In technologies such as SONET/SDH, proper network synchronization is a critical factor in maintaining quality of service.
By LEE COSART, Symmetricom--Proper synchronization in a telecommunications network is critical to its operation. Inadequate synchronization compromises quality of service leading to such impairments as data retransmission, digital video freeze and distortion, and severe degradation of encrypted services.
This is especially true in SONET/SDH networks. Frequent pointer movements--the SONET/SDH mechanisms for maintaining data integrity when subjected to clocking imperfections--can lead to high levels of jitter, and wander, and compromised service quality.
Characterizing network operation and equipment by measuring synchronization in a live network and in the laboratory is an important means of ensuring networks and equipment meet national and international synchronization requirements. Standards bodies and organizations such as the American National Standards Institute (ANSI) and Telcordia in North American, the European Telecommunications Standards Institute (ETSI) in Europe, and the International Telecommunication Union-Telecommunication (ITU-T) internationally, publish documents outlining synchronization limits both for equipment and for networks. Thus, for example, separate jitter and wander limits are specified for primary reference sources such as cesium clocks and global positioning system (GPS) timing receivers, building integrated timing supply (BITS) clocks (equipment distributing synchronization), and SONET/SDH network elements operating in a network.
Measurement theory
Standards addressing synchronization refer to phase, frequency, maximum time interval error (MTIE), and time deviation (TDEV) in discussions of jitter and wander. The understanding of phase, in particular, is critical because the other measurements--frequency, MTIE and TDEV--are all derived from phase.
A number of terms equivalent to phase are used by the standards bodies. Time interval error (TIE) is the most common term used for phase in the standards. Another term is "phase deviation" which acknowledges that phase is always calculated with the measured signal compared to some other actual or ideal reference clock.
To measure phase, some kind of phase detector is required. A phase detector senses when a signal passes through a set voltage threshold, timing the threshold crossing relative to a reference. This procedure is illustrated in Figure 1, where positive edges of the signal are timed with respect to a reference with these differences plotted as phase. An ideal signal, one without jitter or wander, is represented as a series of zeroes and hence lies on the x-axis. The dividing line in the standards between noise classified as jitter, or wander, is 10 Hz--noise components above 10 Hz are jitter and noise components below 10 Hz are wander.
With slower moving wander components removed with a high-pass filter, jitter is then generally characterized by subtracting the minimum value from the maximum value resulting in a measure of peak-to-peak jitter. The phase unit used in this case is "unit interval," (UI) where one UI is equal to a 360° phase movement. In the case of wander characterization, additional steps are required. First, jitter components are removed by a low-pass filter, and then two separate calculations are performed for two alternative views of wander.
MTIE is computed by repeated searches through the filtered phase data for maximum phase excursions over a series of time windows. The larger the time window or "tau" is, the larger the MTIE. MTIE is a monotonically increasing function of tau; it can only increase or stay the same as tau increases. MTIE is thus a kind of peak detector. TDEV or "time deviation" by contrast is a root-mean-square measure of wander over various integration times or "taus." TDEV is a root variance based on Allan Variance, a measure commonly used to characterize atomic clock stability. TDEV was specifically designed for the characterization of wander noise processes operating within the network. Both MTIE and TDEV characterize wander over a range of values ranging from short-term wander to long-term wander.
In Figure 2, a 24-hour phase measurement is separated with filters into jitter and wander, with wander forming the basis for MTIE and TDEV calculations. In the MTIE and TDEV graphs the measurements (the lower curves) are compared to standards (the upper curves with markers). The MTIE and TDEV measurements are below the limits specified by these particular standards--in this case ANSI--so this measurement meets the wander requirements.
Also shown in Figure 2 is a second phase measurement showing a quadratic shape characteristic of a drifting oscillator. Frequency is the rate of change of phase over time. In mathematical terms, frequency is derived from phase by differentiating phase. In graphical terms, frequency is the slope of phase. Thus as the slope of the phase plot on the left increases, so does the frequency in the plot on the right. In this case frequency increases linearly, a common occurrence in drifting oscillators. Standards organizations have universally set a very strict frequency accuracy limit for telecommunications networks, a part in 1011, which accounts for the necessity of primary reference sources such as cesium clocks or GPS timing receivers.
Synchronization measurement and monitoring equipment
A variety of equipment is used to measure and monitor network synchronization. In addition to specialized jitter and wander test sets, general purpose measurement instrumentation such as counters and time interval analyzers is used with commercially available special-purpose software. All this test equipment requires a reference clock, either a primary reference source such as a GPS timing receiver, a suitable atomic oscillator, or a signal traceable to a primary reference source.
Maintenance of quality synchronization is of such importance that equipment designed to provide network synchronization has, in modern designs, included the built-in capability of measuring phase, frequency, MTIE, and TDEV. This includes both BITS synchronization distribution equipment and GPS primary reference sources. Miniaturization of the circuitry used for synchronization measurements is leading to the emergence of compact synchronization modules for multiple channel synchronization measurements.
Lee Cosart is a research and development engineer at Symmetricom (San Jose, CA).