Satellite Time Transfer

Why do satellite signals work better than ground signals for time and frequency transfer? Path delay is easy to estimate and calibrate for timing applications. The variation in path delay is small due to a clear, unobstructed path between the receiver and transmitter. The coverage area is usually much larger. Interference due to weather and ground based noise is usually less of a problem.

Ground-based signals skywave groundwave line-of-sight

HF Radio Propagation (skywave)

LF Radio LF (low frequency) is the part of the spectrum from 30 to 300 kHz, also known as longwave. Used to send time codes via simple modulation schemes. The carrier frequency is also used as a frequency reference. Groundwave signals are more stable, and the delays are easier to estimate than the HF skywave signals. Two LF signals widely used for time and frequency: LORAN-C (100 kHz) NIST Radio Station WWVB (60 kHz)

LF Radio Propagation (groundwave)

Disadvantages of LF Limited coverage area. Subject to diurnal phase shifts at sunrise and sunset over long paths, skywave can interfere with groundwave. When receiver is unlocked, cycle slips equal to the period of the carrier (16.67 s for WWVB, 10 µs for LORAN) are introduced in the data. User must calibrate path delay for time transfer, and even then is limited by the cycle ambiguity. Cycle ambiguity is a much larger problem with WWVB than LORAN.

Line-of-Sight Radio Propagation

Line-of-sight signals (VHF/UHF) VHF (very high frequency) is defined as the spectrum from 30 to 300 MHz. UHF (ultra high frequency) is defined as the spectrum from 300 to 3000 MHz. Generally speaking, signals transmitted from the ground in the VHF and UHF spectrums tend to be line-of-sight. In other words, they don’t bounce off the ionosphere or follow the curvature of the Earth, but instead are used for local transmissions with limited coverage area where there is a clear path between the transmitter and the receiver. Line-of-sight signals are stable, but the coverage area is usually small. Several line-of-sight signals have potential applications in T&F, including FM radio signals, television signals, and cellular phone and pager signals.

Satellite Signals The best signals for time transfer. Since the signals originate high above the Earth, there is an clear path between the transmitter and receiver. Coverage area can be worldwide with global navigation systems like GPS. Small path delay changes occur as signal passes through ionosphere and troposphere, but these are measured in nanoseconds. Satellite signals used for time and frequency include: GPS GLONASS (Russian version of GPS) Galileo (European GPS, coming in future years, first satellite launched on December 28,2005)

Satellite Radio Propagation

Summary Table Type of Signal Spectrum Coverage of single system Stability Reliability of Reception Skywave MF, HF Good Poor Groundwave VLF, LF Line-of-sight VHF, UHF Excellent Satellite UHF

The GPS Infrastructure and Signal Format

GPS Satellite Constellation Semi-synchronous, circular orbits (~20,200 km/10,900 nautical miles altitude) Six orbital planes, inclined at 55 degrees, four vehicles per plane Orbital period is 11 hours, 58 minutes Spares can bring number of satellites up to 32 – new satellites are launched as necessary, lately 2 or 3 per year Designed to cover entire earth, with at least four satellites always in view Cesium and/or rubidium oscillators are on board each satellite

GPS History Developed by the US Department of Defense Earlier satellite timing systems existed Transit GOES Timation (first atomic frequency standards flown in space) USAF 621B Program (PRN codes for ranging) First prototype GPS satellite launched in 1978 First Block II (Operational) GPS satellite launched 1989 Full Operational Capability declared in late 1993

Precise Timing is Fundamental to GPS Assume the maximum acceptable error contribution from GPS satellite clocks is 1 meter Light travels 3 x 108 m/s, thus a one meter requirement is equivalent to 3.3 ns ranging uncertainty Clock error must be maintained below this level over 12 hour period (time between satellite uploads) This requires a clock with < 1 part in 1013 stability, which can only be met by an atomic standard (3.3 x 10-9 s / 43200 s = 0.8 x 10-13)

Atomic Clocks in Space GPS has become the primary system for distributing accurate time and frequency globally GPS satellites carry rubidium or cesium oscillators (or both) Precise frequency standard provides a reference for generating the ranging signals transmitted by the satellites Clocks on the satellites are steered by U. S. DoD ground stations to UTC as maintained by the US Naval Observatory (USNO) UTC(USNO) is usually within a few nanoseconds of UTC so GPS provides a real-time link to the UTC time scale

Relativistic Effects in GPS Einstein would be proud of GPS, because it is a real world application for his theory of relativity. The oscillators onboard the GPS satellites are given a fixed frequency offset of -4.4645 x 10-10 to compensate for relativistic effects in the GPS satellite orbits. Second-order Doppler shift – a clock moving in an inertial frame runs slower than a clock at rest. Gravitational frequency shift – a clock at rest in a lower gravitational potential runs slower than a clock at rest in a higher gravitational potential. Without this frequency offset, GPS satellite clocks would gain about 38 microseconds per day relative to clocks on the ground. GPS receivers apply an additional correction of up to 23 ns (6 meters) to account for any eccentricity in the satellites orbit.

GPS Signal Structure Two L-band carrier frequencies Two PRN Codes L1 = 1575.42 MHz L2 = 1227.60 MHz Two PRN Codes P(Y): Military Code 267 day repeat interval Encrypted – code sequence not published Available on L1 and L2 C/A: Coarse Acquisition (Civilian) Code 1 millisecond repeat interval Available to all users, but only on L1 Code modulated with Navigation Message Data Provides ephemeris data and clock corrections for the GPS satellites Low data rate (50 bps)

Built by Lockheed Martin GPS Satellites Currently (January 2008) there are 30 GPS satellites in orbit All slots filled except PRNs 7 and 32 7 running off cesium oscillators 23 running off rubidium oscillators Oldest satellite is PRN 24, launched in July 1991 Block IIR/IIR-M Built by Lockheed Martin Launched 1997 - 2007 13 satellites are Block IIA 12 satellite are Block IIR 5 are Block IIR-M and transmit the new L2C carrier Two were launched in 2007 (PRNs 15 and 29) Block II/IIA Vehicles

GPS Monitor Station Network Five stations added in 2005, five more planned Monitor the GPS satellites for operational health Track the GPS satellites for orbit determination Upload satellite almanacs, ephemeris messages, and clock corrections Alaska United Kingdom St. Louis, MO Colorado Springs USNO Korea Hawaii Austin, TX Bahrain Kwajalein Ecuador 1. NGA operates a worldwide network of 11 unmanned fully automated GPS monitoring stations, represented in RED on the slide. We exercise command and control remotely from St. Louis. Our network provides coverage that complements and enhances the USAF network because of the geographic and numeric spread of our network. As I mentioned, we provide raw tracking data from our stations directly to the MCS 365 days/year, which is used to baseline (quality check) their post-processing analysis. -Note six of our stations are underlined. These have dedicated comm lines which provide NRT (near real time?) data flow to the NCC. More about this in a minute. 2. With NGA data, plus the data from the AF network, which they send to us 7/365, we also produce a set of post-fit orbital parameters (called a precise ephemeris) which consists of position, velocity, and time values for each satellite. This P.E. is used as THE DoD truth for GPS. For post-launch assessments, navigational assessments, or surveys, when anyone wants to know where a satellite actually was on a certain day at a certain time, they use the NIMA PE as the gold standard. The MCS uses it as their standard as well. 3. Posted on internet, SCEN, JWICS, Data to a select group of government and university research facilities, Radar Calibration ephemeris (U.S. Space Command - Cheyenne Mountain), Earth Orientation Parameter Prediction (correction for polar movement and earth rotation), Misc. data analysis as requested. Ascension Diego Garcia Tahiti South Africa Argentina Australia GPS Monitor Stations NGA Site (11) NGA Test Site (2) USAF Site (5) New Zealand

Corrections are uploaded to the clocks in space SPACE VEHICLE Broadcasts the SIS PRN codes, L-band carriers, and 50 Hz navigation message stored in memory SPACE-TO-USER INTERFACE CONTROL-SPACE INTERFACE MASTER CONTROL STATION Checks for anomalies Computes SIS portion of URE Generates new orbit and clock predictions Builds new upload and sends to GA MONITOR STATION Sends raw observations to MCS GROUND ANTENNA Sends new upload to SV

GPS Signals

GPS Modulation The carriers are pure sinusoids. Two binary codes are modulated onto them: the C/A (coarse/acquisition) code and the P (precise) code. Binary biphase modulation (also known as binary phase shift keying [BPSK]) is the technique used to modulate the codes onto the carrier. There is a 180 degree carrier phase shift each time the code state changes. The modulation requires a much wider frequency band than the minimum bandwidth required to transmit the information being sent. This is known as spread spectrum modulation. It allows lower power levels to be used.

Spread Spectrum Communication “Spreads" the power spectrum of the transmitted data over a wide frequency band Same principle is used for household cordless telephone (voice is the data) Each satellite is assigned unique Pseudo-Random Noise (PRN) Code Allows Multiple Access – All GPS satellites transmit at the same frequency but are identified by their PRN codes

Spread Spectrum Communication - II The signal transmitted by the satellites is the product of the navigation data, a spread spectrum code, and the RF carrier (either L1 or L2). In order to detect the GPS signal and recover the navigation data, the receiver must produce a replica of the PRN code to mix with the incoming signal. Thus, the firmware inside a GPS receiver has to be able to generate all 32 PRN codes and to match codes received over the air to the generated codes. The measured phase offset between the incoming and replica PRN code is the GPS range measurement.

GPS Signal in Space L1 Signal L2 Signal Frequency Spectrum P[dBW] 2.046 MHz L1 Signal C/A-CODE -160 P-CODE -163 f [Hz] 1575.42 MHz P[dBW] 20.46 MHz L2 Signal The Frequency picture of the L1 and L2 signal. See the suppressed carrier frequency. This is the received signal and this is the minimum power levels for the signal. Can see that the power levels are well below the thermal noise level of RX (??) 2.046 MHz ( +/- 1.023 MHz) carrier spread for C/A code and 20.46 MHz (+/- 10.23 MHz) for the P-code due to BPSK. Modulation - BPSK - bi-phase shift keying which reverses the phase of the carrier when digital PRN code foes from 1 to zero or vice versa. The signal is recovered by multiplying received satellite codes in the receiver with codes generated within the receiver and will collapse the signal into the original carrier frequency band. Then the signal power is back in a narrow frequency band and the user can demodulate the 50 Hz navigation message. The new GPS Civil Frequency Spectrum is still up for discussion. Some people propose to put it in the nulls of the P-code and some leave it the same. P-CODE -166 f [Hz] 1227.6 MHz 20.46 MHz Frequency Spectrum

CA/Code The C/A code stands for Coarse Acquisition It is available to anyone, worldwide, as part of the Standard Positioning Service (SPS) of GPS The C/A code is on the L1 carrier Timing specification is 40 ns, 95% of time, averaged for 1 day over entire constellation Used by most commercially available GPS receivers, including SIM network receivers

GPS L1 signal (C/A code) in Frequency Domain 1. Data Message Spectrum 0 Hz 50 Hz frequency 2. Data*Code Spectrum Signal is “Spread” 0 Hz 1.023 MHz frequency 3. Data*Code*Carrier Spectrum 1.023 MHz This is the transmitted signal 0 Hz 1.57542 GHz

GPS L1 C/A Signal (Time-Domain) +1 20ms GPS Data Message 50 bps -1 +1 Repeating 1023 Chip “Spreading Code” (20 per data message bit) 1.023 Mbps -1 Carrier 1.57542 GHz +1 -1 +1

How GPS provides position and time

GPS Positioning GPS-based positioning is fundamentally based on: The precise measurement of time The constancy of the speed of light GPS-based positioning uses the concept of trilateration GPS satellite positions are known Receiver position is not GPS-to-receiver range measurements are used to compute position

Positioning Example with 1 Transmitter Receiver (location unknown) Locus of points on which the receiver can be located Measured Range Transmitter (location known)

Positioning Example with 2 Transmitters True Receiver Location r1 r2 T1 T2 False Receiver Location

Positioning Example with 3 Transmitters True Receiver Location r3 T3 r1 T1 T2

GPS Positioning - II The position solution involves solving for four unknowns: Receiver position (x, y, z) Receiver clock correction Remember: Position accuracy of ~10 m implies knowledge of the receiver clock to within ~30 ns Requires simultaneous measurements from four GPS satellites The receiver makes a range measurement to the GPS satellite by measuring the signal propagation delay The data message modulated on the GPS signals provides the precise location of the GPS satellite and corrections for the GPS satellite clock errors

Pseudo-Random Noise (PRN) Codes Each GPS satellite transmits its own unique Pseudo-Random Noise (PRN) Code on L1 and L2 The C/A Code repeats every millisecond The receiver generates replicas of the C/A code and uses code correlation to distinguish between different satellites Feedback shift register - 10 stage C/A code 1023 bits long P(Y) P code with Anti-spoofing P code week long By using the time the code was sent and when it was received we can calculate the distance from the user to the satellite. ephemeris - location of satellite epoch - marks new period (repeat point of code) Show how use distances to get “good” time

Pseudorange Dt GPS transmitted C(A)-code Receiver replicated C(A)-code Finding Dt for each GPS signal tracked is called “code correlation” Dt is proportional to the GPS-to-receiver range Four pseudorange measurements can be used to solve for receiver position

Ranging ( xs, ys, zs, ts ) --- [m] Satellite PRN sequence Receiver ( x, y, z, t ) pr Receiver pseudo-range --- [s]

Although the primary purpose of GPS is to serve as a positioning and radionavigation system, the entire system relies on precise timing. After the receiver position (x, y, z) is solved for, the solution is stored. Then, given the travel time of the signals (observed) and the exact time when the signal left the satellite (given), time from the clock on the satellite can be transferred to the receiver clock. The measurement made by the GPS receiver reveals the difference between the satellite clock and the receiver clock by measuring the transit time of the signal: time of signal reception, (based on receiver clock, can be significantly in error) time of transmission, encoded in signal by GPS satellite clock (known precisely)

p = ρ + c × (dt − dT ) + dion + dtrop + rn This measurement, when multiplied by the speed of light, produces not the true geometric range but rather the pseudorange, with deviations introduced by the lack of time synchronization between the satellite clock and the receiver clock, by delays introduced by the ionosphere and troposphere, and by multipath and receiver noise. The equation for the pseudorange is p = ρ + c × (dt − dT ) + dion + dtrop + rn where p is the pseudorange c is the speed of light ρ is the geometric range to the satellite dt and dT are the time offsets of the satellite and receiver clocks with respect to GPS time dion is the delay through the ionosphere (an estimate can be obtained from the GPS broadcast) dtrop is the delay through the troposphere rn represents the effects of receiver and antenna noise, including multipath.

Finding Position & Time Two main factors determine accuracy of the position and time solution UERE (User Equivalent Range Error) DOP (Dilution of Precision) Overdetermined Least Square Estimator Matrix Inversion gives position error and time offset Once you have your position you only need one satellite to get time. SPS SA introduces errors - can get 20 to 50 ns according to RX specs but this is w/ smoothing over 10 - 60 minutes. NO SA - receiver position known -C/A or P code -smooth or raw data used SA 31.3 m UERE for PR ~= 104 ns * TDOP (1.7) ~= 175 ns 1 ns = 30 cm (~ 1 foot)

User Equivalent Range Error (UERE) The accuracy of the pseudo-range measurements between a particular satellite and a particular user UERE is the result of several factors the quality of broadcast signal in space, which varies from satellite to satellite and time to time stability of particular satellite’s clock predictability of the satellite’s orbit

URE Performance History Year

Dilution of Precision (DOP) second accuracy limiting factor depends on the geometry of satellites, as seen by the receiver critical for determining accurate position and time used in cooperation with the UERE to forecast navigation and time errors Good (Low) DOP Conditions: Second Accuracy Limiting Factor - independent of the quality of the broadcast signals and the type of user equipment set, as long as the same 4 satellites are used Three-dimensional User Navigation Error (UNE) (drms accuracy values)** UNE = UERE * PDOP One-Dimensional User Time Error (UTE) ** UTE=UERE * TDOP/c (speed of light) Poor (High) DOP Conditions:

The Future of GPS: New signals are being added to the broadcasts

L2C, a new civil GPS signal L2C code Enables higher civilian accuracy when combined with existing civil GPS signal (L1 C/A) Overcomes some limitations of L1 C/A Allows receiver to measure and correct for ionospheric delay Higher power reduces interference, speeds up signal acquisition, enable miniaturization of receivers, may enable indoor use Now broadcast by satellites launched since September 2005, available to entire constellation by about 2014

Third Civil Signal (L5) L5 code Begins with IIF sats First launch: 2008 New signal structure for better accuracy Higher power than other GPS civil signals Higher power (no less than -154.9 dBW) Wider bandwidth (1176.45 MHz +/- 10 MHz) Improves resistance to interference Co-primary allocation with Aeronautical Radionavigation Services at WRC-2000 (1164-1215MHz) Available to entire constellation by about 2016

L1C L1C Begins with GPS III sats First launch: ~ 2012 Modernized L1 civil signal In addition to C/A code to ensure backward compatibility Increased robustness and accuracy for civil users Designed with international partners so that it can work with other satellite navigation systems – will use same type of coding as Galileo Begins with GPS Block III First launch: ~2013; 24 satellites: ~2021

GPS Disciplined Oscillators

GPS Disciplined Oscillators (GPSDO) Operated as standalone time and frequency standards, often as the primary standard for the lab. GPSDOs discipline a local oscillator (quartz or rubidium) to GPS and serve as a self-calibrating standard that will perform at a high level if the receiver hardware and the GPS constellation are functioning normally. Short term stability limited by the local oscillator, long term accuracy and stability provided by the GPS signal Produce on-time pulse (1 pps). Produce standard frequency outputs (such as 1, 5, and 10 MHz). Some units produce frequencies used in telecommunications (1.544 MHz or 2.048 MHz, for example).

GPSDO (steered, no synthesizer) In this design, the local oscillator phase is continuously compared to GPS. A PLL is then used to “pull” the local oscillator into phase with GPS. The PLL is usually (but not always) software based. Performance depends on the response time of the PLL and the quality of the local oscillator.

GPSDO (unsteered, synthesizer) In this design, the local oscillator is used as the time base for a frequency synthesizer (called a translator in the diagram). The phase of the synthesizer is continuously compared to GPS and the frequency offset is measured. A correction is sent to the synthesizer to compensate for the frequency offset and eliminate the phase difference, but no corrections are applied to the local oscillator.

GPS Antennas Small and inexpensive, higher gain units (> 35 dB typical) generally used for timing to drive long cables. These antennas are normally active, with internal amplifiers powered by 5 V dc from the antenna cable. Most bring the 1575 MHz L1 carrier straight to the receiver, without any down conversion. Omnidirectional, need to have clear sky view on all sides for best results.

GPSDOs are easy to install and use The antenna needs to be mounted on a roof, with a clear view of the sky on all sides. Make sure that the antenna has enough gain to drive the antenna cable. A combination of high-gain antenna, low-loss cable, and inline-amplifiers might be needed for some long cable runs. Survey position. You can enter the position, but most receivers can do a self survey of their antenna position. This might take up to 24 hours. The GPSDO always uses the same coordinates after the survey, because the antenna is not moving.

Getting Time from a GPSDO

What is GPS Time? Controlled by the United States Naval Observatory (USNO), but not exactly the same thing as UTC(USNO). GPS time differs from UTC by the number of leap seconds that have occurred since the origination of the GPS time scale (January 6, 1980); this value is equal to 14 s as of January 2008, it will increase each time there is a leap second. The navigation message contains a leap second correction, however, and GPS receivers automatically correct the time-of-day solution. It also differs from UTC(USNO) by a small number of nanoseconds that continuously changes. The current difference between the UTC(USNO) estimate and GPS time is also in the navigation message, and this correction is applied to the 1 pps signal. After the leap second and UTC(USNO) corrections are applied, GPS time as broadcast is nearly always within 30 nanoseconds of UTC, UTC(NIST), and UTC(USNO). This makes it the best estimate of UTC being broadcast anywhere, and it is available free of change to anyone, worldwide. The frequency offset between the UTC(USNO) estimate and UTC(NIST) is very small, typically a few parts in 1015 or less when measured over a one month interval.

GPSDOs get the time from one-way time transfer Need position of GPS receiver from antenna survey Errors introduce a diurnal variation (from coordinates, multipath, ionospheric delays, temperature, etc., but variations are usually small) Receiver applies corrections from 50 Hz GPS navigation message Position, velocity of satellite Broadcast ionospheric model Satellite clock offset from UTC(USNO) GPSDO outputs local clock offset from GPS broadcast of UTC(USNO) ionosphere troposphere Now the receiver can convert each value of the time difference between the receiver and the satellite clocks (obtainable from C/A code PR measurement) to a value of the time difference between the receiver clock and GPS time.

No more SA, a Great Time Source Gets Even Better GPS performance improved dramatically after May 2, 2000, when the Selective Availability (SA) program was deactivated, removing the intentionally introduced jitter from the signal SA had been previously implemented by the United States government for reasons of national defense.

Factors that limit the uncertainty of GPS time received by users Errors in the antenna coordinates, particularly errors in elevation Poor estimates of cable, antenna, and receiver delays Multipath Ionospheric and Tropospheric corrections Changing delays in hardware and cables due to temperature and other environmental factors All of these factors make it difficult to validate time received from GPS to within better than about 50 ns. The typical uncertainty limit is about 100 ns, and nearly all receivers will provide time within 1 microsecond with any attention to the details.

Uncertainty due to Antenna Coordinates GPS computes dimensions in Earth-Centered, Earth-Fixed X, Y, and Z coordinates. Position in XYZ is converted in the receiver to geodetic latitude, longitude, and elevation. Errors in the X,Y,Z coordinates translate to timing errors as large as 3 nanoseconds per meter, depending upon the satellite’s position in the sky. GPS excels at finding horizontal position (latitude/longitude) Most receivers can quickly survey latitude/longitude to within 10 meters, and to < 1 meter after several hours of averaging. GPS is weak at determining vertical position (elevation) GPS provides distance from the center of the earth and then by using the radius of a model of the Earth’s surface, provides elevation. There is nearly always a bias in the elevation. The vertical position error is usually several times larger than the horizontal position error. A 10 meter altitude error (timing error of up to 30 nanoseconds) is not uncommon, even if the receiver surveys the antenna coordinates by averaging position fixes for several hours or more.

Average position error of TMAS survey was 5 Average position error of TMAS survey was 5.37 m, with nearly all of this error (5.30 m) in the vertical position

Uncertainty due to Antenna/Cable Delays Most GPSDOs allow you to enter a delay constant for the antenna cable. The antenna cable delay be easily measured with a time interval counter, or estimated fairly accurately using data from the cable vendor. However: The antenna delay is very difficult to measure unless the entire system (receiver, antenna, and cable) is calibrated by NIST or another timing laboratory as a unit. Some GPS receivers advance the on-time pulse to at least partially compensate for the cable delay. Thus, the delay entered into the receiver should actually be less than the cable delay. Other problems arise to impedance mismatches, connectors, etc. For all of the above reasons, it is unlikely that the cable delay can be estimated to better than 10 nanoseconds, unless the manufacturer provides guidance or instructions.

Uncertainty due to Multipath Multipath is caused by GPS signals being reflected from surfaces near the antenna. These signals can then either interfere with, or be mistaken for, the signals that follow the straight line path from the satellite. If the antenna has a clear, unobstructed view of the sky, the uncertainty due to multipath is usually very small (a few nanoseconds or less), but some uncertainty due to multipath is nearly impossible to avoid and detect.

Uncertainty due to Environment Although not really of concern for most applications, the receiver, antenna, and cable delays can change over the course of time, sometimes by as much as several nanoseconds. This is usually due to temperature. Receivers probably have the most sensitivity to temperature, but they are normally located in a laboratory with a relatively small temperature range, so this is usually not a problem. The antenna and cable are outdoors, and the temperature range can be very large over the course of a year. Sometimes temperature compensated antennas are used (see photo), but that is usually not practical. If possible, however, use a high quality antenna cable with a low temperature coefficient.

Uncertainty due to ionospheric conditions The ionosphere is the part of the atmosphere extending from about 70 to 500 km above the earth. The troposphere is the lower layers of atmosphere, where clouds form. When radio signals from the satellites pass through the ionosphere and troposphere, their path is bent slightly, changing the delay. The delay changes are largest for the satellites at low elevation angles. GPS broadcasts a ionospheric and a tropospheric correction, which most receivers automatically apply (or which can be turned on by the user). This reduces the effect. These corrections are called modeled ionospheric corrections, or MDIO For the very best results, the ionospheric conditions are measured at a receiving location on the ground by a dual-frequency GPS receiver (one that receives both L1 and L2). These measurements are used to generate local corrections that are used in place of the broadcast corrections. In some cases, this can reduce the uncertainty by a few nanoseconds. These corrections are called measured ionosphere corrections, or MSIO.

Interference GPS Signals are easy to interfere with, or “jam” due to their low power levels Under certain condition, the amplifier in a GPS antenna will oscillate in the GPS frequency bands, resulting in interference to adjacent GPS equipment The owner may not notice the antenna is in oscillation Locating the interference is difficult due to the intermittent nature of the oscillation

Interference from a GPS Antenna Antenna oscillated around 1.579GHz (about -90dBm) in low temperature, affecting all the GPS antennas in view All the GPS antennas in the line of sight were affected by the interference. The interference had different effect on close by receivers (the CV-6 vs. CV-11 plot). Symptoms include: data outage, large data variation.

GPSDO Time Uncertainty The time uncertainty relative to UTC usually less than 1 s for nearly any receiver. An uncertainty of less than 100 ns can usually be achieved if the receiver and antenna delays are calibrated. It is difficult to prove an uncertainty of < 50 ns even if the GPSDO has been calibrated. Time stability is very good, often just a few nanoseconds at 1 day, as measured using TDEV. Now the receiver can convert each value of the time difference between the receiver and the satellite clocks (obtainable from C/A code PR measurement) to a value of the time difference between the receiver clock and GPS time.

Getting Frequency from a GPSDO

Getting Frequency from GPSDOs Antenna survey doesn’t matter as much for frequency. Altitude error has no real effect on frequency. Lat/Lon errors will cause bigger short-term variations, but average frequency over long intervals won’t change substantially. Quality of local oscillator matters, particularly when signal is lost and receiver goes into “holdover” mode. The lowest priced units use TCXOs or even simpler crystal oscillators Better units use OCXOs The best units for holdover use rubidiums Some GPSDOs are optimized for short-term stability, others for long-term stability. Others are good for timing, but not so good for frequency.

OCXO with and without GPS

GPSDO “optimized” for short-term stability

Test of GPSDOs vs. UTC(NIST) NIST tested four commercial GPSDOs in the spring and summer of 2005 against the UTC(NIST) time scale. Two had rubidium time bases, two had OCXO time bases. We measured the 1 pps and 10 MHz outputs at the same time with two different measurement systems. Results showed very different behavior amongst different GPSDOs.

Test of GPSDOs vs. UTC(NIST)

10 MHz Phase of GPSDOs vs. UTC(NIST)

10 MHz Stability of GPSDOs vs. UTC(NIST)

Common-View GPS

Common-View Time Transfer

Common View GPS Two users, A and B, compare their clock to the same satellite at the same time. Two data sets are recorded (one at each site): Clock A - GPS Clock B - GPS Data is exchanged. Subtract A from B to get difference between clocks. Common errors cancel. Used for international computation of UTC and by SIM Network

All-in-view GPS ionosphere troposphere Receivers at remote stationary locations track all the satellites in view Each receiver makes the all-in-view measurements, (REFstation_i – GPS): time difference between a local reference clock and the received composite timing signal from all the satellites being tracked The all-in-view measurements from two receivers are differenced to obtain the time and frequency difference of two remote clocks Works when no satellites are in common-view Performance is about the same as common-view for short baselines (2500 km or less), better than common-view for long baselines (5000 km or longer) ionosphere troposphere A B

Keys to Successful Common-View Measurements Same type of receiver (manufacturer and model) should be used at each site All antenna, cable, and receiver delays must be calibrated and used as delay constants The antenna at each site should be surveyed with the least amount of uncertainty possible, ideally less than 1 meter for both 2D and 3D position. A good survey of altitude is especially important. The ionospheric, tropospheric, and multipath delays should be nearly equivalent at each site. For the absolute best results, corrections for these delays can be generated at each site. The relative delays at both sites should be as close to zero as possible.

Common-View works best if the same model of GPS receiver is used at both sites

Sample Common-View Track

Sample Common-View Track (A-B)

SIM Receiver Calibrations SIM systems are calibrated at NIST prior to shipment. Calibrations are performed using the common-view, common-clock method. The SIM laboratory installs the same antenna cable and antenna that were used during the calibration. Calibrations last for 10 days. The time deviation (Type A uncertainty) of the calibration is less than 0.2 ns after one day of averaging. The combined uncertainty is estimated at 4 ns, because a variety of factors can introduce a systematic offset.

Uncertainty Component Estimated Uncertainty SIM Network Uncertainty Analysis Type B Uncertainty Component Explanation Estimated Uncertainty Calibration of SIM unit at NIST Absolute accuracy of delay calibration is limited to about 4 ns. 4 ns Environmental variations Receiver delays can change due to temperature or voltage fluctuations from antenna cables or power supplies. 3 ns Antenna Coordinates Error Assumes that antenna position (x, y, z) is known to within 1 m. Propagation delay changes due to multipath Multipath is caused by GPS signals being reflected from surfaces near the antenna. 2 ns Propagation delay changes due to ionospheric conditions The SIM system uses the ionospheric corrections broadcast by the satellites, and does not apply measured ionospheric delay corrections. This uncertainty represents the typical difference between the modeled and measured correction. Cable delay measurements made by the SIM laboratory. Usually done with a time interval counter and is subject to small errors. 0.5 ns Resolution Uncertainty Software limits the resolution of entered delay values to 0.1 ns. 0.05 ns Uncertainties are expressed using a method complaint with the ISO GUM standard. We use the time deviation (TDEV) at an averaging time of 1 day as our Type A uncertainty (1.5 ns in this example). Type B uncertainties are summarized in the table. Combined standard uncertainty (k = 2) is < 15 nanoseconds for time, and < 1  10-13 for frequency after 1 day of averaging.

CGGTTS Format for submission to BIPM

BIPM Circular T (www.bipm.org) Published monthly, it contains the official results of international time comparisons. Five labs in the SIM network have their standards listed on the Circular-T. The Circular-T numbers are post processed and obtained with completely independent receiving equipment. The real-time numbers obtained through the SIM network are in good agreement with the Circular-T numbers, well within our stated uncertainties. This helps validate our results.

GPS All-in-view Primary method for most national timing centers in the world to contribute clock data to the computation of International Atomic Time (TAI) and Coordinated Universal Time (UTC) PTB in Germany is the pivot laboratory Coordinated by the BIPM (Bureau International des Poids et Mesures located near Paris, France) About 60 laboratories contribute by submitting data in the CGGTTS format* * Consultative GPS and GLONASS Time Transfer Sub-committee

Multi-channel Common-view Track Schedule Starting at 0:00 (UTC) on the reference date (October 1, 1997), the 24 hours of a day are divided into 90 16-minute intervals. The first 89 intervals are used for common-view. Start time of each 16-minute interval is shifted 4 minutes earlier everyday. The 90th interval is reserved for handling the 4-minute start time update. The 13-minute common-view measurement starts 2 minutes after the beginning of the 16-minute interval. The multi-channel common-view track schedule contains the single channel common-view track schedule as a subset. lock up measurement data processing 2 13 1 90 1 2 3 4 89 1 2 t 0:00 0:16 0:32 0:48 1:04 23:28 23:44 23:56 0:12 0:28 Day 1 Day 2

The CGGTTS Common-view Data Format GPS RCVR: NBS10 V9809 MJD= 51658 YR=00 MONTH=04 DAY=24 HMS=14:47:20 (UT) GGTTS GPS DATA FORMAT VERSION = 01 REV DATE = 2000-04-03 RCVR = NBS10.................... CH = 01 IMS = 99999 LAB = NIST X = -1288398.27 m Y = -4721698.10 m Z = +4078625.68 m FRAME = ITRF.... COMMENTS = NO COMMENTS.............. INT DLY = 53.0 ns CAB DLY = 0199.9 ns REF DLY = 0066.7 ns REF = UTCNIST CKSUM = 74   PRN CL MJD STTIME TRKL ELV AZTH REFSV SRSV REFGPS SRGPS DSG IOE MDTR SMDT MDIO SMDI CK hhmmss s .1dg .1dg .1ns .1ps/s .1ns .1ps/s .1ns .1ns.1ps/s.1ns.1ps/s 3 08 51655 105800 780 380 760 -1058301 -1131 -571 -1098 415 163 107 +2 76 +0 02 8 32 51655 111400 780 319 2933 -7071115 -3061 -246 -3082 290 074 125 -20 85 -9 34 13 28 51655 113000 780 415 3083 +6965884 -30 -94 -241 625 019 100 -12 71 -7 FB 3 74 51655 114600 780 296 530 -1058331 +929 -503 +962 470 163 133 +19 92 +24 17 31 08 51655 121800 780 498 706 -7572 -400 -197 -390 470 180 87 +4 99 +14 DD 13 32 51655 123400 780 569 2693 +6966345 +171 -440 -40 424 011 79 +0 90 +9 F0 18 68 51655 125000 780 279 1829 -341335 +18 -132 +22 698 182 141 +35 152 +44 16 31 74 51655 132200 780 283 472 -7436 +2669 -73 +2678 441 206 139 +29 190 +36 24

BIPM-Compatible Common-view Receivers A single channel GPS receiver records a maximum of 48 common-view tracks each day according to a schedule. The receiver starts tracking the satellite 2 minutes before the beginning of each 13-minute measurement. Receiver models include: AOA TTR-6 (might be discontinued) A multi-channel receiver continuously tracks all the satellites in view. For each satellite tracked, the receiver groups the measurements into the 13-minute interval according to the multi-channel schedule. Receiver models include: AOS TTS-2 AOS TTS-3 (dual frequency) NPL TimeTrace Novatel (dual frequency) PolaRx2eTR (dual frequency) SIM system with conversion software (possibly)

GPS Performance Comparison

Other Satellite Navigation Systems GLONASS (Russia) Galileo (European Union) QZSS (Japan) COMPASS (China) India and Australia among other nations developing GNSS augmentations

GLONASS: GLObal NAvigation Satellite System Operated by the Coordinational Scientific Information Center of the Russian Federation Ministry of Defense First satellite launched in 1982, constellation still not fully populated today Russian Government has renewed commitment to replenish and modernize the GLONASS constellation

GLONASS Constellation Completed system would have 24 satellites in 3 orbital planes ascending nodes 120 degrees apart 8 satellites equally spaced in each plane Two frequency bands L1 = 1602 + n*0.5625 MHz L2 = 1246 + n*0.4375 MHz Where n is frequency channel number (n=0,1,2,…) Circular 19,100 km orbit inclination angle of 64.8 degrees Cesium clocks on board satellites

Key Differences between GLONASS and GPS Signal Structure: GPS: CDMA (different satellites transmit different PRN codes on the same carrier frequencies) GLONASS: FDMA (different satellites transmit the same PRN codes on different carrier frequencies) Period of Satellite Orbit: GPS – 11:58 (same satellite can be observed at the same position, with same velocity every sidereal day -- 23:56) GLONASS – 11:15 (satellite of the next slot of the same plane can be observed at about the same position every sidereal day) PRN Code for Civilian User: GPS (L1) C/A code, chipping rate: 1.023MHz (L1) C/A code, chipping rate: 511KHz

GLONASS Status: January 2008 13 operational satellites Plane I : 4 satellites operating Plane II: 4 satellites operating Plane III: 5 satellites operating Four satellites were launched in 2007 Russian Space Agency Information Analytical Centre: http://www.glonass-ianc.rsa.ru

Performance of GLONASS Time GLONASS time, UTC(SU) is not nearly as accurate as GPS. GLONASS time is canceled in the common-view method and GLONASS CV is an accepted BIPM time transfer method.

GALILEO European Global Navigation Satellite System Galileo is a joint initiative of the European Commission (EC) and the European Space Agency (ESA). Completed system will have 30 satellites Only 3 orbital planes compared to 6 for GPS Will offer a basic service for free (open service), but will charge user fees for premium services. T First signal in space transmitted by GIOVE-A satellite which became operational in January 2006 Should be operational around 2013-2014

Constellation Configuration

Galileo System Design Galileo System Time (GST): Shall be a continuous coordinate time scale steered towards the International Atomic Time (TAI) with an offset of less then 33 ns. Offset between GST and the GPS system time is monitored and broadcast to users, but may also be estimated in the receiver. Each Spacecraft will have 4 onboard clocks 2 Rubidium Vapour 2 Passive Hydrogen Maser

Galileo Frequency Structure