Appendix B
Issues in Digital Network Time and Frequency Synchronization

1. INTRODUCTION

This appendix explores issues in the synchronization of digital communications networks used for telephony and data transmission. It is based on briefings presented to the National Research Council’s Committee on Review of Switching, Synchronization and Network Control by several organizations, including the American Telephone and Telegraph Company (AT&T), Bell Communications Research, MCI Communications Corporation, US Sprint, CONTEL/ASC, and the U.S. Coast Guard,1 together with material provided by the committee members themselves. In addition to this information, the appendix includes background material gathered from many sources, as documented in the notes at the end of this appendix.

The most important area of network synchronization for the committee’s purposes has to do with how various digital telephone networks interoperate using synchronous data streams at the T-1 rate (1.544 Mbits/s) with DS-1 frames (193 bits). The electrical interfaces between such networks use a double-buffered technique to compensate for the different framing relationship ordinarily encountered between them. If the networks use timing sources not exactly synchronized in frequency (phase locked), the frames sent by one network will precess slowly with respect to another, and frames must



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness Appendix B Issues in Digital Network Time and Frequency Synchronization 1. INTRODUCTION This appendix explores issues in the synchronization of digital communications networks used for telephony and data transmission. It is based on briefings presented to the National Research Council’s Committee on Review of Switching, Synchronization and Network Control by several organizations, including the American Telephone and Telegraph Company (AT&T), Bell Communications Research, MCI Communications Corporation, US Sprint, CONTEL/ASC, and the U.S. Coast Guard,1 together with material provided by the committee members themselves. In addition to this information, the appendix includes background material gathered from many sources, as documented in the notes at the end of this appendix. The most important area of network synchronization for the committee’s purposes has to do with how various digital telephone networks interoperate using synchronous data streams at the T-1 rate (1.544 Mbits/s) with DS-1 frames (193 bits). The electrical interfaces between such networks use a double-buffered technique to compensate for the different framing relationship ordinarily encountered between them. If the networks use timing sources not exactly synchronized in frequency (phase locked), the frames sent by one network will precess slowly with respect to another, and frames must

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness occasionally either be discarded or replicated in order to maintain overall synchronization. The appendix consists of nine sections, including this one. The following two sections describe the means for generating and distributing the national standards of time and frequency in the United States. In particular, Section 2 discusses general aspects of standard time and frequency scales used for navigation and space science, while Section 3 describes the primary services operated by the National Institute of Standards and Technology (NIST), formerly the National Bureau of Standards (NBS), for the dissemination of standard time and frequency. The next three sections describe how synchronizing information is distributed throughout the United States and utilized as timing sources by various switches and transmission systems in the U.S. telephone networks. In particular, in Section 4 several time and frequency distribution systems are presented, with special emphasis on LORAN-C, which is fast becoming the system of choice used by U.S. exchange and interexchange carriers. Section 5 discusses issues important for the understanding of synchronization errors and how they may affect the operation of the various switches and other components of the telephone network. Section 6 describes the synchronization networks operated by the various exchange and interexchange carriers in the United States, including AT&T, the Bell Operating Companies (BOCs), and selected independents. The final three sections contain the committee assessment of the impact of synchronization impairments on the National-Level Program/National Security Emergency Preparedness (NLP/NSEP) programs. In particular, Section 7 analyzes the effects of these impairments on transmission, switching, and user applications, while Section 8 discusses the impact on the NLP/NSEP programs. Section 9 states the committee’s conclusion and recommendation. 2. DETERMINING STANDARD TIME AND FREQUENCY2 For many years the most important use of time information was for worldwide navigation and space science, which depend on astronomical observations of the moon and stars. Ephemeris time is based on the revolution of the earth about the sun with respect to the vernal equinox on the celestial sphere. In 1956 the tropical year, or one complete revolution, was standardized at its value at the beginning of this century, when it had a period of 31,556,925.9747 seconds, or

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness 365 days, 5 hours, 48 minutes, 46 seconds; however, the actual year has been increasing by about 5.3 milliseconds (ms) per year since that time. Sidereal time is based on the rotation of the earth about its axis with respect to the vernal equinox point. The mean sidereal day is about 23 hours, 56 minutes and 4.09 seconds of the tropical year, but is not uniform due to variations in earth rotation. In 1967 the Thirteenth General Conference of Weights and Measures decided that the unit of time of the International System of Units is the second defined as follows: The second is 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.”* The transition referred to had been declared in 1964 by the International Committee of Weights and Measures to be that between the hyperfine levels F = 4, M = 0 and F = 3, M = 0 of the ground state 2S1/2 of the cesium-133 atom, unperturbed by external fields…. Here, F is the total angular momentum quantum number and M is the magnetic quantum number associated with F. The International Atomic Time (TAI) scale, used for astronomy and physics, is based on the standard second. On the other hand, the Coordinated Universal Time (UTC) scale, used for other purposes, is based on the rotation of the earth about its axis with respect to the sun, indexed to the prime meridian, which passes through Greenwich, England. In recent times UTC has been slow relative to TAI by a fraction of a second per year. UTC is coordinated throughout the world by the Bureau International de l’Heure (BIH) at Paris, which issues various corrections to TAI on a regular basis. On 1 January 1972 the TAI and UTC time scales were made coincident and have been diverging slowly ever since. The UT-0 day of 24 hours is defined as the mean sidereal day converted to mean solar day by ephemeris tables. The UT-1 day is determined from the UT-0 day by including regular corrections on the order of 30 ms due to seasonal changes in winds and tides. The UT-2 day is determined from the UT-1 day by including irregular *   Source: National Bureau of Standards. 1977. The International System of Units (SI). NBS Special Publication 330. Washington, D.C.: U.S. Government Printing Office.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-1 Characteristics of Primary Standards Type Stability Drift Hydrogen maser < 2 × 10−14/day < 5 × 10−12/day Cesium beam < 3 × 10−13/day < 1 × 10−12/yr Rubidium gas cell < 3 × 10−12/yr < 3 × 10−11/mo corrections as reported by various observatories to the BIH. UTC is derived from UT-1 as described later. Primary Frequency Standards In order that both atomic and civil time can be coordinated throughout the world, it is expected that national administrations will operate publicly available primary time and frequency standards and maintain UTC cooperatively by observing various radio transmissions and through occasional use of portable atomic clocks. A primary frequency standard is an oscillator that can maintain extremely precise frequency relative to a physical phenomenon, such as a transition in the states of an orbital electron. Presently available standards are based on the transitions of the hydrogen, cesium, and rubidium atoms. Table B-1 shows performance data for typical units. For reasons of cost and robustness, frequency standards based on cesium are used worldwide for national standards. In principle then, the frequency standards of the world should not drift apart by more than 43 nanoseconds (ns) per day or 95 microseconds (µs) per year. For instance, The NIST Primary Time and Frequency Standard, which consists of multiple cesium beam clocks and computer-controlled measurement and computation methods, is held to within 10−12 with daily variations even less. Primary Time Standards Since 1972 the various national time scales have been based on UTC, as determined by the BIH using astronomical observations provided by the U.S. Naval Observatory and other observatories. However, it is desirable that the UTC oscillator run in synchronism with the TAI oscillator. Thus, when the magnitude of correction approaches 0.7 s, a leap second is inserted or deleted in the UTC time scale on the last day of June or December.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-2 Months of Leap-Second Insertion Occasion Month of Insertion 1 June 1972 2 December 1972 3 December 1973 4 December 1974 5 December 1975 6 December 1976 7 December 1977 8 December 1978 9 December 1979 10 June 1981 11 June 1982 12 June 1983 13 June 1985 14 December 1987 For the most precise coordination and time stamping of events since 1972 it is necessary to know when leap seconds were inserted or deleted in UTC and how the seconds are numbered. A leap second is inserted following second 23:59:59 on the last day of June or December and becomes second 23:59:60 of that day. A leap second would be deleted by omitting second 23:59:59 on one of these days, although this has never happened. Leap seconds were inserted on the following 14 occasions prior to January 1988, as shown in Table B-2 .11 BIH corrections consist not only of leap seconds, which result in step discontinuities in UTC, but 100-ms adjustments, which provide increased accuracy for navigation and space science. The current time-scale formats used by NIST radio broadcast services do not include provisions for advance notice of leap seconds, so this information must be determined from other sources. Various specification and standards documents stipulate that the primary timing sources used by digital networks must be verifiable with respect to UTC; however, for digital network synchronization, only the frequency information is used—the time information is not used. This distinction is a minor one, since the U.S. standard frequencies distributed by NIST are based on atomic time, while the standard times distributed are based on UTC.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness 3. PRIMARY TIME AND FREQUENCY DISTRIBUTION2 Most seafaring nations of the world operate some sort of broadcast time service for the purpose of calibrating chronographs, which are used in conjunction with ephemeris data to determine navigational position. In many countries the service is primitive and limited to seconds-pips broadcast by marine communication stations at certain hours. For instance, a chronograph error of 1 s represents a longitudinal position error of about 0.23 nautical mile at the equator. The National Institute of Standards and Technology operates three radio services for the distribution of primary time and frequency standard information. One of these uses high-frequency (decametric) transmissions on various frequencies from Fort Collins, Colorado (WWV) and Kauai, Hawaii (WWVH). Propagation of these signals is usually by reflection from the ionosphere F layer, which varies in height and composition throughout the day and season and results in large phase fluctuations at the receiver. The time code is transmitted over a 60-s interval at a data rate of 1 bit/s using a 100-Hz subcarrier on the broadcast signal. While these transmissions and those of Canada (CHU) and other countries can be received over large areas in the Western Hemisphere, the accuracies attainable are considered insufficient for telephone network synchronization. A second service operated by NIST uses low-frequency (kilometric) transmissions on 60 kHz from Boulder, Colorado (WWVB), which can be received over the continental United States and adjacent coastal areas. Propagation of these signals is between the earth and the ionosphere D layer, which is relatively stable over time. The time code is transmitted over a 60-s interval at a rate of 1 pulse per second using periodic reductions in carrier power. With appropriate receiving and averaging techniques and corrections for diurnal and seasonal propagation effects, frequency comparisons to within 10−11 are possible. However, there is only one station and it operates at modest power levels. The third service operated by NIST uses ultra-high-frequency (decimetric) transmissions on 468 MHz from the Geosynchronous Orbiting Environmental Satellite (GOES). The time code is interleaved with messages used to interrogate remote sensors and consists of 60 4-bit binary coded decimal (BCD) words transmitted over an interval of 30 s. The time code information includes the UTC time of year, satellite position, and UT-1 correction. There is some speculation on the continued operation of GOES (one of the two satellites has

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness failed), especially if the Global Positioning System (GPS) continues to evolve as expected. 4. SECONDARY TIME AND FREQUENCY DISTRIBUTION Present time and frequency dissemination services operated by NIST are not sufficiently ubiquitous and reliable as the basis of digital network synchronization in the United States. Accordingly, a high-power, widely distributed and replicated secondary means of distribution would be highly desirable. Among the various means possible to do this are the Long-Range Navigation System-C, or LORAN-C, operated by the U.S. Coast Guard (USCG), the OMEGA system operated by the U.S. Navy, and various satellite services now in operation or planned for the future. These services are described in following sections, with specific attention to LORAN-C, which is particularly suitable for use by U.S. digital networks. LORAN-C1,6 LORAN-C is a wide-area radionavigation system intended for maritime, aeronautical, and land navigation and positioning. It was first used about 1962 and has been operated since then by the USCG in North America and several overseas areas. Coverage is determined by geometry, range, time of day, and receiver characteristics, and presently includes U.S. coastal areas and large portions of the continent, with the exception of a midcontinent gap that is to be plugged by two new chains with four new stations and linked to existing chains. While originally intended for ships and aircraft on intercontinental routes, LORAN-C has domestic applications in aviation for nonprecision approaches, area navigation, and direct instrument flight rules (IFR) routing, as well as automatic vehicle monitoring, electronic maps, and resource management. The USCG estimates there are 40,000 users in the aviation services alone. However, LORAN-C can also be used for the distribution by radio of precise time and frequency, which is the topic of this section. The LORAN-C system operates in the low-frequency (kilometric) band of 90 to 110 kHz using pulse-coded modulation. For navigation purposes a LORAN-C chain consists of a master and three or more slave stations, all operating at a designated repetition rate in the 100-ms range. A chain provides differential time-of-arrival measurements that establish position in a hyperbolic coordinate system

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness to within 500 m under most conditions and to within 30 m (100 ns) under the best conditions. During its designated repetition interval, each LORAN-C master and slave station emits a group of eight phase-coded 100-kHz pulses transmitted at 1-ms intervals, with different phase codes used for the master and slaves and for even and odd repetition intervals. Accurate time measurement to the order of 100 ns requires precise envelope and zero-crossing determination to provide adequate signal-to-noise ratio and to discriminate against multipath due to sky waves (ionospheric reflections), especially at night. A precisely controlled pulse shape is used to maximize accuracy with achievable transmitter power and bandwidth constraints. To retain precise timing, each LORAN-C master station is equipped with three cesium clocks and two sets of timing equipment, which are continuously displayed and compared with each other. Slave stations synchronize to the master transmissions. The signals transmitted by the master and slave stations of a chain are monitored by antenna sensors and by remote receivers at various locations in the service area and along the base lines. Monitor updates including time differentials and received power and noise levels are sent via landline to the stations, which compute phase adjustments in 20-ns increments. Station timekeeping within a chain is usually better than 50 ns relative to the master cesium clock; however, monitored deviations of 100 ns or more are indicated in the transmitted signals by “blinking” certain pulses. The master cesium clocks may drift 60 ns on a day-to-day basis, but are maintained within 2.5 µs of NIST standard time using corrections determined manually and published weekly. With automatic means it is estimated that this accuracy can be improved to 500 ns. The design of the present generation of LORAN-C stations uses solid-state devices extensively, but includes no specific protection against the electromagnetic pulse (EMP) phenomenon from high-altitude nuclear explosions. However, these stations are usually located in remote areas and if necessary can operate from independent power sources for weeks without onsite operators or coordination. Since for timekeeping purposes only one station of a chain is necessary and most areas of the country are within the service area of multiple stations, a considerable degree of redundancy is available.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness According to U.S. Department of Defense (DoD) and U.S. Department of Transportation (DoT) policy and plans for future radionavigation systems, use of LORAN-C by the government may be phased out over the next 10 years in favor of GPS, which is satellite based. The following quotations13 are relevant: LORAN-C provides navigation, location and timing services for both civil and military air and surface users. It is the Federally provided navigation system for the United States Coastal Confluence Zone (CCZ). LORAN-C is approved as a supplemental air navigation system. Signal monitors necessary for LORAN-C guided nonprecision approaches will be installed and become operational in 1989. By 1990, additional transmitting stations will be installed to complete signal coverage over the 48 conterminous states. The LORAN-C system serving the continental United States (including Alaska) and the coastal areas will remain a part of the navigation mix into the next century. DoD will phase out military use of overseas LORAN-C transmitting stations established for military use that do not serve the North American continent. GPS is a DoD developed, worldwide, satellite-based radionavigation system that is scheduled to provide three-dimensional coverage by 1991. The GPS Precise Positioning Service (PPS) will be restricted, due to national security considerations, primarily to the military. However, under certain circumstances, PPS will be available to qualified civil users. OMEGA6 OMEGA is a worldwide very-low-frequency (myriametric) radionavigation system for maritime and aeronautical enroute navigation. The system comprises eight high-power transmitting stations operating on frequencies in the range 10.2 to 13.6 kHz. Navigational position is determined by comparing the relative phase differences of received signals; however, this results in lane ambiguities that must be resolved by other means. The accuracy of these comparisons is limited by propagation corrections, which depend on location and time, and result in a navigational accuracy of 2 to 4 nautical miles. In principle, the worldwide coverage and relatively stable propagation conditions possible at OMEGA frequencies would make this system highly useful for worldwide dissemination of time and frequency. Unfortunately, as mentioned in the section on LORAN-C, future operation of the OMEGA system is in doubt and may be discontinued if GPS proves reliable and economically viable.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness Global Positioning System6 The Global Positioning System is a worldwide satellite-based radionavigation system developed by the DoD and operating on two L-band (decimetric) microwave frequencies at 1227.6 MHz and 1575.42 MHz. When completely deployed as planned over the next few years, the system will consist of a constellation of 18 satellites in near-earth orbit, with at least four satellites necessary to provide accurate horizontal and vertical position information. The key to accurate position determination with GPS lies in accurate determination of satellite position, which is aided by an on-board ephemeris table in each satellite. The tables are continuously updated by information transmitted to the satellite by the system control station and relayed to the users via the L-band transmissions. These transmissions are also modulated in quadrature by two pseu-dorandom sequences for range determination from the satellite to the user. One of these sequences provides accuracy to within 500 m and is intended for civil use. The other provides accuracy to within 20 m horizontally and 30 m vertically, but is currently classified and available only for U.S. military use. There remain considerable uncertainties about the accuracy, reliability, and availability of satellite-based secondary time-distribution systems such as GPS, especially in areas where LORAN-C is available. Satellite-based systems such as GPS can provide differential time measurements to an extraordinary precision; however, with current DoD policy the accuracy achievable with GPS for civil users is in the same range as LORAN-C. Portable Clocks and Transfer Standards6 Portable cesium clocks have been constructed for the purpose of calibrating local time and frequency standards when other means are not available and as a backup for these means when available. These clocks are intended for equipment calibration only and not as a substitute for the regular, in-service methods based on LORAN-C and other systems discussed in previous sections. At one time NIST advocated calibrating local time and frequency standards using the 3.579545-MHz color-burst signal transmitted by the television networks and, indeed, the New York studios of all three networks were equipped with precision oscillators for this purpose. Assuming the offset of these oscillators was known (published peri-odically, for example), then it would be a simple matter to calibrate a

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness local standard, as long as the signal available locally was synchronous relative to the oscillator. Unfortunately, the advent of frame buffers, in which the television frame is buffered locally and made synchronous with the localstation timing generator, destroyed the accuracy of this method and it is no longer viable in most areas. Disciplined Frequency Standards1 Quartz crystal oscillators have been used as frequency references for many years, since they are compact, relatively inexpensive, and stable. A suitably designed and temperature-stabilized crystal oscillator should be stable within a few parts in 1010 per day and be adjustable to a precise reference, such as a cesium clock. However, typical crystal oscillators will show a gradual departure from nominal frequency with time, known as the aging rate. Thus, an uncorrected crystal oscillator may not satisfy the requirements for telephone network synchronization. A disciplined frequency standard (DFS) incorporates a precision quartz crystal oscillator together with a mechanism to measure its departure from a primary reference source and generate corrections accordingly. In the form used by several digital networks the corrections are generated by a LORAN-C receiver and implemented in the form of a digital phase-locked loop. The loop can include provisions to estimate the particular crystal aging rate, as well as ensure stable operation during intervals when the primary reference signal is not available (holdover). The design of typical stratum-2 and stratum-3 clocks (see definitions below) is based on the same principles of DFS, except that in these cases the primary reference signal is not a LORAN-C receiver, but the chosen timing reference signal at either the same or lower stratum. A typical clock design for the DMS-100 family of telephone switches has been described.8 5. GENERAL SYNCHRONIZATION ISSUES1,3,5 The primary reason for worrying about synchronization is to avoid frame slips due to mismatched clocks at the ends of a digital transmission link. General issues on the design and stabilization of clock-distribution networks are discussed in publications cited in notes 3, 4, and 5. In the case of a 1.544-Mbits/s DS-1 link and mismatched

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness daily wander of 18 µs, which has been adopted as an American National Standard. Equipment normally operating at stratum 2, including the 4ESS switch, is engineered to this specification. Equipment normally operating at stratum 3, including the 5ESS switch and digital automatic cross-connect systems (DACS), is engineered to a looser standard of 90 µs. The long-term equipment accuracy is expected to be in the range of 10−11 with 4 to 10 transmission links. AT&T does not intend to operate this network indefinitely. The BOCs (see below) are planning their own synchronization networks and to discontinue use of the BSRF when these become operational. Over the next 10 years, AT&T intends to replace the existing analog network with 12 timing islands, each with a PRS consisting of two rubidium-controlled timing generators, a GPS receiver, and a monitor and control computer to provide performance verification. The stratum-1 accuracy is in concurrence with the International Consultative Committee on Telegraphy and Telephony (CCITT) standards and will be better than 10−11 over the long term (that is, 20 years). The new AT&T sources should not introduce any impairments into the local exchange networks, since the stratum-1 accuracy will be maintained and there are no direct connections to this network. The choice of clock sources, a local stratum-1 clock or acceptance of BSRF, is basically a business decision of the LECs, since both alternatives are technically suitable. Accepting synchronism from the BSRF network requires no additional equipment, no specialized installation, and no specialized maintenance. A stratum-1 clock requires special surveillance and maintenance, as well as trained personnel to operate the system. From now to the year 2000 there should be little change in synchronism strategy among the BOG LECs. Some might implement stratum-1 clocks as test beds should any major problems arise. As switches, the key for trouble-free synchronization will be accurate maps and records of transmission facilities utilized for synchronization purposes so that loops will be avoided. Each BOC LEG has one or more synchronization coordinators whose function it is to keep the maps current and provide technical help as required. Bell Communications Research, Incorporated Prior to divestiture the RBOC facilities were an integral part of the AT&T synchronization network. There were two digital synchronization networks under AT&T control, one for switched digital services

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness (4ESS) and the other for the DDS. Following divestiture the BOCs are responsible for synchronizing their own local access transport areas (LATA), which now number over 160. Some of the BOCs make use of AT&T facilities, while others maintain a PRS using cesium clocks or a DPS slaved to LORAN-C. In 1986 Bell Communications Research, Incorporated (Bellcore) published a plan for network synchronization9 that was later incorporated as an ANSI standard.10 The standard establishes tolerances in frequency, jitter, and wander for each clock stratum, as well as service objectives, impairment allocations, and strategies for interconnecting clocks in a network. It also specifies strategies for deployment and evolution, as well as protection strategies and use of nonstratified clocks. The Bellcore plan includes an extensive discussion of synchronizing principles for use within a physical facility or building. Each such facility uses a single building integrated timing supply (BITS) clock, which obtains timing from a clock of equal or lower-numbered stratum and has duplicated circuitry and provisions for backup timing via diverse routes. The BITS clock is distributed to all equipment in the facility in such a way that no timing loops will occur, either in normal operation or under abnormal operation involving any combination of backup links. A timing loop occurs when a timed clock receives timing from itself via a chain of timed clocks. Timing loops are undesirable for two reasons. First, all the clocks in the timing loop are isolated from the timing source (that is, a timing path does not exist from a timed clock to the timing source). Second, frequency instabilities may arise because of the timing reference feedback. MCI Communications Corporation MCI Communications Corporation (MCI) currently has six major switching centers operating in 12 plesiochronous islands. Each of these islands has a PRS consisting of a DPS with a disciplined oscillator, LORAN-C receiver, and antenna. The DFS operates as a stratum-1 clock and generates 308 kHz for analog equipment and 2,048 kHz for digital equipment to an accuracy of 10−11. The synchronizing tree is organized as master-slave with backup and has an expected service life beyond the year 2000. The MCI design pays careful attention to the multiple-station LORAN-C deployment. The LORAN-C receiver locks to the strongest station available, but needs only one station (master or slave)

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness in a chain. If this station is lost, the receiver automatically searches for and locks to one of the remaining stations. If no station can be found after 2.5 hours, the system enters holdover mode and switches to the disciplined oscillator. The disciplined oscillator uses a temperature-stabilized high-precision crystal oscillator normally locked to the LORAN-C receiver. Its control circuitry memorizes the intrinsic crystal drift and aging rate and corrects for these quantities during periods when the LORAN-C signal is lost. The DFS normally stays within 10−10 of the initial frequency during these periods for up to 10 days. Timing distribution within the island uses a pre-engineered spanning tree. The design avoids long synchronizing paths and allows few clock nodes on each path. Although each island operates with its own DFS and would ordinarily be considered plesiochronous, different islands may be synchronized to the same LORAN-C chain and thus be considered synchronous. Obviously, this would not be possible in all failure scenarios. MCI plans in the future to use BITS principles. The BITS design imposes a master-slave hierarchy for timing distribution within a physical facility or building. The design of the clock distribution equipment (CDE) includes provisions to smooth and “deglitch” the received timing signal, usually in the form of one or more DS-1 signals, and distribute it within the facility over a loop-free synchronization tree with backup. The CDE is also expected to provide higher-order synchronization for DS-3, Synchronous Optical Network (SONET), and so forth. CONTEL/ASC The CONTEL/ASC network includes extensive use of satellite and microwave facilities, in addition to digital fiber. Because of the Doppler shift inherent in satellite systems, special consideration must be given to buffering and timing issues. CONTEL/ASC uses a single dual-redundant DFS slaved to a LORAN-C receiver as the PRS for the national network. Each major central office is synchronized directly to the PRS with claimed minimum stability of 6 × 10−12 per day. US Sprint Communications Company The network of the US Sprint Communications Company (US Sprint) includes 45 switches at 28 locations interconnected by over 23,000

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-6 Synchronizing Failure Estimates Cause of Failure   PAMS Master/Slave No Diversity Diversity MTBFa 10 months 3 years 94 years MTTRb 2–8 hours 8 hours 2 hours Common failures Cable cuts, repeater failure Cable cuts, circuit failure, interface failure Change-over route miles of digital transmission facilites, mostly fiber operating at up to 565 Mbits/s. In comparison with other carriers, US Sprint has many long spans and few diversity routes. After study of several alternatives, US Sprint decided on an approach using a dual-redundant PRS at every switch location. The PRS plan is similar to MCI and involves the eventual deployment of duplexed LORAN-C receivers at all 28 switch locations. a MTBF: mean time between failure. b MTTR: maximum time to repair. 7. IMPACT OF SYNCHRONIZATION IMPAIRMENTS 1,12 An analysis of network reliability, based on various considerations of topology and route diversity, is shown in Table B-6. The mas-ter/slave column presumes a synchronizing tree with no diversity or alternate routing. PAMS is a distribution system that provides alternate routing with and without route diversity. A failure assumes the loss of all primary and secondary synchronization paths to clocks of lower strata and implies the use of local clocks operating at the stratum level of the equipment itself. The effects of synchronization impairments (disruption or failure) depend strongly on the type and severity of the underlying cause and on the particular user application. In the following subsec-tions, the effects of synchronization impairments will be assessed on transmission, network elements, and user applications. Subsequent sections will address the implications on NSEP survivability.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness Effects on Transmission Imperfect digital network synchronization can cause two principal types of transmission impairments: controlled slips and burst errors. Controlled Slips A controlled slip, which is the deletion or repetition of one frame (193 bits) of a DS-1 bit stream, occurs if the timing source of the transmitting network equipment at the sending end of a digital link is not synchronized with the timing source of the network equipment at the receiving end of that link. The interval between controlled slips is inversely proportional to the frequency offset between the two timing sources. The typical end-to-end performance objective for digital transmission under normal network conditions is one slip every 5 hours. Burst Errors A burst error, which is the transmission of a stream of errored bits, can be due to faulty transmission equipment (for example, broken line cards), protection switching, lightning strikes, and maintenance operations. Burst errors, while not caused by synchronization impairments, can be magnified and propagated by certain synchronization configurations. If a burst error of sufficient severity occurs on the incoming line that is providing timing to transmission equipment with a stratum-4 clock, then all of the output lines from that equipment may suffer magnified burst errors. (Channel banks, T-1 multiplexors, and digital PBXs typically have stratum-4 clocks.) This magnification and propagation of burst errors typically do not provide phase continuity when switching from one timing source (primary input, secondary input, or internal oscillator) to another. Reframes A reframe is the operation of recovering or initially finding the reference bit in a 125-µs DS-1, DS-2, and so on, frame. Reframes can occur when equipment is first turned on, when a protection switch occurs, and when the maintenance operations are performed. As a practical matter, reframes are fairly rare on today’s telephone network and are of brief duration from a few milliseconds to several seconds. However, reframes may be expected to arise in

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-7 Dual-Tone Multifrequency Signaling Survivability System Stratum Days to Exceed Frequency Offset L5E 2 170 TD radio 3 120 TD radio 4 12 NSEP scenarios when the damage level is high enough to produce any of the above situations and may last longer than several seconds in extreme cases. Effects on Network Elements Synchronization impairments can affect any frequency-sensitive network element, including both digital and analog systems. The following analog systems are most frequency sensitive, in decreasing order of sensitivity: Dual-tone multifrequency signaling (DTMF) Multifrequency signaling (MF) Single-frequency signaling (SF) Voice. DTMF, when used to address the local switching office, almost never undergoes frequency translation and, hence, is dependent only on the telephone set from which it originates. When DTMF is used for end-to-end signaling, such as in a Nationwide Emergency Telecommunications Service (NETS) application, then the frequency offsets must be controlled within ±10.5 Hz. Assuming worst-case scenarios for selected transmission systems, the estimated interval this requirement can be met, following loss of outside timing source, is shown in Table B-7. Synchronization impairments due to controlled slips result in a loss or replication of a 125-µs frame. The impact on DTMF signaling could be a missed digit if a minimum 50-ms DTMF signal was being transmitted. Speech is virtually unaffected. The maximum worstcase slip rate in digital systems is about 265 slips/hour. This is an order of magnitude better than the bit error rate (BER) limit of 10−4 on a T-1 trunk. There may be some equipment operating with

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-8 Impairments Due to Controlled Slips Application Effect on Transmission Quality Voice None Voiceband data < = 1,200 bits/s, none; > 1,200 bits/s, errors Secure voice Loss of secure connection, session rekey required Digital data Single-byte dropped or repeated Facsimile Small, illegible areas Video Mild picture breakup and freezing, garbled audio TABLE B-9 Impairments Due to Isolated Burst Errors Application Effect on Transmission Quality Voice Mild noise Voiceband data Data errors Secure voice Loss of secure connection, session rekey required Digital data Severe data loss Facsimile Large, illegible areas Video Severe picture breakup and freezing, severely garbled audio 2,400-bits/s modems that may well be affected by 265 slips/hour, but new designs planned will eliminate the known problems. Effects on User Applications It is not possible, in general, to specify with certainty the effects of particular transmission impairments on user applications, because these effects often depend on the exact timing of the impairment and on the exact contents of the user information being transmitted. It is possible, however, to specify what effects on user applications are typical for different types and severity of transmission impairments. Tables B-8 and B-9 summarize these effects for three levels of transmission impairment for typical user applications. The three transmission impairments considered in the table are isolated controlled slips, isolated burst errors (for example, a 100-ms period with a BER of 10−2 every 4 s), and consecutive burst errors

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness TABLE B-10 Impairments Due to Reframes Application SES < = 1 s OSES > 5 to 7 s Channel bank, cross-connect Audible but inoffensive Disconnect after 5 s Data transmission   Re-login required Analog facsimile Some unreadability in patches Unreadable Video codec Detectable effect Still usable (for example, a 250-ms period with a BER of 10−2 each second for eight consecutive seconds). The six user applications considered are voice 64 kbits/s pulse code modulation (PCM) or 32 kbits/s Adaptive Delta PCM (AD-PCM), voiceband data (with modem), secure voice (STU III with 2,400-bits/s-modem), digital data (64 kbits/s), facsimile (group 3), and video (1.544-Mbits/s) coder-decoder (codec). Table B-8 shows the effects of controlled slips on these applications. While users might notice these effects, most would probably elect to continue the present connection, especially if error-detection and correction procedures were incorporated in the protocol design. Table B-9 shows the effects of isolated burst errors on the applications. Users would certainly notice these effects and may choose to abandon the connection and retry. In the case of consecutive burst errors the user most likely would find the connection unusable and abandon it. In fact, the transmission equipment, noting the severely degraded state of the link, usually declares it inoperable and drops the connection itself. The effects of reframes are summarized in Table B-10. The kind of application is listed in the first column and the second and third columns list two successively more severe disruption classes. SES means severely errored (more than 10−3 BER) seconds and CSES means consecutive occurrences of SES seconds. 8. SENSITIVITY OF NATIONAL SECURITY EMERGENCY PREPAREDNESS TO SYNCHRONIZATION IMPAIRMENTS 12 The preceding discussion has emphasized the mechanism and effects of synchronization impairments within the telephone network itself

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness and the resulting impact to the various user applications. This section assesses the impact of these impairments to the programs of the NLP/NSEP. These programs include NETS, Commercial Satellite Interconnectivity (CSI), and Commercial Network Survivability (CNS). The main motivation in this paper is to determine whether there may be problems due to synchronization impairments in the public switched networks that could lead to unacceptable performance or loss of useful NSEP capabilities of any of the various services required by the National Communications System (NCS). This is to be contrasted with the question of whether there may be perceptible degradation under natural or man-made abnormal stresses—for example, measurable increase in bit error or message delivery time which, of course, may be most important items under normal conditions for public network users. Degradation or loss of services due to natural effects such as storms and earthquakes will be of limited geographical extent. Probably the worst case that one can consider is a complete loss of the primary synchronization of stratum-1 digital switches. However, such loss would have negligible effect on NSEP services because the system architecture and other strata in the system would be adequate for very long time periods. Synchronization degradation from man-made events includes vandalism, sabotage, direct attack with nuclear weapons, and so forth. Reflection on the attractiveness of attacking synchronization elements compared to other system components such as common channel signaling (CCS), large switches, and so forth, led to a consensus that synchronization was not a major player in such postulated events and will not be through the year 2000. Nationwide Emergency Telecommunications Service In general, there are two potential synchronization and timing concerns in NETS: (1) frame slips, when the divided network must be used in a plesiochronous mode and (2) resynchronization. Overall, neither concern is of major consequence if network issues are separated from terminal device (or customer-premises equipment) issues. The network will hold up quite well under frame slips. Even under severe conditions, slips appear to cause little network impact. Resynchronization is the more catastrophic event. Here links

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness are interrupted, circuits are lost, and surviving islands may be forced to operate with relatively unstable timing. Commercial Satellite Interconnectivity The only new issue the CSI plan raises relative to synchronization are the effects of the satellite transmission characteristics of the connected T-1 link. The timing hierarchy algorithms used on the public switched networks takes synchronization from the highest surviving stratum. A connection between two main islands by satellite will encompass two more switches. This means that it is highly likely that there will be a stratum-2 clock, or better, in one of the two islands. As synchronization timing is taken from the better of the two sources, synchronization does not appear to be an issue other than through the jitter corruption of the timing through the T-1 link and through the satellite channel delay. This does not appear to be a significant issue. Commercial Network Survivability Since the CNS program essentially offers only a skinny analog bandwidth voiceband connection, the usual concerns about timing and synchronization are not applicable. (It is noted that in the older single sideband [SSB] radios used, crystal frequency setting was marginal. Human operator tweaking was required to keep the links going. Parenthetically, this is standard operating practice for these older radio units. More important than the results of the early make-do type experiments was the concept itself.) Better technology radio equipment would permit less manual interaction in setting up connections. As the evolution to T-1 is moving along rapidly, digital multiplexed carriers may be the more likely long-term direction of evolution. The analog connection may be an interim step along the way, but one not to be overlooked. 9. CONCLUSION AND RECOMMENDATION From the foregoing analysis the committee reaches the following conclusion and recommendation: No significant synchronization timing issues for national security emergency preparedness appear to exist, because timing is set by the connected surviving access tandem.

OCR for page 91
Growing Vulnerability of the Public Switched Networks: Implications for the National Security Emergency Preparedness As existing network synchronization levels already exceed those required for national security emergency preparedness, no action need be taken to increase the robustness of network synchronization beyond existing standards for normal network operation; designers of terminal devices should engineer them to operate satisfactorily under system synchronization standards. NOTES 1.   Briefing material on file with the Committee on Review of Switching, Synchronization and Network Control in National Security Telecommunications. 2.   Time and Frequency Dissemination Services. 1979. NBS Special Publication 432. Washington, D.C.: U.S. Department of Commerce. 3.   Lindsay, W.C., and A.V.Kantak. 1980. Network synchronization of random signals. IEEE Transactions on Communications COM-28 (8 August): 1260– 1266. 4.   Braun, W.B. 1980. Short-term frequency effects in networks of coupled oscillators. IEEE Transactions on Communications COM-28 (8 August): 1269– 1275. 5.   Mitra, D. 1980. Network synchronization: Analysis of a hybrid of master-slave and mutual synchronization. IEEE Transactions on Communications COM-28 (8 August): 1245–1259. 6.   Jordan, E.C., ed. 1985. Reference Data for Engineers, 7th ed. New York: H. W.Sams & Co. 7.   Davies, K. 1966. Ionsopheric Radio Propagation. NBS Monograph 80. Washington, D.C.: National Bureau of Standards. 8.   Munter, E.A. 1980. Synchronized clock for the DMS-100 family. IEEE Transactions on Communications COM-28 (8 August): 1276–1284. 9.   Bell Communications Research, Incorporated. 1986. Digital Synchronization Network Plan. Technical Advisory TA-NPL-000436. Livingston, N.J.: Bell Communications Research, Incorporated. 10.   American National Standards Institute. 1987. ANSI T1.101–1987: Synchronous Interfaces for Digital Networks. New York: American National Standards Institute. 11.   U.S. Naval Observatory (private communication). 1988. 12.   Information provided by expert committee members. 13.   Beser, J., and B.W.Parkinson. 1982. The application of NAVSTAR differential GPS in the civilian community. Navigation 29(Summer).