DAVID W. ALLAN Allan's Time Inc.
NEIL ASHBY University of Colorado &
CLIFF HODGE National Physical Laboratory
Precise timing is at the heart of much of modern society more so than most people may realize. In this evolving Information Age, precise timing is used to manage information flow through many network nodes so that it is reliable robust and inexpensive.
Unprecedented navigation accuracy is now right at our fingertips using hand-held receivers that are part of the Global Positioning System, which is based on a synchronized atomic clock system.
Literally billions of timing devices go into service each year. Understanding how they all play together is an intriguing story and adds to an appreciation of "what makes things tick." What is time based on? How do we know we have set a clock correctly? Where is the best clock in the world? How can we get the "right time"?
Practical precise timing in the modern era came with the invention of the quartz crystal oscillator and quartz-crystal filters in the I 920s. Atomic clocks were invented in the '40s, and since then their accuracy has on average doubled every two years-interestingly, the same rate of improvement as for computer memory density.
The most accurate measurement known to humanity now is time-related-the duration of a second. The ultimate international reference for accurate time and frequency is Coordinated Universal Time or UTC (Temps Universal Coordunne' in French), which receives its time and frequency input from timing centers around the globe.
In 1970 the Coordinated Universal Time system was devised by an international advisory group of technical experts within the International Telecommunication Union (ITU). The ITU felt it was best to designate a single abbreviation for use in all languages in order to minimize confusion. Since unanimous agreement could not be achieved on using either the English word order, CUI or the French word order, TUC, a compromise of using neither, UTC, was adopted. Currently, the best accuracy for the determination of the second for UTC is equivalent to +1 second in 10 million years.
Because time and frequency can be measured so accurately, they are often used to determine other fundamental quantities, such as the volt, ampere, ohm, and meter. For example, in 1983 the General Conference of Weights and Measures (Conference Cenerale des Poids et Mesures), in Paris, redefined the meter as "the length of the path traveled by light in vacuum during a time interval of 1/299 792 458 of a second." It is expected that all of the base units in metrology for the support of technological development will eventually be traceable back to the second.
Perhaps there is no better example of how precise timing is used for accurate navigation, positioning, and time transfer than
the Global Positioning System (GPS). The system features a set of 24 orbiting satellites, each with a synchronized atomic clock on board, which transmit signals carrying precise time tags. At a GPS receiver, the computer on board uses a precision quartz-crystal clock to calculate the time of arrival, and hence the time of flight, of the signals from each of the observable satellites.
Since the signals travel at the speed of light, and the transmission times are known, the receiver's computer can turn the time of flight into a very accurate estimate of the distance to each satellite. In this process, the high accuracy of the GPS atomic clocks is transferred to the clock inside the receiver. The result: easy access to highly accurate position and time measurements, which a multitude of users are capitalizing on.
In fact, the number and variety of uses of GPS have exploded, transforming almost every endeavor in which accurate position and time are needed. For example, the January 1998 issue of GPS World magazine highlights a receiver survey that lists over 400 kinds of receivers available from more than 70 companies. In 1996, well over a half-million GPS receivers were marketed by Japan alone, many of them for vehicle tracking. In addition, many types of massive engineering and cbmmunications systems are now being developed that largely depend on GPS. Many countries, for instance, are planning to use GPS with augmentation from other systems for navigation and control of aircraft. ecl3


Atomic clocks in space in lock-step with Coordinated Universal Time disseminate precisely the right time round the world


What is time?

No unique "true" time derivable from natural phenomena exists. So, by international agreement, Coordinated Universal Time was set up as the world's official time-by definition. The word "coordinated" means that UTC is generated through the cooperation of about 50 nations that are now signatories of the Convention of the Metre, which was first established by agreement of 17 nations in 1875. Timing data is supplied to the Bureau International des Poids et Mesures, near Paris, by responsible timing laboratories and observatories around the world. In turn, timing experts at the bureau combine that data into an official time scale, designated as UTC.
Timing and navigation almost always go hand in hand. Accurate longitude determination was a critical maritime issue during the 16th and 17th centuries. Because the earth rotates, the accurate determination of longitude requires determination of the time-that is, how much the earth has turned up to the moment of measurement. It was known as early as 1530 that having a good clock on board ship, were it possible, would be a tremendous aid to navigation.
In 1675 the observatory at Greenwich, England, was established to conduct the astronomical observations needed by ships at sea to determine their longitude. During the 17th century, several nations offered prizes for an acceptable method of making this determination. In 1714 the British Parliament offered a prize of 20000 pounds to ships sailing from England to the West Indies for the measurement of the longitude of the journey-accurate to within 30 nautical miles. This goal translates into a clock that can keep time to better than 3 seconds per day. The difficulty was the sea, whose motion precludes the use of pendulum- based devices.
Coming to the rescue was clockmaker John Harrison, who through his persistence and genius, proved that a seaworthy chronometer could be built that could determine accurate longitude. Harrison's chronometers overcame harsh environmental problems and worked to within about a second per day. The ship's log of Captain James Cook, who used a copy of a Harrison chronometer while exploring the South Pacific from 1772 to 1775, expresses great praise for the advantage gained in using the clock: "our never failing guide, the Watch."
In 1884 a treaty agreed to by 25 countries established Greenwich Mean Time (GMT) as the world's official time and the meridian through Greenwich Observatory as the prime meridian [Fig. 1]. GMT, the local time at the observatory, was determined using astronomical measurements. Each nation then adjusted the time for its own time zone(s) by adding or subtracting the right number of hours.
As clocks have improved and as needs have changed, Greenwich Mean Time has evolved into UTC, which relies on atomic rather than astronomical clocks. Although UTC is the correct time by definition and convention, there are many sources of time. The question, "What time is it?" should be changed to, "Whose time is it? - especially if there is an interest in the precision and accuracy of the source and how well that source ties in to UTC.
Understanding how Coordinated Universal Time.is generated requires some grasp of clocks and timekeeping techniques. Almost any clock may be considered as a two- part device [Fig. 2]. One part is an oscillating device for determining the length of the second or some other desired time interval. This is usually referred to as the clock's frequency standard, which oscillates at some rate determined by physical laws.
Historically, the pendulum was the classic time interval standard (although Galileo used his own pulse in many experiments). A pendulum's frequency depends on its length and the gravitational acceleration at the pendulum's location. An ideal pendulum that swings through its lowest point once per second will have a length of approximately 1 meter. A pendulum's frequency will, however, depend on its environment since the pendulum's length will change when the temperature changes.
Currently, the typical wristwatch has as its frequency standard a quartz-crystal oscillator with a frequency of 32 768 Hz. This number of oscillations is convenient for the associated digital electronic circuit, because if the number is divided by 2'~, which is easy for a digital chip divider, the result is one pulse per second.
An atomic clock supplies a much more accurate frequency than a pendulum or a quartz crystal.The atomic clock uses as its reference the frequency associated with a quantum transition between two energy levels in an atom. For a given quantum transition, the photons emitted or absorbed have a unique frequency proportional to the energy difference between the two energy levels, with little variability around this value. The achievement of atomic clock metrology is to harness the frequency of these photons while affecting the natural atomic resonance only slightly.
The current official definition of the second is much more elegant than that based on the environmentally sensitive pendulum. At the 13th General Conference of Weights and Measures (Resolution 1, 1967), the second was defined as "the duration of 9 192 631 770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the cesium-133 atom." In other words, one official second elapses when the defined number of cycles transpiring for the electromagnetic signal associated with the photon are either given off or absorbed by this quantum transition.
The second part of a clock is a counter that keeps track of the number of ticks or clock cycles that have occurred since the clock was set. This part is represented, for example, by the gears and clock face in a pendulum clock, or by an accumulating register and a display in an atomic clock. After being set initially, the clock provides its estimate of the correct time.
In principle, if a clock were set perfectly and if its frequency or rate remained perfect, it would keep the correct time indefinitely. In practice, this is impossible for several reasons. One is that the clock cannot be set perfectly. Another is that random and systematic variations are intrinsic to any oscillator, and when these random variations are averaged, the result is often not well behaved. Also, time on a clock depends on its position and motion (there are relativistie effects). And finally, but invariably, environmental changes cause the clock's frequency to vary from the ideal. So, if a clock is measured with sufficient precision, its reading will almost never agree with UTC.
To compensate for some of these timing  instabilities, modern techniques can be used to couple a clock with sensors and a microprocessor.  The quartz crystal oscillator provides a cost-effective means of achieving reasonably good clock stability, and can be very effective if properly interfaced with a computing capability and set of sensors.

A quartet of quality measures

Four useful measures help to describe a clock's quality: frequency accuracy, frequency stability, time accuracy, and time stability. These measures are not all independent. Frequency (or rate) accuracy measures how well the clock can realize the defined length of the second.
One yardstick that is commonly used is the change in the error of a clock's time divided by the elapsed time over which the change occurred.
The goal of Harrison's chronometers was to have a frequency accuracy of better than 3 seconds per day Modern primary frequency standards have frequency accuracies of about a nanosecond per day.
Frequency stability, on the other hand, measures the change in frequency from one time interval to the next. A particular value of this time interval, T, is chosen, called the averaging time. The frequency stability is then ascertained for different values of T. A clock can have a high frequency error and still be very stable, that is, the frequency or rate error stays about the same. For example, a clock may have a rate inaccuracy and gain I second a day, but if that rate remains exactly the same, it would have perfect frequency stability, and consequently perfect time predictability.

Time accuracy means how well a clock agrees with UTC. Often, what is needed is time accuracy at several locations in a system or network, as with the GPS. The GPS time broadcast by each of the satellites must be synchronous with that of all the other satel - lites in the constellation, to within a small number of nanoseconds, in order for the system to work properly. This is accomplished by having atomic clocks on board the satellites and then synchronizing the clocks.

Time stability is correlated with frequency stability since the time error of a clock is proportional to the integral of its frequency error. But time stability is often useful as a measure of change with respect to some uniform flow of time in a time measurement system or in a time-distribution or dissemination system. If the time or frequency errors of a clock can be estimated, then compensating corrections can be made.
If the time and rate (frequency) of a clock with good stability are calibrated against some better clock, the former clock can provide an estimate of the better clock's reading, should the better one not be immediately available. In a hierarchy of calibrations, the best clocks and frequency standards are often referred to as "primary."
In determining the accuracy of some primary frequency standards, all factors affecting the accuracy are evaluated to obtain a best estimate of the second. Often this procedure precludes these standards from running continuously, as is needed for clock operation. In such a case, a secondary clock that has been calibrated in rate by a primary frequency standard is used. The technique-used extensively in the generation of UTC-allows the secondary clock to perpetuate an estimate of the time of the primary frequency standard as if it had been operating as a clock on a continuous basis.
An important modern timekeeping concept focuses on the power df properly used clock ensembles, that is, sets of many clocks. Historically, astronomical time was based on the movements of one earth, one moon, one sun, one solar system, and one celestial sphere of moving stars and planets. Even though there was only one of each, reliability was not an issue,- the earth was not expected to stop spinningi. With the introduction of atomic clocks, many timekeepers, each of which could generate its own estimate of time, came into existence. The concept of a clock ensemble has great merit for reliability since one clock could quit working and for improved average performance.
From measurements of the time difference between just two clocks, it is impossible to tell which one is deviating. Three independent clocks allow an independent estimate of the stability of each, but four are needed in case one quits. Common systematic errors can be removed by comparison with primary frequency standards. Once the individual clock stabilities are known, optimum weights can be assigned to each of the clocks.
What's more, it has been shown that if the clocks are well-characterized and optimum weights are given to them in a properly developed combining algorithm, the weighted ensemble can perform better than the best clock in the set. If one clock fails or has a bad reading, it can be detected and rejected so that it does not perturb the ensemble's output. These ensembles can generate a highly reliable and stable real-time output. The clock ensemble at the National Institute of Standards and Technology (NIST), in Boulder, Cob., has been running in such an ensemble mode since 1968.


Distortions in timekeeping

Deviations in the observed time of a clock can be caused by measurement noise internal clock deviations, and external environmental perturbations affecting the clock. Measurement noise arises from imperfections in the system that is used to observe the clock readings. Internal clock deviations can arise from deviations either in the clock's frequency standard or in the counting mechanism that keeps~track of the hours, minutes, seconds, and so on. Environmental effects can arise from variations in such factors as local temperature and barometric pressure.
To illustrate how proper clock characterization and ensembling can overcome some of these problems, the authors compared an ensemble of three cheap stop- watches with the UTC time scale generated by NIST for a period of 145 days [see "In praise of cheap stopwatches," opposite]. In this experiment the measurement noise arose principally from uncertainties in the authors' ability to trip a camera shutter at a precise moment. Environmental effects, due to seasonal variations in temperature during the observing period, could be clearly discerned, and temperature variations could be assessed by comparing the residual errors of two of the three clocks. Also measured were frequency inaccuracies and frequency drift, and appropriate corrections were applied to each of the clock readings.
After removal of such systematic and environmental effects, the improvement in predictability was well over a factor of a thousand, giving a time predictability of the ensemble of about I second per year- for just $181


Official time

Because most of the users of UTC want official time to tie to earth time (earth's rotation on its axis gives a time scale called UTI), a dilemma arises. People interested in precision time and frequency want the most uniform and accurate time possible. Yet, because the planet speeds up and slows down milliseconds in a day, earth time is not useful for precision metrology. For that reason, it was decided to make UTC a compromise time scale.
Changes due to instabilities in the earth's spin rate are accommodated by employing leap seconds in UTC, while on the other hand, the length of the UTC second is kept as close as possible to the definition based on the cesium atom. UTC was set to be synchronous with UTI at 0000 hours on 1 January 1958. Until 1972, frequency steps and 0.1-second time steps were used to chase the instabilities in earth time.   From 1972 through 1996, there were 20 leap seconds.
As mentioned earlier, UTC is generated near Paris at the Bureau International des Poids et Mesures. There, data from about 230 clocks from 60 laboratories around the world is collected. Each clock has its rate and stability measured and is tested for abnorrrial behavior. Each clock also receives a weighting factor corresponding to its individual stability. This procedure provides uniformity (stability) in combining data from the entire set of clocks. Reliability is thus achieved by the sheer number of clocks involved. The entire process to generate UTC takes about one month.
Currently, the length of the second is being determined by evaluations from 10 laboratory cesium-beam primary frequency standards. To have the UTC second agree as closely as possible to the definition and to obtain an overall world-best estimate, a weighted combination of these primary frequency standards is taken according to their individual accuracies. The time scale generated by combining this set of primary frequency standards from around the world is called International Atomic Time (TAI, or Temps Atomique International). TAI is Coordinated Universal Time without leap seconds.
While the sotirce of the second for International Atomic Time is derived from the primary frequency standards throughout the world, the flywheel that remembers the calibrations provided by these standards is the ensemble of the 230 or so contributing clocks.
Many laboratories need good estimates of UTC before it is available from the international bureau of weights and measures. There are 47 timing centers around th~ world that generate their own current estimate of UTC, for use in their respectivt countries. These are called UTC(k)s, with the "k" denoting the particular timing center. For example, the UTC estimate generated by the United States Naval Observatory's Master Clock in Washington D.C., is UTC(USNO MC). The Nava Observatory has responsibility for supply mg time and frequency to the Departmen of Defense. Another government group NIST, generates UTC(NIST) and has res ponsihility for determining the length 0 the second for the United States. Both organizations also supply time for other U.S. government entities and their time scales usually agree to within about 20 ns.
By international agreement, all timing centers worldwide strive to keep their UTC(k) within 100 ns of UTC.
Currently, theses predictions are within approximately 10 ns.

Each of the 230 clocks that contribute to Coordinated Universal Time reports its data with respect to the UTC(k) available to that clock. Each UTC(k) including a clock at the Paris Observatory is measured against GPS time. When these time differences are substracted, each UTC(k) is known in comparison to the Paris Observatory clock using GPS time as a transfer standard, which drops out in the substraction.
The French bureau then sends out a monthly bulletin, reporting the time differences of each of the UTC(k)s with respect to UTC for the previous month.
 This bulletin is like a report card, telling each timing center how good its prediction was. When it is received, it tells a center "officially" what time it was!


Advantages with satellites

Measuring and comparing time and frequency of clocks remote from each other can be accomplished in a variety of ways and with a variety of accuracies and stabilities. Usually, the better the accuracy and stability, the more expensive the system. Satellite techniques, in general, have demonstrated many advantages over terrestrial techniques, outdoing them in accuracy, integrity, availability, continuity, coverage, and perhaps, more importantly, cost.
With the advent of the GPS, great progress has been made in transporting time and frequency The atomic clocks on board the satellites are like portable clocks in the sky, continuously available anywhere near earth through inexpensive receivers. Accuracies better than I us with respect to coordinated time are made readily available by the proper use of GPS timing receivers.
The GPS has become the world's principal supplier of accurate time. It is used extensively both as a source of time and as a means of transferring time from one location to another. Three kinds of time are available from the system; GPS time, UTC as estimated and produced by the USNO, and the times from each free-running GPS satellite's atomic clock.
The Master Control Station at Falcon Air Force Base near Colorado Springs, Colo., gathers the GPS satellites' data from five monitor stations located around the globe. A software program estimates the clock time error, frequency error, frequency drift, and orbit for each of the satellites. This information is uploaded to each satellite so that it can broadcast estimated errors as correction terms, in real time, as part of the data message. The process provides GPS with time consistency across the constellation to within a small number of nanoseconds, and accurate position determination of the satellites to within a few meters.
Any discontinuity in GPS time would throw receivers out of lock, so the system cannot tolerate the introduction of leap seconds. GPS time must be kept within 1µs of UTC(USNO MC), ignoring the leap seconds. Typically, the steering performance is within 20 ns. In 1910, when the Department of Defense started keeping time on the satellites, its system time was synchronized with UTC. Since then, GPS minus International Atomic Time has been very close to 19 seconds, while UTC has been delayed many leap seconds.
To provide an estimate of Coordinated Universal Time derivable from a positionmg satellite signal, a set of UTC corrections is also supplied as part of the broadcast signal. These are the time difference estimate between GPS time and the UTC(USNO MC) modulo I second. With these corrections a receiver can, in principle, calculate an estimate of UTC(USNO MC). Except for the purposeful degradation of the GPS signal, called selective availability, by the Department of Defense for security reasons, this calculation may have an accuracy of about 10 ns. Since the USNO predictions have been successful to within about 10 ns, combining these two independent error sources yields a real-time potential uncertainty (ignoring selective availability) level of about 14 ns.


Handling relativity effects

The great accuracies now being attained with clocks require that relativistic effects be included in comparing the times and frequencies of clocks at different primary frequency standards laboratories. Relativistic effects are also important for navigation using the GPS. Basically, relativity effects on clocks are of three types; slowing of the rate of a moving clock, frequency shifts due to differences of gravitational potential, and the inconsistencies of synchronization in a rotating reference frame, such as on the spinning earth.
The synchronization problem can be handled by introducing an inertial reference frame that has its origin at the earth's center, which is not rotating. Since the speed of light is constant in this earth-centered inertial frame, a hypothetical, but self-consistent, set of synchronized clocks could be distributed throughout the frame. Relativistic corrections could then be applied to real clocks so that their readings agree with the hypothetical clocks at each of their locations in the earth-centered inertial frame. The resulting self-consistent time system in the inertial frame can then be used by earth- fixed observers. Called "coordinate time," this approach is used for UTC.
Mean sea level is approximately a surface of constant total gravitational potential, called the geoid, in the earth-fixed rotating frame. Conveniently, several relativistic effects combine so that ideal clocks at rest on the rotating geoid all tick at the same rate. This rate defines the SI second for UTC as currently given by the cesium atom.
Corrections must be applied to primary frequency standards laboratories to compensate for their distances from the geoid. The primary frequency standard for the United States, NIST-7, runs fast with respect to an ideal clock at the geoid by about 15.55 ns per day, which is significant, given its current accuracy of about O.4 ns per day.
As for slowing the rate of a moving clock, the Global Positioning System has at least eight relativistic effects that must be accounted for. The slowing of a moving clock means that GPS satellite clocks lag behind clocks on the geoid by 7.11 us per day. Gravitational frequency shifts cause GPS satellite clocks to run fast relative to clocks at the geoid by 45.7 us per day. The net effect on atomic clocks in a GPS orbit is that they run fast by 38.59 us per day. GPS satellite clock frequencies are corrected downward by this amount before launch, in order for the broadcast signal frequencies to be correct when received.
Also, since GPS orbits are not perfectly circular, additional relativistic effects arise from second-order Doppler shifts, and from gravitational frequency shifts, as the satellite moves toward and away from the earth. The system's receiver software must correct for relativistic effects due to this eccentricity, sometimes by as much as  several dozens of nanoseconds. When the system's transmitter and receiver clocks are corrected in this way, they will be self-consistently synchronized. GPS time is thus another example of coordinate time.


Accuracy and stability of UTC

In recent years, the accuracy and stabi lity of UTC have been been dramatically improved. These advances are primarily due to the 1991 introduction of the Hewlett- Packard model 5071A cesium-beam clock. Besides being about 10 times more accurate than any other commercial frequency standard, this clock also has less sensitivity to the environment-by more than a factur of 10. The result is a clock with an exceptionally long-term stability~usually averaging well below I ns per day. From 1991 to 1996, the stability of UTC improved about tenfold.
With these same clocks, timing centers are nowadays able to do a much better job of predicting coordinated time. Because of the outstanding performance of the HP 5071As, the weight assigned them is nearly three-fourths that for the entire UTC ensemble~ven though their total number is less than half of all the contributors to coordinated time [Fig. 3].
Moreover, in the last few years, emerging technologies such as laser energy-state pumping and detection have brought about additional improvements in cesium primary frequency standard accuracies. Other advances are laser cooling down to near absolute zero using photon pressure, followed by a photon pressure pulse producing a controlled low-velocity atomic fountain. Improvement in accuracy of nearly a factor of 10 has already been achieved.
Though it does not exist in nature, an ideal clock can be conceptualized. If three or more independent clocks are compared, the individual instabilities of each with respect to this ideal clock may be estimated. A clock's instabilities are directly related to its predictability [Fig. 4].
Since UTC is not directly available as a clock, real-time approximations to it are made available from 47 timing centers around the world. Several of these timing centers have means of broadcasting the time from their UTC(k)s. Aside from prediction errors, another problem in accessmg the different UTC(k)s arises from instabilities in the methods employed for transmitting and making available this standard time and frequency information.
Considerable effort has gone into measuring the instabilities of various time and frequency transfer techniques. When the frequency stability of different methods of time and frequency dissemination is measured, it is found that essentially all techniques improve with increased averaging time.
One of the best time transfer techniques is GPS common-view, which provides a fractional frequency stability of about 10.14 with one day of averaging. To explain the technique, suppose the time of a clock in Boulder, Cob., is being communicated to the Paris Observatory. At a predetermined instant, the clock in Boulder measures its time difference with respect to GPS time, B-C. At the same instant, the GPS receiver in Paris, listening to the same satellite, measures its time difference with respect to GPS time, P-G. These data are exchanged and subtracted, yielding the time difference, B-P. With all the delay corrections properly accounted for, the accuracy of this technique has been shown to be a few nanoseconds. The effects of selective availability drop out in the subtraction, as do many other common-mode errors.
Even though the GPS common-view technique presently limits the short-term stability of UTC, its introduction in the early '80s refined the accuracy and stability of time transfer by more than a factor of 20. It has been designed into special GPS timing receivers, for automatic operation, and is currently the principal operational means of communicating the times of the contributing clocks to the Bureau International des Poids et Mesures for generation of UTC.
In recent years, the use of GPS as a source of UTC has increased rapidly. Although 1µs accuracies are achieved with minimal effort and little cost, there are many applications that require better timing, most of which need a real-time output, such as for telecommunications or for synchronizing a power grid. However, the degradation in the GPS signal caused by selective availability can cause peak-to-peak variations of hundreds of nanoseconds. The current GPS common-view technique, which avoids most of the selective availability degradation, only gives time differences between two clocks after the fact.
Anovel approach was developed a few years ago in which the nature of the instabilities of selective availability were studied. Understanding how greatly these instabilities differed in character from the instabilities in precision clocks allowed real-time filters to be designed for timing receivers in fixed locations. Using, for example, the HP 5071A clock with an appropriate filter essentially eliminates the selective availability in real time. lfclocks based on quartz crystal 9scillators or rubidium gas-cell frequency standards are used instead, a good level of select availability filtering is still achievable. Products based on such filtering concepts have become very popular and useful for precision network timing.


Global navigation

Airline safety is a key issue internationally While navigation and timing for all of avionics has taken a giant step forward with the availability of GPS, that system by itself does not provi  deI the accuracy integrity, availability, and a continuity of service that is needed for civil aviation. But some exciting new concepts for improvements with augmentations of GPS are being planned.
One involves the Global Navigation Satellite System (GNSS), a satellite-delivered global navigation and positioning system that meets civilian users' requirements in a cost-effective way. Precise, accurate timing- including Coordinated Universal Time-is built into its designs, and atomic clocks are at the heart of its success.
  In the offing is a set of geostationary satellites that will be used to augrnent the signals already available from GPS and Glonass, Russia's Global Navigation Satellite System. These new satellites are being sponsored by Inmarsat, the London-based international telecommunications cooperative.
Also for civilian use, the U.S. government plans to implement a new GPS signal on the Block hF GPS satellites, scheduled for replenishment launches around the year 2008. This signal would provide civilian users with an accurate measurement of time delays due to the ionosphere, which is now one of the biggest uncertainties in precision airline navigation. But, because of the congestion of the l~2-GHz region of the electromagnetic spectrum, securing an allocation for this new frequency is by no means certain.
Other improvements in satellite position prediction and clock accuracy are under way. For instance, intersatellite links are being implemented with the GPS.

Block IIR program, which should increase the time and position accuracy available from the GPS. The Block IIR satellites are next in the launch schedule.
What's more, the Federal Aviation Administration is in the process of developing and implementing a GPS Wide-Area Augmentation System (WAAS) for all U.S. air traffic, involving GPS-like signals from geostationary satellites. WAAS could bring about large fuel savings, increased in-flight safety, and all-weather precision landings. Ultimately, with future local augmentations near airports, the goal of such programs for position determination for precision landings is 0.8 meters.
Asimilar augmentation is being implemented in Europe, which cannot be covered at an adequate level by transmissions from the WAAS geostationary satellites. Egnos, the European Geostationary Navigation Overlay Service, will augment the existing GPS and Glonass signals with transponders on geostationary satellites. Sometimes also called GNSS- 1, the first-generation Global Navigation Satellite System, Egnos will ultimately support five levels of service over different service areas. It is designed to provide the best performance over Europe, while guaranteeing a minimum level of service with global earth coverage. Initial operation is expected in 1999 for levels 1, 2 and 3, with the launch of two satellites. Full operation is scheduled for 2002 at all five service levels.
The Russian authorities appear to be committed to making Glonass available for civil aviation and acceptable to the international community. A study is under way, under the auspices of the Tacis Telrus project, to identify options for cooperation between Western Europe and Russian industry involving Clonass and other related navigation systems.
Besides being a provider of CNSS services at the international level with WAAS and Egnos, Inmarsat is proposing to add a dual-frequency navigation function to its Intermediate Circular Orbit communications satellites, which are due to be launched around 1999~2000. The proposed system, the International Satellite Navigation Service (ISNS), would offer both an overlay service and an independent dual-frequency navigation signal, which will refine the accuracy in accounting for the timing signal's propagation delay. ISNS is expected to be capable of providing nonprecision approach capability worldwide without differential ground augmentation.
For users in the Asia Pacific region, the Japanese Space Agency plans to launch two geostationary satellites for relaying navigation, integrity, ranging, and differential messages. The service will mainly cover the Pacific Ocean, which is at present poorly served by the lnmarsat-3 overlay system. The satellites, called MTSATs and already in the development stage, will also provide communications and weather capabilities to users.
The initial MTSAT satellite is scheduled for launch in 1999, with a replacement satellite in 2005. This service alone will not meet requirements for primary means of navigation, but will probably provide a component to a future Asian wide-area system. The Indonesian government is planning to launch two additional geostationary satellites with navigation payloads, to complement MTSAT in the South-Eastern Asia region. The first satellite is expected to be launched in 2000.
In Europe, the European Space Agency has made much progress in its development of the wholly civil GNSS-2. Following an initial GNSS-2 Mission Analysis study of system architectures in 1995, the European agency is now concentrating on such key technical issues as satellite orbit configurations. The program is aimed at demonstrating a flight experiment by 1999.
This ambitious project is expected to be funded through public and private initiatives, supported by a Europe that is keen to see implementation of a developmental system as soon as possible. The plan is for GNSS-2 to achieve full operation around 2007-2010 and remain operational until at least the year 2025.
Ultimately, the decision to fabricate the next-generation GNSS will be made at political and institutional levels. Who needs it? Who will pay for it? Ultimately, only through international cooperation and coordination can there be a truly seamless, international, next-generation GNSS. The European Space Agency is currently engaged in discussions with Russia, Japan, and the United States to ensure that the project has a common goal and to minimize the duplication of effort. Without the cooperation of all partners, the cost of providing such a system would be prohibitive.
For example, the GPS and Glonass are not currently synchronized and use different coordinate reference frames. In 1996 the Russian Federation agreed to improve synchronization of its time scales with UTC. This will become an important factor for one-way time-dissemination for GNSS- I where the time scales for the three systems will need to be coordinated. There is also the additional problem that Clonass includes leap seconds and GPS does not.
At present, the GPS coordinate reference frame complies to within a few tens of centimeters with the International Terrestrial Reference Frame. From the last report of the Russian authorities, getting the Glonass reference frame to also comply poses a challenge and may not happen for some time.
Because the next~generation CNSS user market will be automotive as well as avionic, the accuracy requirements will be much stricter-lO cm is one of the long- term goals, for collision avoidance on smart highways. Light travels 30 cm in 1 ns. Already methods that could accomplish this incredible goal have been proposed and are being studied, but propagation delay inaccuracies pose one of the biggest problems.
Finally, the issue of who controls the next-generation GNSS is complex, and beyond the scope of this article. Some of the challenges ahead include encryption of the signal, which would need agreement from all parties owing to the global naturc of the system. Ensuring a seamless system will also be difficult. And for safety-critical applications, responsibility and liability for the operation of the system is currently an active area of research.


UTC prospects and problems

Since UTC is the common reference for all national time scales, and recent advancements have boosted its accuracy and stability, all the timing needs throughout the world are essentially being satisfied through the current methods of generation and distribution. The prospect of making UTC a real-timeservice is a burden that will fall on each nation. But, now that GPS has international coverage, this burden is minimal.
Presently, GPS can be used to obtain an estimate of the UTC(USNO MC) clock. If the time output of a good multi- channel C/A (clear access) code GPS receiver is averaged for one day against a sufficiently stable local clock, such as a cesium standard, the resulting estimate of UTC(USNO MC) will be within 20 ns about 95 percent of the time. With that kind of performance, the GPS broadcast correction should be within 30 ns of UTC about 95 percent of the time.
Typically the frequency excursions of UTC(USNO MC) through GPS average below 10 ns per day. As for the GPS clock ensemble, it has a long-term frequency stability of about 2 ns per day. The frequency stability of UTC (USNO MC) is about 0.2 ns per day, and its rate is a predicted-forward estimate of Coordinated Universal ~me, well within I ns per day.
In the future, there should continue to be much improvement in the accuracy of the coordinated time second as well as that of the SI second generated by the primary standards, which are being constructed now with anticipated accuracies of 0.01 ns per day. These improved accuracies will benefit the precision user community within the limitations of the methods of communicating time and frequency.
But there are two other developments that could greatly improve the usefulness of UTC: making it a real-time service, such as Creenwich Mean Tme used to be, and decreasing the measurement noise in the time and frequency transfer techniques. The measurement noise associated with communicating time from one location to another is of such a nature that the performance of state-of-the-art clocks cannot be utilized at a distance. The noise also degrades the short-term stability of coordinated time.
The fact that UTC is not available in real time means that none of the users having real-time synchronization needs can use official world time. Still the number of such applications is increasing rapidly-for example, telecommunications network nodes need highly accurate synchronization~o alternative solutions are needed for synchronizing timing networks.
The traditional GPS common-view technique is used to transfer the times and frequencies of most of the standards contributing to the generation of International Atomic Time and UTC This procedure removes most of the effects of selective availability and employs one-day averages, but requires a large amount of time for processing data from many laboratories before the comparisons are published.
Improvements are anticipated by enhancing GPS receiver stabilities and accuracies and by bettering the characterization of atmospheric delays. One technique that has the potential to provide short-term performance in real-time is the GPS advanced common-view (ACV) technique, which uses more than one GPS satellite at a time. Another is the two-way satellite time and frequency transfer technique, which exhibits excellent short-term stability-limited by only 200 or 300 picoseconds of white-noise phase modulation taken over 1-second averages.
However, environmental effects on the transmit and receive equipment can cause degradation of the longer-term performance. The enhanced GPS technique is a systems approach in which the attributes of the reference clock are best utilized to filter the effects of the GPS selective availability degradation, to obtain a real-time estimate of UTC.
Both GPS common view and GPS advanced common view use the broadcast time code. An alternative approach using the system's carrier phase for transferring time and frequency information between two remote sites shows promise, but more research is needed.
The long-term accuracy and stability of UTC can take advantage of the best clocks in the world and are currently doing so. This situation is possible because over a long enough time, the measurement noise can be averaged away. The problem is that the more clocks improve, the more time is needed to average away the measurement noise. The result is a need for both real-time access to UTC and better methods of communicating time and frequency between clocks and to the user community.
These difficulties are not insurmountable and reasonable solutions have been proposed. As clocks improve along with methods for communicating time and frequency, available accuracies will also improve, providing additional benefits to an ever-growing community of users of precise time.

To probe further

A comprehensive treatment of the Global Positioning System (GPS), including its background and history can be found in "Global Positioning System: Theory and Applications, Vols. I and II," eds. B. W. Parkinson and J. J. Spilker Jr., in Progress in
Astronautics and Aeronautics (American Institute of Aeronautics and Astronautics, 370 L'Enfant Promenade, Washington, DC
2002~2518 1996; ISBN 1-56347-106-X, 1~56347-107-8).

A more detailed discussiori of the topics addressed in this article is given in "The SPACE AGE Science of Timekeeping," Hewlett-Packard Application Note 1289, by D. W. Allan, N. Ashby, and C. Hodge, available from Jerrie Kerins, Hewlett-Packard Corp., Santa Clara Division, Box 58059, Santa Clara, CA 95052.

Several annual congresses publish proceedings that can provide much additional information on time, frequency, and navigation. Among these are Proceedings of the Institute of Navigation-GPS-93, IONGPS-94, ION-GPS-95, ION-GPS-96, and IONGPS-97 (1800 Diagonal Rd., Suite 4801, Alexandria, VA 22314);Proceedings of the European Frequency and lime Forum (Institution of Electrical Engineers, London); Precise Time and Time Interval Applications and Planning Meetings, U.S. Naval Observatory (NASA Center for Aerospace Information, Springfield, Va.); National Technical Information Service-NTIS, distributor; and Proceedings of the Annual Symposium on Frequency Control (sponsored by the IEEE, published by NTIS).

Articles on the Global Positioning System and its augmentations can be found in the periodical GPS World., Box 6139, Duluth, MN 55806~139; 800-346-0085 or218-723- 9477; Web, wvvw.gpsworld.coml.
Among the Internet sites of relevance to GPS, with links to related topics, are http://physics/nist.gov/Genint/time.html
and http://wvvvvhost.cc.utexas.edu/ftp/pub/grg/gcraft/notes/gps/gps.html.

A monograph from the National Institute of Standards and Technology (N 1ST) that is also very useful is "NIST Technical Note 1337, Characterization of Clocks and Oscillators (1990)," U.S. Government Printing Office, Washington, DC  20402~9325.

About the authors

David W. Allan worked at the U.S. National Institute of Standards and Technology (NIST) from 1960 until 1992, focusing on time and frequency research, development, and dissemination. He is currently in Africa on sabbatical from his position as president of Allan's Time Inc., Fountain Green, Utah, a time and frequency metrology company. His contributions include the development of an internationally adopted method of characterizing the performance of clocks, known as the Allan variance.
Neil Ashby is a professor of physics at the University of Colorado, where he joined the faculty in 1962. He also consults for NIST. His research formed the basis for relativistic corrections being properly included in the Global Positioning System.
Clifford C. Hodge is currently a project leader at the National Physical Laboratory in Teddington, Middlesex, England, where he is involved in developing the international Global Navigation Satellite System (GNSS). He previously was a resident research associate at NASA's Jet Propulsion Laboratory in Pasadena, Calif.

IEEE Spectrum editor: Tekla S. Perry