Judah Levine

Judah Levine wants everyone to have access to the correct time, as determined by atomic clocks at NIST and other national laboratories around the world. Since 1967, the Bureau International des Poids et Mesures (BIPM) has been responsible for determining Coordinated Universal Time, or UTC. The BIPM is located near Paris and has no clocks of its own. It determines UTC by averaging the data from 300 atomic clocks, which are located at NIST and other timing laboratories throughout the world. The BIPM computes this average using measurements made every 5 days at 0 UTC and publishes the difference between the UTC time scale maintained by each laboratory and this overall average. Because data from BIPM arrives at NIST about a month late, Levine is responsible for NIST's predictions of UTC, or UTC (NIST). NIST scientists use UTC (NIST) as a real-time estimate of both UTC and International Atomic Time (TAI).

Levine oversees multiple approaches for communicating UTC (NIST) to anyone who wants to know the exact time or evaluate the frequency of a device. First, he and his colleagues devise methods for distributing frequency information with the highest possible accuracy. These methods are used to compare widely separated primary reference clocks, for example, which have accuracies and fractional frequency stabilities approaching 10^-15. Comparing devices that can realize this level of accuracy or stability requires that the error budget of the measurement process itself be significantly smaller than this value. Two techniques are under investigation: (1) two-way satellite time/frequency transfer that uses commercial broadcast satellites to exchange data between two sites with atomic clocks and (2) GPS carrier phase time or frequency transfer that uses measurements of the phase of radio-frequency carrier waves transmitted by Global Positioning satellites. By comparing GPS measurements at two atomic clock sites, Levine's group can evaluate the clocks' frequencies.

The GPS technique is much more cost effective than the two-way satellite technique because users don't have to transmit signals up to the satellites or pay to receive the signals. However, it has the disadvantage that data can only be processed in batches of 1-5 days. Discontinuities often appear between the end of one data set and the beginning of the next. Levine's group recently developed an improved analytical approach for dealing with these discontinuities.

Because the two methods are affected by very different noise sources, using them simultaneously provides insight into the limitations of both. For example, the Levine group has compared the frequency of the cesium fountain atomic clock at NIST with a similar device operating at the Physikalisch Technische Bundesanstalt (PTB) in Braunschweig, Germany. Using an averaging time of 42 days, the combined uncertainty of the two techniques is about 4.2 x 10^-16, which is about a factor of 5 smaller than the accuracy of the devices themselves. These results show that the techniques are adequate for comparing the frequencies of the current generation of primary frequency standards, but they are unlikely to remain adequate as higher-accuracy clocks are developed in the near future.

Second, the Levine group designs and operates the public Network Time Service, which currently handles about 20,000 requests per second for time in a variety of digital formats. To satisfy both the current number of requests and anticipated service growth, Levine and his colleagues are working on new techniques to improve the cost/benefit ratio of their client systems. That is, they are figuring out ways to improve the accuracy of the time of the client systems, while at the same time requiring fewer requests to the servers. They are also developing software that can support an explicit trade-off between the accuracy of the time of the client system and the number of calibrations that is required to realize this accuracy. For example, users who require only modest time accuracy (e.g., ~0.1 s) can configure their software so that the cost of realizing this accuracy can be almost a thousand times less than the cost of realizing the highest possible accuracy of about 10-30 ms. (In both cases, the cost is a function of the network bandwidth and number of computer cycles used.) This is a significant advance, since many users require only relatively modest time accuracies and cannot effectively make use of the higher accuracy that can be realized with more frequent calibrations. Possibly as a result of new techniques such as this one, the growth of the time service has slowed from about 7% per month to about 1% per month.

Third, Levine and his colleagues are improving the stability of UTC (NIST) with respect to UTC itself. New enhancements include the design and construction of new measurement hardware with a noise floor of about 2 x 10^-13 s and upgraded measurement software to support the improved hardware. As a result of these improvements, the stability of UTC (NIST) will be about 10^-15 (or a bit better) at an averaging time of 30-40 days. This is about a factor of 2 better than the current system. When the upgrade is completed, the stability of UTC (NIST) over the averaging time needed for an evaluation will be comparable to the accuracy of the NIST cesium fountain atomic clock itself.

Once UTC (NIST) is known, it is transmitted to radio station operators in Fort Collins, Colorado, and Hawaii who, in turn, broadcast the information to the public via shortwave radio station WWV. A second station, WWVB, which is also located at Fort Collins, transmits a low frequency signal at 60 kHz. Many wall clocks and other consumer devices can receive these signals. UTC (NIST) is also available via a talking clock at 303-499-7111 and in digital format at 303-494-4774. It is also available over the Internet. At present, 14 Internet servers, all synchronized to the NIST atomic clock ensemble, answer about 1.5 billion requests a day for the exact time. All of the NIST time services are provided at no extra charge beyond the cost of the network connection or telephone call.

Currently, Levine is evaluating a new generation of hardware and software that can work with the ultraprecise optical atomic clocks currently under development. The new techniques use the phase of the carrier transmitted by the satellites of the Global Positioning System. He wants to ensure that next-generation communications systems provide redundancy and fall-back strategies in case of failure.