Within decades of the invention of the first atomic clock and first computer in the mid-20th century, global time keeping would be forever transformed. In 1967, time keeping went from relying on the fluctuating motions of the Earth to being based on precise transitions within a cesium atom. The communication of time soon left the telegraph behind, and radio time transmissions were improved. During the 1980s and 1990s, time communication also began to rely on telephone networks, the Internet, and satellite networks, which together now respond to billions of requests per day for the exact time. Since 1990, there has been a veritable revolution both in determining the exact time and in communicating it to increasingly sophisticated personal computers, telecommunications networks, electric power grids, financial markets, scientific institutions, and navigation systems on Earth and in space. JILA Fellow Judah Levine, a member of the National Institute of Standards and Technology’s (NIST’s) Time and Frequency Division, has been a key participant in this revolution.
The determination of the exact time requires more than just one clock. In the United States, it calls for computing the average time of an ensemble of clocks, known as a Time Scale, at NIST in Boulder, Colorado. These clocks are calibrated to the nation’s primary standard (the F-1 cesium fountain atomic clock). The NIST time scale delivers Coordinated Universal Time (UTC) via NIST’s time services. Levine has not only helped improve the time scale, but also spearheaded the creation of both an automated computer time service that relies on telephone networks and an internet-based time service. He has also developed advanced methodologies for comparing the accuracy and stability of atomic clocks in different locations using the phase of global positioning satellite (gps) carrier waves. This work has been performed in anticipation of a new generation of optical atomic clocks that will be the most stable and precise in history. Comparisons of these next-generation clocks will require time-transfer methods that are even more stable and precise than the clocks themselves.
The NIST Time Scale
When Judah Levine came to the National Bureau of Standards (NBS), now NIST, the Time Scale as it exists today had not yet been created. NBS had a primary frequency standard that ran infrequently — about once a year — because calibrating it was a monstrous job that took about two months. NBS also had a primitive time scale consisting of four or five early cesium atomic clocks. Technicians measured the time on these clocks by hand and calculated their average time.
In the early 1970s, NIST took the first step in automating the measurement of the differences in time among the clocks by having each clock produce a paper tape of its time readings over the past month. These tapes were then carried upstairs to the NBS computer system, which at that time occupied an entire room. The computers used the data on the tapes to calculate the average time of the clock ensemble for the month. These monthly time averages were sent to the Bureau International de l’Heure (BIH) at the Paris observatory via telex (a precursor to the teletype).
In 1978 Levine and NBS colleague Dave Allan began the job of converting this rudimentary system into a modern Time Scale featuring a computer-controlled time measurement system. After purchasing two new computers, Levine focused on writing the first of what would become 200 analysis programs (software) for the measurement system. Allan developed the system’s statistical underpinnings. With the help of colleagues at NIST, Levine and Allan had the nation’s first automated time scale up and running by 1980. The paper tape era was over.
In 1983, Levine received a U. S. Department of Commerce Gold Medal for his role in developing the automated Time Scale. The now 28-year-old measurement system remains the official measurement system today. However, because the Time Scale now depends on old (and unsupported) computers, NIST will have to upgrade both the computers and the measurement system in the near future. In the meantime, though, monthly time reports are sent to the BIH’s time-keeping successor, le Bureau International des Poids et Mesures (BIPM). The reports are now sent via email and using file transfer protocol (FTP).
Although the measurement system has stayed the same for nearly three decades, the clocks in the Time Scale have undergone continual improvement. The time scale currently has four improved cesium atomic clocks, which provide long-term stability, and six hydrogen masers, which are extremely stable in the short term. One of these is the original maser purchased for the Time Scale and the rest were subsequently added over 3–4 year intervals.
In 2007, Levine installed a back-up Time Scale at a remote site near Fort Collins, Colorado, (which also houses the transmission towers for NIST’s radio stations WWV and WWVB). The back-up Time Scale consists of four atomic clocks and a measurement system. It will become fully operational once enhancements to the remote site, including telephone lines and high-speed internet connections, are installed. In the meantime, Levine carries data from the back-up Time Scale on his laptop computer for comparison with the primary NIST Time Scale in Boulder.
Automated Computer Time Service (ACTS)
With the new automated Time Scale up and running in 1983, Levine and NIST colleagues Mark Weiss, Dickie Davis, Dave Allan, and Don Sullivan turned their attention to modernizing time transfer. Until then, the only time-transfer services were the short-wave radio stations WWV and WWVB, which transmitted time signals to wall clocks and other devices. At that time there were no a “low-level” time services for ordinary people who needed access to the precision of atomic-clock-based time.
Levine and his colleagues decided to build a telephone-based time-transfer system. Davis took on the control of the phone lines (firmware), Weiss wrote the software that ran the firmware, and Levine took charge of the client programs, or user interface. Levine’s first user interface ran on early MS DOS systems, but it evolved to work with personal computer (PC) operating systems, including XT and AT, as these systems progressed over time.
ACTS came online in the late 1980s running on two new MS DOS servers. The service grew rapidly at first as personal computer users took advantage of their dial-up modems to access atomic clock time. Today, ACTS runs on 6–8 PC servers at NIST. The system has 30 phone lines available serving the continental United States and another four in Hawaii.
ACTS is currently undergoing an upgrade. During 2006–2007, Levine undertook the design of a modernized ACTS to be located in a small control building at a remote site near Fort Collins, Colorado. This ACTS will be maintained by technicians at the site, but can be controlled and run remotely. It will come online once telephone lines are installed at the site.
Internet Time Service
As early as 1990, it was clear that ACTS alone was not going to be able to handle the burgeoning number of time requests. Levine soon came up with a novel idea for solving the problem: He proposed synching a computer with ACTS so the computer could disseminate time information over the Internet. Initially considered a minor addition to the existing time service, the idea soon became the foundation for NIST’s new Internet Time Service (ITS), which was developed by Levine.
In its first year of operation, ITS relied on servers in the NIST-Boulder facility; soon servers were added at NIST-Gaithersburg (MD), the National Center for Atmospheric Research (NCAR) in Boulder, and Redmond, Washington, where Microsoft donated space to the new time service. Today, ITS uses 24 servers located across the country. Server hosts provide power, cooling, and bandwidth to the tune of several thousand dollars a month.
With the exception of a single network standard program for clients, Levine wrote all the software for the new time service, whose popularity grew exponentially after it was introduced in 1993. Since then, Levine has not only overseen software improvements, but also several hardware upgrades, including one in 2008 featuring the installation of high-capacity rack-mounted commercial servers. In 2007, Levine was awarded a second Gold Medal by the U. S. Department of Commerce for his many accomplishments related to the NIST time scale and time services.
During the past decade, Levine has also worked to improve a technique for comparing the frequencies of different atomic clocks based on measurements of the phase of the carrier waves transmitted by global positioning satellites at sites where the clocks are operating. To make these comparisons, the methods must have a lower uncertainty than the clocks themselves. Currently, the techniques for comparing the accuracy and stability of atomic clocks are just barely ahead of those of the clocks. However, improvements in atomic-clock technology are now outpacing the ability of the clock-comparison community to devise better comparison methods.