Method and apparatus for iterative synchronization of two or more electronic devices

• US 8,634,341 B1
• Filed: 06/12/2006
• Issued: 01/21/2014
• Est. Priority Date: 10/11/2005
• Status: Active Grant

• Current Assignee: Atmel Corporation (Microchip Technology Incorporated)
• Original Assignee: Atmel Corporation (Microchip Technology Incorporated)
• Inventors: Vleugels, Katelijn, Thompson, Barry
• Primary Examiner(s): Rinehart, Mark
• Assistant Examiner(s): YEUNG, MANG HANG
• Application Number:
  US11/423,499
• Time in Patent Office: 2,780 Days
• Field of Search: 370/503, 370/324, 370/350, 370/395.62, 370/484, 370507-520, 375354-376, 714/498
• US Class Current: 370/324
• CPC Class Codes: G06F 5/16 Multiplexed systems, i.e. using two or more similar devices which are alternately accessed for enqueue and dequeue operations, e.g. ping-pong buffers
• CPC Class Codes: H04J 3/06 Synchronising arrangements for television systems H04N5/04; bit-synchronisation H04L7/00
• CPC Class Codes: H04J 3/0664 unidirectional timestamps
• CPC Class Codes: H04W 56/00 Synchronisation arrangements
• CPC Class Codes: H04W 56/002 Mutual synchronization

Abstract

Dynamic adjustment of offset between timebases of separated electronic circuits is provided based on an amount of time between synchronization events, wherein a synchronization event is an event involving communication across a separation between the separated electronic circuits in order to synchronize the timebases. The dynamic adjustment can be dynamically or adaptively adjusting a margin of error to account for drift that may have occurred since a prior synchronization event. Adjustments might be done by applying a learning sequence to quantify and adjust for static or slowly varying delays between the timebases and coordinating the timebases using output of the learning sequence. The learning sequence might quantify and adjust for static or slowly varying delays and frequency drift in a communication system having a master timebase device and slave timebase devices wherein slave timebase devices are configured to adjust their local timebases to coincide with the master timebase by obtaining timing/synchronization information during one or more events, the events being communication and/or synchronization events, analyzing the timing/synchronization information to derive timing errors due to system nonidealities, and applying timing correction based on timing errors derived by the analyzing.

Claims

1. A method comprising:

obtaining timing or synchronization information during a plurality of synchronization events, wherein a synchronization event is an event involving communication across a separation between at least two separated electronic circuits of a plurality of electronic circuits in order to synchronize timebases of the separated electronic circuits, the plurality of electronic circuits each having at least one component used to output a signal that represents a timebase, wherein which events occur in each of the plurality of electronic circuits according to the timebase of the respective electronic circuit;

synchronizing the timebases of the separated electronic circuits as part of at least one of the plurality of synchronization events, using the obtained timing or synchronization information; and

providing for a dynamic adjustment of offset between the timebases of the at least two separated electronic circuits at an adjustment time, wherein the dynamic adjustment is based on an amount of time since a prior synchronization event, wherein the adjustment time is distinct from each of the plurality of synchronization event times and wherein the dynamic adjustment takes into account estimated offsets for a time period between the time of one of the synchronization event times and the adjustment time, and wherein the estimated offsets are determined without communication across the separation occurring at the adjustment time.

2. The method of claim 1, wherein the dynamic adjustment comprises at least one of dynamically or adaptively adjusting a margin of error to account for drift that may have occurred since a prior synchronization event.

3. The method of claim 1, further comprising adjusting the offset based on requirements of a communication system.

4. The method of claim 3, further comprising adjusting the offset to improve robustness of the communication system.

5. The method of claim 1, further comprising:

scheduling a wake-up event at the slave device taking into account the dynamic adjustment of offset such that the timing of the wake-up event depends on the dynamic adjustment, wherein the wake-up event is an event that triggers a transition to a signal-listening mode from a not-listening mode.

6. The method of claim 1, wherein the synchronization event is part of a communication sequence that is usable for other than synchronization.

7. A method of determining an adjustment between timebases of separated electronic circuits comprising:

receiving the timing signal at one of a plurality of synchronization event times, a synchronization event time being a time of one of a plurality of synchronization events wherein a synchronization event is an event involving communication across a separation between at least two separated electronic circuits of a plurality of electronic circuits in order to synchronize the timebases of the separated electronic circuits, the plurality of electronic circuits each having at least one component used to output a timing signal that represents a timebase, wherein which events occur in each of the plurality of electronic circuits according to the timebase of the respective electronic circuit;

applying a learning sequence to quantify and adjust for static delays or varying delays between the timebases; and

coordinating the timebases using output of the learning sequence at an adjustment time that is distinct from each of the plurality of synchronization events event times, such that at least one component used to output a signal that represents the timebase is altered so that the timebase is altered, thus taking into account estimated offsets for the time period between the time of at least one of the synchronization events and the adjustment time and wherein the estimated offsets are determined without communication across the separation occurring at the adjustment time.

8. The method of claim 7, wherein the learning sequence comprises:

scheduling synchronization events spaced in time such that frequency drift creates less of a timing error than non-drift timing errors, and

quantifying the static delay or varying delay in the system as a timing error that is introduced between a master timer of a master device and a slave timer of a slave device, over a time that expires between two synchronization events.

9. The method of claim 8, further comprising:

measuring the timing error once or multiple times; and

performing additional processing to determine nonsteady variances.

10. The method of claim 9, wherein the additional processing comprises computation of a minimum timing error in a series of multiple error measurements.

11. The method of claim 9, further comprising:

sending, from the master device at a start of a synchronization event following a medium access procedure, a signal to the slave device; and

measuring, at the slave device, the timing error as a difference between a start of a synchronization event as indicated by the slave device'"'"'s local timebase and the reception of the signal from the master device.

12. The method of claim 11, wherein timing error is measured at the slave device, and information is sent back to the master device such that the additional processing is entirely implemented at the master device.

13. The method of claim 11, wherein timing error is measured at the slave device, and information is sent back to the master device such that the additional processing is partially implemented at the master device and is partially implemented at the slave device.

14. The method of claim 9, wherein the slave device sends a signal at the start of a synchronization event as indicated by the slave device'"'"'s local timebase following a medium access procedure, and the master device measures the timing error as the difference between the start of a synchronization event as defined by the master timer and the reception of the signal from the slave device.

15. The method of claim 9, wherein the slave device sends a signal to the master device, the signal containing timing information that defines a start of the synchronization event as indicated by the slave device'"'"'s local timebase.

16. A method of quantifying a frequency drift between a master timer of a master device and a slave timer of a slave device, the master device and the slave device being devices that include electronic elements that rely on timing, the method comprising:

obtaining a plurality of timestamps, each timestamp obtained from a packet received at the master device or the slave device and each timestamp representing a time of sending the packet containing that timestamp, thereby defining a plurality of timestamp times;

using at least one of the plurality of timestamps to adjust timing of one of the electronic elements;

applying a learning sequence to the plurality of timestamps to quantify the frequency drift at a MAC level, such that the frequency drift can be measured independent of timing values of a PHY level, and wherein the frequency drift is quantified as of a sample time that is other than-one of the plurality of timestamp times, and wherein the learning sequence takes into account estimated frequency drifts for the nonzero time period between-one of the plurality of timestamp times and the sample time, without the packet being received at the sample time.

17. The method of claim 16, wherein the learning sequence comprises scheduling of synchronization events far enough apart in time such that master-slave frequency drift is a dominant error component and frequency drift is derived from an error measurement.

18. The method of claim 17, wherein the timing error is measured at least once and additional processing is performed to compute the master-slave frequency drift in the presence of other timing uncertainties.

19. The method of claim 18, wherein the additional processing comprises computation of a maximum error or an average error in the series of multiple error measurements.

20. The method of claim 18, wherein the slave device sends a signal at the start of a synchronization event as indicated by the slave device'"'"'s local timebase following a medium access procedure, and the master device measures the timing error as the difference between the start of a synchronization event as defined by the master timer and the reception of the signal from the slave device.

21. The method of claim 17, wherein the slave device sends a signal containing timing information that defines a start of the synchronization event as indicated by its local timebase to the master device.

22. The method of claim 21, wherein the master device computes the frequency drift based on the timing information received from the slave device and the start of the synchronization event as defined by its master timer.

23. A method of applying a learning sequence to quantify and adjust for static or varying delays and frequency drift in a system comprising at least a slave device and a master device, comprising:

obtaining timing or synchronization information during a plurality of synchronization events, wherein a synchronization event is an event involving communication across a separation between the master device and the slave device in order to synchronize a master timer and a slave timer, wherein a plurality of electronic circuits each has at least one component used to output a signal that represents a timer, wherein which events occur in each of the plurality of electronic circuits according to timer signals and wherein at least two have distinct timers and one of the plurality of electronic circuits is the master device having the master timer and one of the plurality of electronic circuits is the slave device having the slave timer;

synchronizing the master timer and the slave timer, as part of at least one of the plurality of synchronization events, using the obtained timing or synchronization information;

applying a one-time or periodically repeated learning sequence to derive static or varying delays in the system;

correcting for such static or varying delays in the system;

applying a learning sequence to learn master-slave frequency drift; and

applying learned frequency drift to static or dynamic adjustment at an adjustment time that is distinct from each of the plurality of synchronization event times, thus taking into account estimated frequency drifts for a time period between a time of one of the synchronization events and the adjustment time, wherein the learned frequency drift is determined without communication across the separation.

24. The method of claim 23, wherein the learning sequence comprises:

scheduling multiple synchronization events;

changing time spacing between synchronization events;

making error measurements; and

applying additional processing to derive the master-slave frequency drift and static or varying delays in the system from the error measurements.

25. The method of claim 23, further comprising:

scheduling a wake-up event at the slave device taking into account the static or dynamic adjustment such that the timing of the wake-up event depends on the static or dynamic adjustment, wherein the wake-up event is an event that triggers a transition to a signal-listening mode from a not-listening mode.

26. A method of applying a learning sequence to quantify and adjust for static or varying delays and frequency drift in a communication system comprising at least a slave timebase device and a master timebase device, wherein slave timebase devices are configured to adjust their local timebases to coincide with the master timebase device, comprising:

obtaining timing/synchronization information during one or more events, the events being communication and/or synchronization events involving communication between the master timebase device and the slave timebase device in order to synchronize a master timer and a slave timer, wherein a plurality of electronic circuits each has at least one component used to output a signal that represents a timer, wherein which events occur in each of the plurality of electronic circuits according to timer signals and wherein at least two have distinct timers and one of the plurality of electronic circuits is the master timebase device having the master timer and one of the plurality of electronic circuits is the slave timebase device having the slave timer;

synchronizing the master timer and the slave timer as part of at least one of the communication and/or synchronization events, using the obtained timing/synchronization information;

analyzing the timing/synchronization information to derive timing errors due to system nonidealities; and

applying timing correction based on timing errors derived by the analyzing, wherein timing correction is applied at a correction time that is distinct from any one of the times of the one or more events, with the timing correction taking into account drifts or offsets for the nonzero time period between an event time and a correction time, without a communication between the master timebase device and the slave timebase device occurring at the correction time.

27. The method of claim 26, further comprising dynamically reducing a timebase error following a slave device wakeup, wherein the reduction occurs over time as timing error information becomes available.

28. The method of claim 26, wherein the timing/synchronization information is obtained continuously, or on a regular basis, and correction for timing errors is adjusted continuously or on the regular basis to reduce a margin of error.

29. The method of claim 26, further comprising:

scheduling a wake-up event at the slave device taking into account the timing correction such that the timing of the wake-up event depends on the timing correction, wherein the wake-up event is an event that triggers a transition to a signal-listening mode from a not-listening mode in which the timing/synchronization information is unavailable.

30. The method of claim 26, wherein the communication and/or synchronization events are events that are part of a communication sequence that is usable for other than synchronization.