Classifications

• • G—PHYSICS
  • G04—HOROLOGY
  • G04G—ELECTRONIC TIME-PIECES
  • G04G7/00—Synchronisation
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04J—MULTIPLEX COMMUNICATION
  • H04J3/00—Time-division multiplex systems
  • H04J3/02—Details
  • H04J3/06—Synchronising arrangements
  • H04J3/0635—Clock or time synchronisation in a network
  • H04J3/0638—Clock or time synchronisation among nodes; Internode synchronisation
  • H04J3/0641—Change of the master or reference, e.g. take-over or failure of the master
• • G—PHYSICS
  • G04—HOROLOGY
  • G04R—RADIO-CONTROLLED TIME-PIECES
  • G04R40/00—Correcting the clock frequency

Definitions

• the present invention relates generally to a method, time consumer system, and computer program product for maintaining accurate time on a timing device during periods when the reference clock is not accessible and/or is compromised.
• the field of use includes but is not limited to
• the problem of "holdover” is the problem of maintaining an accurate time even when updates from the reference clock are interrupted for a relatively extended interval.
• the problem of compromise detection involves detecting frequency changes in the reference clock. Providing a method for holdover and protecting against compromise is important in
• Particular embodiments of the present invention include methods, systems, and computer program products for providing holdover capability and compromise detection by deriving high quality "secondary reference frequency" information from a secondary clock (or collection of clocks) despite inaccuracies in clock signal and/or high signal delay variation.
• the methods, systems, and computer program products of the present invention permit the use of a time signal delivered via a network, a network within an enterprise or even the Internet, as a source of precise frequency information.
• Particular embodiments of the disclosed solution derive the secondary reference frequency and apply this derived frequency information to holdover and compromise detection.
• the methods, systems, and computer program products of the present invention enable ubiquitous availability and low cost of access to such time sources.
• the secondary clock is used to measure both the current frequency of the local clock and the reference clock. If the reference clock frequency changes, the possibility of compromise can be detected. If the reference clock becomes unavailable, the frequency measure derived from the second clock can be used to
• a method for maintaining accurate time on an ideal clock of a timing device includes receiving time information from a local clock, a reference clock, and one or more secondary clocks. The method further includes calculating frequencies for the local clock, the reference clock, and the one or more secondary clocks. The method further includes comparing the calculated frequencies of the reference clock to the calculated frequencies of the one or more secondary clocks. The method further includes detecting a holdover and/or a compromise situation based on the comparison. The method further includes syntonizing the ideal clock to one or more of the calculated frequencies.
• a computer program product includes a non-transitory computer readable medium having stored thereon executable computer instructions for maintaining accurate time in an ideal clock of a timing device.
• the computer instructions are configured to cause a computer to receive time information from a local clock, a reference clock, and one or more secondary clocks.
• the computer instructions are configured to cause a computer to calculate frequencies for the local clock, the reference clock, and the one or more secondary clocks.
• the computer instructions are configured to cause a computer to compare the calculated frequencies of the reference clock to the calculated frequencies of the one or more secondary clocks.
• the computer instructions are configured to cause a computer to detect a holdover and/or a compromise situation based on the comparison.
• the computer instructions are configured to cause a computer to syntonize the ideal clock to one or more of the calculated frequencies.
• a time consumer can use the secondary clock to measure a time interval, detect the number of "ticks" counted by the local clock during that time interval and compute local clock frequency. If the local clock counts N ticks during the interval and the secondary clock measures elapsed time T, the average duration of a clock tick is T/N and the average frequency is N/T.
• holdover can be implemented by using the secondary reference frequency to correct for drift in the local clock or local clocks.
• a computing device connected to a GPS time source as the reference clock may also be able to receive time over the Intemet from the National Institute for Standards and Technology (NIST) clocks using NTP.
• NIST National Institute for Standards and Technology
• the NIST intemet clock in general, is not suitable for the purpose of synchronizing the ideal clock because packet delay variations can be several milliseconds long on current Internet technology. However, over intervals of time significantly longer than the delay variation, the elapsed time will be accurate. This permits the ideal clock computation to update its frequency estimates for the local clock and accurately compensate for variations in local clock frequency.
• the time consumer may track multiple overlapping intervals from the secondary clock and use these to find patterns in the frequency changes of the local clock to improve frequency
• the time consumer is a server computer
• the reference time is a bus card with a GPS clock module
• the local clock is an oscillator connected to a processor "time stamp register" which operates as a counter
• the secondary clock is an internet connection to the National Institute of Standards and Technology (NIST) time via a network device connected to a local network and then out to the Internet via switches and routers.
• NIST National Institute of Standards and Technology
• the time consumer is a base station for a cellular telephone system
• the reference clock is a GPS receiver
• the local clock is a TCXO oscillator
• the secondary clock is a receiver for a time signal encoded in a radio signal from a neighboring base station.
• a server computer may use multiple secondary clock sources connected to multiple remote clocks over a network.
• the elapsed time may be computed as a weighted average of the measured elapsed times or via a voting mechanism or some other means of increasing the reliability of the frequency measure from the secondary sources.
• the secondary clock frequency may come from the same physical device as the primary external reference clock.
• a number of commercial "network time clock" devices that act as PTP master clocks synchronized to a GPS source in one mode, can switch to modes where the quality of the synchronization is degraded or the protocol changes.
• One such device will switch to a holdover mode using an internal temperature controlled oscillator or rubidium timer when GPS reception is interrupted or of poor quality.
• the quality of that holdover time may not be as high as the quality of the holdover time that can be produced using the method of the present invention by discarding the synchronization information from the PTP source while using its frequency information as in the example above.
• some devices have a failure mode in which they cannot produce PTP but fall back on a less accurate NTP implementation.
• the present invention permits the client system to use the NTP source for frequency information during a holdover.
• FIG. 2 is a block diagram showing a time consumer connected to a reference clock and a plurality of secondary clocks, in accordance with exemplary embodiments of the present invention.
• Fig. 4 is a timeline illustrating the concept of signal delay variation, in accordance with exemplary embodiments of the present invention.
• Fig. 5 is a timeline illustrating the simultaneous recording of secondary time and local frequency information, in accordance with exemplary embodiments of the present invention.
• Fig. 6 is a timeline illustrating multiple secondary time and local frequency measurements, in accordance with exemplary embodiments of the present invention.
• Fig. 7 is a bus card that contains a GPS receiver which produces a running time value, in accordance with exemplary embodiments of the present invention.
• Fig. 8 is a block diagram of a time consumer, in accordance with exemplary embodiments of the present invention.
• Fig. 9 is a flow chart illustrating the steps performed by the time consumer to calculate local frequency drift, in accordance with exemplary embodiments of the present invention.
• Fig. 10 is a flow chart illustrating the steps performed by the time consumer to perform either holdover or detect a compromise, in accordance with exemplary embodiments of the present invention.
• Fig. 11 is a flow chart illustrating a method for maintaining accurate time on an ideal clock of a timing device.
• the methods and devices of the present invention can derive frequency data from a possibly inaccurate source of time and use that data to detect frequency changes in a reference clock and to maintain correct time during failures of the reference clock.
• a radio which serves as a reference clock
• a highly accurate chronometer might be purchased and synchronized with radio updates so that when the radio is unavailable, e.g., during a holdover, a wrist watch may be periodically synchronized to the chronometer. If the accuracy of the chronometer is higher than the wrist watch, the wrist watch time can be maintained close to the reference time until the radio is available again.
• the clock in the radio station temporarily slows, e.g., a compromise, and the wrong time is announced.
• a secondary time source may be used to detect such compromise.
• a “clock” is a device or combination of devices and software that produces a time signal (that may, for example, be an electrical signal or a value in a register or in a network packet).
• a “time consumer” is a computing or communications or control device that reads or receives or detects time signals from clocks.
• a “time consumer” may contain one or more "local clocks” and may communicate with one or more external clocks.
• a time consumer generates an "ideal clock” that is in some way derived from the signals or values obtained from external clocks in
• a reference clock is a clock that has been chosen as the authoritative time source.
• the problem of "synchronization” in this context is the problem of keeping the ideal clock adjusted so that it produces a time as close as possible to the time produced by the reference clock.
• the problem of "holdover” concerns maintaining synchronization during an interruption or degradation of the signal from the reference clock.
• “compromise” involves detecting changes in the frequency of the reference clock.
• the time consumer is a computer server
• the reference clock is a device following signals received from the Global Positioning System (GPS) satellites
• the local clock is a computer crystal oscillator driving a counter.
• GPS Global Positioning System
• the crystal oscillators provided on commercial computer boards are known to change frequency during operation, particularly, but not only, as device temperature changes.
• the problem of compromise of the reference time source involves temporary inaccuracies in the external time source. That is, if the external time source changes frequency for some reason, it would be useful to detect this and then go into holdover mode until the drift ceases or some external source (such as an operator) validates the signal.
• the GPS signal has been shown to be susceptible to "spoofing" in which someone deliberately modifies the time signal.
• the provision of a rubidium clock or temperature controlled or “ovenized" oscillator is one method for cross checking the authoritative time source.
• the external reference time may also be obtained from alternative satellite systems (such as COMPASS and GALILEO), from an atomic clock, over an IRIG signal, a Pulse Per Second (PPS) signal, or via a computer network using IEEE 1588 Precision Time Protocol (PTP), the Network Time Protocol (NTP) or some other messaging protocol.
• PPS Pulse Per Second
• NTP Network Time Protocol
• synchronous Ethernet can be used to directly distribute a time frequency that might be used to verify the external time reference or to correct the external time reference.
• the crystal oscillator might be a counter, or a MEMS device, or something else that provides a local time interval.
• Syntonization involves determining the frequency of the local clock so that intervals computed by that source can be adjusted to match the interval that would have been measured by the external reference.
• the time broadcast on the radio might be written down and the offset from the wristwatch computed.
• the drift of the wristwatch over the interval could be computed and this would give information on the difference between the frequency of the reference time source and the frequency of the wristwatch.
• the wristwatch time could be "corrected," at times between radio updates, by using the frequency adjustment.
• the frequency differential between the crystal oscillator and the GPS clock would be computed and used to compute the ideal time.
• Signal Delay Variation is the time variation of the time it takes for clock data (the signal) to travel from the clock to the time consumer and become available to the time consumer for calculation.
• SDV is called Packet Delay Variation (PDV).
• PDV Packet Delay Variation
• the signal from the clock carries a time that is inaccurate by some variable quantity when it arrives at the time consumer.
• SDV is very low for the local clock, and to be usable for synchronization purposes, the reference clock SDV must also be small. If the SDV of the reference clock increases too much, a holdover condition can be triggered.
• Fig. 1 is a block diagram showing a time consumer 100 connected to both a reference clock 110 and a secondary clock 120.
• time consumer 100 may be connected to a reference clock 110 and secondary clock 120.
• the connections between time consumer 100 and reference clock 110 and secondary clock 120 may be hardwired so that the connection is within the same computing device.
• a hardwire connection may be used across different computing device.
• one or both of the connections may be across one or more communications networks via a cable or wireless connection.
• Fig. 2 is a block diagram showing a time consumer 100 connected to a reference clock 110 and a plurality of secondary clocks 120a- 120n.
• the "secondary clock” may be an aggregated collection of clocks.
• the connections between the secondary clocks 120a- 120n and the time consumer 100 may consist of a hardwired connection in the same computing device, or a wired or wireless connection across one or more computing devices, or any combination of hardwired, wired, and wireless connections.
• secondary clock time information is transmitted by a secondary clock 120 to a communications network 360.
• the ideal clock 300 located in time consumer 100, may receive the secondary time information from secondary clock 120 via communications network 360.
• the secondary clock data may be, for example, time information from one or more NIST clocks.
• a processor may be connected to a crystal oscillator 320 that drives a running counter called a Time Stamp Counter (TSC) 340.
• TSC Time Stamp Counter
• the TSC 340 is the local time source.
• a network device connects the time consumer 100, which is a computing device, to a computer network 360 connected to the Internet, and the time consumer 100 periodically requests the time from NIST or another secondary clock 120 using NTP.
• the data for the ideal clock 300 is obtained by periodically polling the reference clock, or GPS 310, for reference time, using that information to compute the actual frequency of the local clock and then, between updates, using the TSC interval since the last time read from the reference time, plus current frequency estimate for the local time source. If the system detects that the reference clock is no longer working, the average TSC count over an interval previously computed, for example over a 5 minute interval, is computed against the time reported by the Internet connection to the secondary clock 120, or NIST clock. In practice, these computations may also involve more sophisticated calculations, including the use of smoothing and filtering methods.
• a time consumer 100 contains a network card with an oscillator 320 that is connected over a high speed local network 360 to a device that acts as a IEEE 1588 PTP "master clock.”
• the processor is connected to a crystal oscillator 320 that drives a running counter called a Time Stamp Counter (TSC) 340.
• TSC Time Stamp Counter
• the TSC 340 is the local time source.
• the same or a second network device connects the time consumer 100 to a computer network 360 connected to the Internet and the device periodically requests the time from a secondary clock 120 using NTP.
• the ideal clock 300 is obtained by some combination of time provided in PTP packets, delay information from the network card oscillator 320 (used to reduce PDV imprecision on PTP packets), historical data with some smoothing/filtering algorithms, and a computation of the actual frequency of the local clock. Then, between PTP updates, the ideal clock 300 is obtained by using the TSC interval, the last time read from the reference time, plus current frequency estimate for the local time source. If the system detects that the FTP reference clock 310 is no longer working, the same method used in the previous embodiment is employed to compute the actual frequency of the local time source.
• the methodology is applicable to any type of time consumer 100 device that can be configured to communicate to a secondary clock 120 via a network, including, for example, a mobile device such as a wireless telephone, a personal digital assistant, an electronic reader (e.g. an Amazon Kindle), or a personal tablet computer (e.g., an Apple iPad); a server computer; a laptop computer; and a personal computer.
• a mobile device such as a wireless telephone, a personal digital assistant, an electronic reader (e.g. an Amazon Kindle), or a personal tablet computer (e.g., an Apple iPad); a server computer; a laptop computer; and a personal computer.
• FIG. 4 a timeline illustrating the concept of signal delay variation is shown.
• a signal 400 carrying time information is produced by the reference clock 110 at time to and only arrives at the time consumer 100 at time t 3 , as illustrated by the actual time timeline 410.
• the point at which the time information, produced at time to, is received at the time consumer 100 at t 3 , the time information is no longer accurate.
• the delay between to and h varies, it is difficult for the time consumer to accurately compensate for delay.
• the JSC count n 0 51 1 could be recorded at time consumer 100 just as a time update from a secondary clock, such as a NIST clock, to 501 is received.
• the TSC count ni 512 may be recorded at the time consumer 100 just as a secondary clock update ti 502 is received, indicating the passage of some computed time interval. This information may be used to update the computed frequency of the local clock with the simplest calculation being that the frequency is (ni-n 0 )/(ti-to).
• the time value in the packet is 8 milliseconds less than the actual time.
• the request took 5 milliseconds to reach the clock
• the clock took 4 milliseconds to produce the response
• the response took 1 millisecond to reach the client
• the time value in the packet is only 1 millisecond less than the actual time.
• the time consumer 100 will use historical data to determine the minimal response time delay that can be expected, and then begin an interval only after some request has produced a response with such a delay.
• the same method can be used by a time consumer 100 to determine when to end the interval - since the important quality is how exactly the interval is measured, not whether the interval is of some preset duration. So, an interval of 122 seconds plus or minus 5 milliseconds is more precise than one of exactly 120 seconds plus or minus 20 milliseconds.
• the time consumer 100 can also track multiple intervals and overlapping intervals to better measure actual frequency changes in the local time source. In cases where network quality permits, the time scale can be reduced: given a high quality connection to a secondary time source, the intervals might be milliseconds, or even of shorter duration. However, the intervals should be long enough to permit accurate frequency time calculation.
• a received NTP time is associated with a delay ranging between 1 ⁇ 2 and 10 milliseconds.
• the reference clock 110 may be a bus card 710 that includes a GPS receiver and module 330, and electronic components including one or more oscillators 720 and control devices 700 that produce a running time value in a register that can be read by the main processor 730 of the time consumer 100.
• the reference time from the bus card can be accurate down to a few tens of nanoseconds.
• time consumer 100 may be a communication device that includes a processor 730, a network interface device 810, a memory 820, and an ideal clock 300.
• Computer readable program code may be stored in a non-transitory computer readable medium/memory 820, such as, but not limited to, magnetic media (e.g., a hard disk), optical media (e.g., a DVD), memory devices (e.g., random access memory), and the like.
• time consumer 100 may be connected to an external or internal database to store, for example, the time and clock frequency information disclose above.
• a flow chart illustrating the steps performed by the time consumer to calculate local frequency drift are displayed.
• the time consumer 100 collects time data from reference clock 1 10 in step S900.
• step S901 if enough time data is collected from reference clock 1 10, then the frequency of the ideal clock 300 is matched to the frequency of reference clock 110 in step S902. If there is not enough time data, then the time consumer 100 continues to collect time data from reference clock 1 10 in Step S900.
• an updated frequency estimate is calculated in step S903.
• the estimate may be calculated using a variety of methods, such as smoothing and filtering methods
• step S1000 a flow chart illustrating the steps performed by the time consumer to perform either holdover or detect a compromise are displayed.
• ideal clock 300 of time consumer 100 will continue to operate in step S1000 based on a certain reference clock 110 frequency.
• step S1010 if the reference frequency has not changed, then the ideal clock 300 will continue to operate in step S1000 based on the currently used reference clock 110 frequency.
• step S1020 it is determined if the reference clock 110 has failed. For example, as described above, the frequencies of the reference clock and the secondary clock can be compared. If the frequencies change with respect to each other, then the frequency of one has changed and thus a possible compromise or holdover has been detected. If the reference clock 110 has not failed, then there is no
• step S1000 if the reference clock 110 has failed, then a holdover procedure is performed in step S1030.
• the holdover procedure is implemented by using the secondary clock 120 reference frequency to correct for drift in the local/ideal clock or clocks 300.
• a flow chart illustrating a method for maintaining accurate time on an ideal clock of a timing device is displayed.
• time information from a local clock, a reference clock, and one or more secondary clocks may be received in step S1100.
• the received time information may be used to calculate the frequencies of the local clock, the reference clock, and the one or more secondary clocks.
• the calculated frequencies of the reference clock may be compared to the calculated frequencies of the one or more secondary clocks.
• step S1130 in some embodiments, a holdover and/or a compromise situation may be detected based on the comparison
• step S1140 in some embodiments, the ideal clock may be syntonized to one or more of the calculated frequencies.

Landscapes

• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Engineering & Computer Science (AREA)
• Computer Networks & Wireless Communication (AREA)
• Signal Processing (AREA)
• Electric Clocks (AREA)
• Synchronisation In Digital Transmission Systems (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (1)

Family

ID=49778030

Family Applications (1)

Country Status (5)

Cited By (22)

Families Citing this family (34)

Citations (3)

Family Cites Families (16)

• 2013
  • 2013-06-28 IN IN593DEN2015 patent/IN2015DN00593A/en unknown
  • 2013-06-28 US US13/931,149 patent/US9348321B2/en active Active - Reinstated
  • 2013-06-28 WO PCT/US2013/048566 patent/WO2014005016A1/en not_active Ceased
  • 2013-06-28 CA CA2877720A patent/CA2877720C/en active Active
  • 2013-06-28 EP EP13810228.0A patent/EP2867740B1/en active Active
  • 2013-06-28 EP EP17001444.3A patent/EP3282340A1/en not_active Withdrawn
• 2016
  • 2016-04-28 US US15/140,784 patent/US9671761B2/en active Active

Patent Citations (3)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events