Logo of the Physikalisch-Technische Bundesanstalt
View of the Mainflingen long-wave radio station for transmission of the DCF77 signal:  transmitter building (in the back), antenna building (yellow bricks) and antenna masts.

GPS time comparisons

The Coordinated Universal Time, UTC, is realized in cooperation with approximately 50 time institutes globally distributed and the International Bureau of Weights and Measures (BIPM), in Sèvres. This cooperation is based on the regular comparison of the atomic time scales UTC(k) realized in the individual institutes (k). For four decades, special GPS time receivers have been used for this purpose. Originally developed at the National Bureau of Standards, USA, (today: National Institute of Standards and Technology, NIST), they have later been industrially manufactured and further developed.

GPS: Overview, signal

The NAVSTAR Global Positioning System GPS is a military satellite navigation system, composed of the space segment (24 satellites in 6 orbital planes, 55° inclination, 20183 km orbit height above ground, 2 orbits during a sidereal day) and the control segment (5 monitor stations, Master Control Station in Colorado, USA). The GPS signal is of course primarily used for position determination. In addition, GPS has developed into an excellent auxiliary means for time dissemination and comparison of time scales, which become clearly by understandable looking at the quality of the available signals. In each satellite, an atomic clock (based on rubidium or caesium) furnishes a 10.23 MHz reference signal. Each satellite transmits navigation signals on the frequencies L1 and L2 (L1 = 1575,42 MHz = 10.23 MHz × 154. L2 = 1227.6 MHz = 10.23 MHz × 120). On both carrier frequencies, pseudo-random noise codes (PRN codes) are transmitted, in which the information relevant to navigation and time transmission is encoded. The L1 signal is modulated with the P code (P for precision, chip-rate 10.23 MHz, repetition rate 267 days) and the C/A code (C/A for coarse/acquisition, chip-rate 1.023 MHz, repetition rate 1 ms), the L2 signal with the P code. Each satellite transmits a code characteristic of it which is synchronized with the respective satellite clock. In a GPS receiver, a signal with the same PRN modulation as on the signal received and synchronized with the receiver clock, is generated. This signal is temporarily shifted until maximum correlation between the two PRN signals is recognized at a temporal displacement τPR. The quantity τPR × c (c = light velocity) is also referred to as pseudo-range, as it indicates the apparent distance between satellite and receiver antenna.
It is only after the maximum correlation has been found that the proper transmission of data to the receiver can take place which is necessary for interpretation of the τPR value found. The L1 and L2 signals are continuously modulated with a modulation rate of 50 bit/s in so-called “frames“ of 1500 bits. The complete navigation message is contained in 25 frames whose transmission requires 12.5 minutes. It contains, among other data, the relation between the individual satellite clock and the system time T(GPS) (see below), the satellite positions (ephemerides), current modelled parameters of the ionosphere and the so-called almanac, the catalogue of available satellites with their plane data.

GPS system time

The GPS system time is derived from the clock ensemble of the monitor stations. It is a continuous time scale without leap seconds with 00:00 h UTC(USNO), January 6, 1980 as zero point (USNO: United States Naval Observatory, Washington DC). At the end of 2022, the time difference T(GPS) - UTC(USNO) amounted to 18 s, apart from that, the deviations are clearly below 20 ns. UTC(USNO) and UTC published by BIPM typically deviate from each other by less than 10 ns. The navigation message contains both the time difference in integral seconds between T(GPS) and UTC(USNO), which changes by one second when a leap second is introduced in UTC, and also the remaining time difference T(GPS) - UTC(USNO) predicted for the current moment (in nanoseconds). With this information, the GPS signal represents a globally uniform time signal which can be used for dating as well as for the control of standard frequency oscillators. For this purpose, a wide palette of commercial devices is available on the market.

GPS time comparison

For the performance of time comparisons, a GPS receiver is synchronized with UTC(k) in each time institute (k). Then the time difference UTC(k) - T(GPS) results from the primary measurand τPR by

UTC(k) - T(GPS) = τPR - (R/c + τS+ τIT) τL+ τSV   (1)


    R :

    distance between satellite and GPS antenna, calculated from the satellite position transmitted and the antenna position assumed to be known,

    c :

    velocity of light,

    τS :

    correction with respect to the rotation of the Earth during time R/c (Sagnac effect),

    τI : additional propagation time due to the refractive index of the ionosphere for electromagnetic waves of L1 or L2 frequencies deviating from 1,
    τT : similar to τI, here caused by the troposphere, but frequency-independent (in the range L1, L2) and calculated on the basis of a model (per software),
    τL : internal signal delay of the GPS signal between antenna and measuring facility for the determination of τPR,
    τSV : predicted time difference of the individual satellite clock and T(GPS) at the moment of the measurement.

The correction terms, except for τL, can be calculated from the information contained in the navigation message, τL must be determined by calibration. When two institutes a and b determine, for example, time differences in accordance with equation 1) from observations of all satellites during one day and exchange the results, this allows the mean time difference UTC(a) - UTC(b) to be calculated.

Selective Availability

The term “selective availability“ (SA) refers to the option to transmit in the above-mentioned navigation message deterministically incorrect information, which is corrected again by selected (military) receivers but leads to a reduced accuracy in the case of civil receivers. For many years, this measure affected the transmitted relation between satellite clock and GPS system time. In time comparisons, SA could be avoided by the so-called “common view“ procedure in which the two institutes a and b of our example determine the measurand (1) using the same satellite at the same moment (exact to the second). The NBS-type single-channel time receivers were used accordingly and, with 13 minutes, the observation time per satellite was selected exactly so that the complete navigation message could be received in the meantime.

The CV mode makes sense even without SA, as then τSV in (1) is in every case equal for both stations and cancels out in the difference formation. As long as SA was activated (until and including May 1, 2000), the CV mode was practically indispensable.

Modern GPS time receivers have several channels evaluated in parallel so that in a specified time window the time difference between two institutes can be determined via up to 6 satellites within Europe and via up to 4 satellites between Europe and America.