Time recovery in a time sensitive network
Time recovery techniques are described. A method comprises receiving messages from the first device by the second device in the first network domain, the messages to comprise time information to synchronize a first clock for the first device and a second clock for the second device to a network time, determining the second clock is to recover the network time for the second device without new messages from the first device, retrieving a first set of timestamps previously stored for events in the first network domain using the network time from the second clock, retrieving a second set of timestamps previously stored for the events in the first network domain using a redundant time from a third clock, where the third clock is not synchronized with the first and second clocks, and recovering the network time using a regression model and the redundant time from the third clock.
1. An apparatus, comprising:
a processing circuitry;
a memory coupled to the processing circuitry, the memory to store instructions that when executed by the processing circuitry causes the processing circuitry to:
establish a data stream between a first device and a second device in a first network domain, the data stream comprising a plurality of switching nodes;
receive messages from the first device by the second device in the first network domain, the messages to comprise time information to synchronize a first clock for the first device and a second clock for the second device to a network time for the first network domain;
determine the second clock is to recover the network time for the second device without new messages from the first device;
retrieve a first set of timestamps previously stored for events in the first network domain using the network time from the second clock;
retrieve a second set of timestamps previously stored for the events in the first network domain using a redundant time from a third clock, wherein the third clock is not synchronized with the first and second clocks;
construct a regression model based on the first and second set of timestamps; and
recover the network time of the first network domain using a redundant time from the third clock as input for the regression model to output a recovered network time for the second clock of the second device.
2. The apparatus of claim 1 , the processing circuitry to establish the data stream in accordance with the Institute of Electrical and Electronics Engineers (IEEE) 802.1AS and/or 802.1Qbv and/or 1588 standards.
3. The apparatus of claim 1 , wherein the messages are synchronization messages or follow up messages for a precision time protocol (PTP).
4. The apparatus of claim 1 , wherein the first device operates in a clock leader (CL) role and the second device operates in a clock follower (CF) role.
5. The apparatus of claim 1 , the processing circuitry to determine the second clock is to recover the network time for the second device without new messages from the first device when the messages are not received by the second device for a defined time interval.
6. The apparatus of claim 1 , the processing circuitry to store the first set of timestamps for events in the first network domain based on the network time from the second clock by the second device.
7. The apparatus of claim 1 , the processing circuitry to recover the network time of the first network domain using the redundant time from the third clock as input for the regression model to output the recovered network time for the second clock of the second device, the processing circuitry to: receive a current redundant time from the third clock in the second network domain not synchronized with the first and second clocks; calculate a recovered network time with the regression model; synchronize the second clock of the second device to the recovered network time.
8. The apparatus of claim 1 , the processing circuitry to send messages from the second device to a fourth device in the first network domain, the messages to comprise time information to synchronize the second clock for the second device and a fourth clock for the fourth device to the recovered network time, the second device to operate in the CL role and the fourth device to operate in the CF role.
9. A computing-implemented method, comprising:
establishing a data stream between a first device and a second device in a first network domain, the data stream comprising a plurality of switching nodes;
receiving messages from the first device by the second device in the first network domain, the messages to comprise time information to synchronize a first clock for the first device and a second clock for the second device to a network time for the first network domain;
determining the second clock is to recover the network time for the second device without new messages from the first device;
retrieving a first set of timestamps previously stored for events in the first network domain using the network time from the second clock;
retrieving a second set of timestamps previously stored for the events in the first network domain using a redundant time from a third clock, wherein the third clock is not synchronized with the first and second clocks;
constructing a regression model based on the first and second set of timestamps; and
recovering the network time of the first network domain using a redundant time from the third clock as input for the regression model to output a recovered network time for the second clock of the second device.
10. The computing-implemented method of claim 9 , wherein the first device operates in a clock leader (CL) role and the second device operates in a clock follower (CF) role.
11. The computing-implemented method of claim 9 , determining the second clock is to recover the network time for the second device without new messages from the first device in response to a message indicating the first device is under a security attack and placed in isolation from the first network domain.
12. The computing-implemented method of claim 9 , wherein the redundant time from the third clock maintains a defined relationship with the network time from the second clock, the defined relationship to comprise an approximate linear relationship.
13. The computing-implemented method of claim 9 , comprising storing the second set of timestamps for the events in the first network domain based on the redundant time from the third clock by a third device in a second network domain separate from the first network domain.
14. The computing-implemented method of claim 9 , comprising recovering the network time of the first network domain using the redundant time from the third clock as input for the regression model to output the recovered network time for the second clock of the second device by: receiving a current redundant time from the third clock in the second network domain not synchronized with the first and second clocks; calculating a recovered network time with the regression model; synchronizing the second clock of the second device to the recovered network time.
15. The computing-implemented method of claim 9 , comprising switching the second device from operating in the CF role to operating in the CL role for the first network domain.
16. The computing-implemented method of claim 9 , comprising sending messages from the second device to a fourth device in the first network domain, the messages to comprise time information to synchronize the second clock for the second device and a fourth clock for the fourth device to the recovered network time, the second device to operate in the CL role and the fourth device to operate in the CF role.
17. A non-transitory computer-readable storage device, storing instructions that when executed by processing circuitry of a manager of a time sensitive network (TSN), cause the manager to:
receive messages from the first device by the second device in the first network domain, the messages to comprise time information to synchronize a first clock for the first device and a second clock for the second device to a network time for the first network domain;
determine the second clock is to recover the network time for the second device without new messages from the first device;
retrieve a first set of timestamps previously stored for events in the first network domain using the network time from the second clock;
retrieve a second set of timestamps previously stored for the events in the first network domain using a redundant time from a third clock, wherein the third clock is not synchronized with the first and second clocks;
construct a regression model based on the first and second set of timestamps; and
recover the network time of the first network domain using a redundant time from the third: clock as input for the regression model to output a recovered network time for the second clock of the second device.
18. The computer-readable storage medium of claim 17 , the instructions, when executed by the processing circuitry, cause the manager to establish the data stream in accordance with the Institute of Electrical and Electronics Engineers (IEEE) 802.1AS and/or 802.1Qbv and/or 1588 standards.
19. The computer-readable storage medium of claim 17 , wherein the messages are synchronization messages or follow up messages for a precision time protocol (PTP).
20. The computer-readable storage medium of claim 17 , wherein the first device operates in a clock leader (CL) role and the second device operates in a clock follower (CF) role.
21. The computer-readable storage medium of claim 17 , the instructions, when executed by the processing circuitry, cause the manager to determine the second clock is to recover the network time for the second device without new messages from the first device when the messages are not received by the second device for a defined time interval.
22. The computer-readable storage medium of claim 17 , the instructions, when executed by the processing circuitry, cause the manager to store the first set of timestamps for events in the first network domain based on the network time from the second clock by the second device.
23. The computer-readable storage medium of claim 17 , wherein the redundant time from the third clock maintains a defined relationship with the network time from the second clock, the defined relationship to comprise an approximate linear relationship.
24. The computer-readable storage medium of claim 17 , the instructions, when executed by the processing circuitry, cause the manager to store the second set of timestamps for the events in the first network domain based on the redundant time from the third clock of the second device.