, CLOCK 2 FFM 200 100 Clock Offset Clock 1 Clock 2 Clock 3 Composit Clock Offset 0-100 -200-300 -400 0 2000 4000 6000 8000 10000 Time
PLOTS OF LOG 10 (σ y ) AGAINST LOG 10 (τ) CLOCKS 1 AND 3 WFM, CLOCK 2 FFM 0 PLOTS OF LOG 10 (σ y ) AGAINST LOG 10 (τ(s)) Clock 1 Clock 2 Clock 3 Composite Theory -0.5 Log10(σy) -1-1.5 0 0.5 1 1.5 2 2.5 3 log 10 (τ)
TESTING ALGORITHM USING SIMULATED DATA CLOCKS 1 AND 3 RWFM, CLOCK 2 FFM Clock Offset 2.5 2 1.5 1 0.5 0-0.5 3 x 10 4 Clock Offset Clock 1 Clock 2 Clock 3 Composit -1 0 2000 4000 6000 8000 10000 Time
TESTING ALGORITHM USING SIMULATED DATA CLOCKS 1 AND 3 RWFM, CLOCK 2 FFM 10 Log10(σy) 0.2 0-0.2-0.4-0.6-0.8-1 -1.2 PLOTS OF LOG 10 (σ y ) AGAINST LOG 10 (τ(s)) Clock 1 Clock 2 Clock 3 Composite Theory -1.4 0 0.5 1 1.5 2 2.5 3 log 10 (τ)
ADEV ESTIMATED USING A THREE CORNERED HAT AND FROM THE NOISE PARAMETERS -14-14.2-14.4-14.6 PLOTS OF LOG 10 ( σ y ) AG AIN ST LO G 10 ( τ (s)) Clock 3 Real AVAR Clock 2 Real AVAR Clock 1 Real AVAR Clock 1 Noise Parameters Clock 2 Noise Parameters Clock 3 Noise Parameters log 10 (σ y ) -14.8-15 -15.2-15.4-15.6 2 2.5 3 3.5 4 4.5 5 log 10 (τ)
APPLICATION TO REAL NPL CLOCK DATA ADEV (COMPOSITE CLOCK) -14 PLOTS OF LOG 10 (σ y ) AGAINST LOG 10 (τ(s)) Composite - Clock 1 Composite - Clock 2 Composite - Clock 3 Composite (Theory) -14.5 Log10(σy) -15 2 2.5 3 3.5 4 4.5 5 log 10 ( τ )
IMPLEMENTING THE CLOCK ALGORITHM FOR CONTINUOUS OPERATION Data will be processed on either a daily or hourly basis. Front end software now complete. Need to combine main body of software and the front end. Phase micro-stepper on order so as to provide for real time implementation. Backup hardware available at NPL.
WORK PACKAGE 1.3 UTC UTC(NPL) CLOCK PREDICTOR NPL reports data on the UK s national time scale UTC(NPL) to BIPM once per month. BIPM uses that data to compute UTC UTC(NPL) and reports data back to NPL usually approximately six weeks in arrears. To relate UTC(NPL) back to UTC in the short term (e.g. τ = 1day) we must estimate the current value of UTC UTC(NPL) from the last value published by BIPM. Clock predictor developed to achieve this.
TWO STAGE ANALYSIS PROCESS Characterise the clock s performance in terms of deterministic parameters, time offset, frequency offset and linear frequency drift and WPM, WFM and RWFM noise types. STAGE 1: Determine the magnitude of the noise type parameters using least squares analysis. STAGE 2: Determine the values of the deterministic parameters and the uncertainty associated with estimate of those values, again using least squares analysis. Kalman filter version of the clock predictor under development.
OPERATION ON SIMULATED DATA TIME OFFSET PREDICTION UNCERTAINTY (STANDARD DEVIATION), 100 SIMULATIONS 30 25 Estimated noise parameters, error estimates from simulation Arbitary units 20 15 10 5 0 0 2 4 6 8 10 12 Prediction Length (Epochs) Estimated noise parameters, error estimates from theory Known noise parameters, error estimates from simulation Known noise parameters, error estimates from theory
APPLICATION TO UTC UTC(NPL) DATA Run clock prediction software once per month, after NPL has received the circular T report from BIPM. Update existing predictions, make new predictions and compare previous predictions with latest measurements. Supply Thales with latest predictions once per month, one estimate per day.
PREDICTION OF UTC-UTC(NPL) 7.00E-08 6.00E-08 UTC - UTC(NPL) 5.00E-08 4.00E-08 3.00E-08 2.00E-08 1.00E-08 0.00E+00-120.00-100.00-80.00-60.00-40.00-20.00 0.00 20.00 40.00 60.00 MJD -52904 Prediction Orginal Measurements Later Measurements
PREDICTION ERROR Error (s) 1.80E-08 1.60E-08 1.40E-08 1.20E-08 1.00E-08 8.00E-09 6.00E-09 4.00E-09 2.00E-09 0.00E+00 0 5 10 15 20 25 30 35 40 45 50 Prediction Length (days) Error Expectation Value Example Error
LINKING THE TUGGS PROJECT TO UTC(NPL) Trimble GPS receiver located at NPL 1PPS input to receiver is related to UTC(NPL) Receiver is outputting RINEX every 30s to local PC Link still needs to be set-up for rapid data transfer to Thales
WORK STILL TO COMPLETE Setting up clock algorithm to operate automatically on a daily basis (component parts complete). Hardware realisation of composite timescale. Agree format to send UTC-UTC(NPL) clock prediction data to Thales.