My research focuses on understanding the statistics of precision clocks and frequency standards and on developing methods for distributing precise time and frequency information. The statistical studies are the basis for the NIST time scale, AT1, which computes the weighted average of the times of an ensemble of cesium standards and hydrogen masers, and uses this weighted average to realize the NIST time scale, UTC(NIST). This time scale is the basis for all of the NIST time and frequency services and is also used to evaluate precision oscillators that are being developed by other groups at NIST and at JILA. For example, the time scale output can be used to characterize an optical frequency using a frequency comb, which links the microwave frequency output of the time scale to the optical frequency. It is also used to characterize oscillators based on cooled, trapped ions. The output of the time scale is typically within a few nanoseconds in time and less than 2 X 10-15 (for up to 30 days of averaging) in frequency relative to UTC, the international time scale computed by the International Bureau of Weights and Measures (BIPM).
A second important application of the time scale is to provide a reference time for the NIST time services, including the radio stations WWV, WWVH and WWVB, which transmit time and frequency information from transmitters in Fort Collins, Colorado and Hawaii. The transmissions from station WWVB in Colorado are very widely used to set consumer devices such as wall clocks, microwave ovens and similar systems. In this application, the advance notice of the transitions to and from daylight saving time and the traceability of the time signals to a national standard are more important than the accuracy of the time signals as received by the users, which is typically on the order of 0.05 s. A related application uses the time signals from the NIST time scale to distribute time and frequency information in digital formats using dial-up telephone lines and the Internet. These services are widely used to synchronize e-mail networks and network routers and gateways. The signals are also used by commercial and financial institutions, and are particularly relevant to providing time stamps for rapid, automated trading in stocks, bonds, and commodities. The Internet time service also supports authentication based on one-way hash functions. This authentication guarantees that the time signal originated from a NIST server and was not modified (either accidentally or maliciously) in transit. The service uses an geographically diverse ensemble of systems that are synchronized to UTC(NIST) in Boulder, and the servers currently receive about 8 X 10http://www.nist.gov/pml/div688.9 requests per day. These services are more fully described on the time and frequency web page,
A third research effort is a study of methods for improving the accuracy of transferring time and frequency information so as the enable the comparison of current primary frequency standards, which can realize the SI second with an uncertainty of less than 10-15, and the next generation of these devices, which may have uncertainties a factor of 10 or more smaller than this value. One promising approach is to use the phase of the carrier of the signals from GPS satellites to perform these comparisons, but there are a number of challenges that must be overcome before this technique can realize the required accuracy level. Although the short-term (second to second) fluctuations in the carrier-phase data can satisfy the transfer requirements, the noise spectrum is not white phase noise, so that the variance does not improve with averaging. The flicker and random-walk nature of the noise actually causes the accuracy of the frequency transfer to degrade as the averaging time increases.
Many of these issues are more fully discussed in my review article, which appeared in the February 2012 issue of the Review of Scientific Instruments and in the references listed at the end of that article.