Precision synchronization of computer network clocks
read more
Citations
Fine-grained network time synchronization using reference broadcasts
System and method for synchronizing operations among a plurality of independently clocked digital data processing devices
A media synchronization survey: reference model, specification, and case studies
Global clock synchronization in sensor networks
Global clock synchronization in sensor networks
References
User Datagram Protocol
Internet time synchronization: the network time protocol
Network Time Protocol (Version 3) Specification, Implementation and Analysis
Network Time Protocol (Version 3) Specification, Implementation
Fault-tolerant clock synchronization in distributed systems
Related Papers (5)
Frequently Asked Questions (15)
Q2. What future works have the authors mentioned in the paper "Precision synchronization of computer network clocks1,2,3" ?
In future work the authors plan to investigate methods to stabilize the local clock and to isolate the cause of the bias observed between two primary servers synchronized to the same PPS signal. Preliminary results suggest that residual frequency wander can be reduced about two orders of magnitude with this scheme.
Q3. What is the main source of timing errors in the quartz oscillators used in modern computers?
The wander component depends primarily on the ambient temperature and is the major source of timing errors in the quartz oscillators used in modern computers.
Q4. What is the offset between the local clock and the on-time epoch?
Since the timestamp is captured at the on-time transition, the seconds-fraction portion is the offset between the local clock and the on-time epoch less the UART delay.
Q5. What is the conclusion to be drawn about the integration interval?
The conclusion to be drawn is that adjusting the integration interval muchbelow or much above τ = 1000 s does not improve the oscillator stability.
Q6. How can a network be synchronized with a number of hosts?
It has been demonstrated in previous work cited that it is possible using NTP to synchronize a number of hosts on an Ethernet or a moderately loaded T1 network within a few tens of milliseconds with careful selection of timing sources and the configuration of the time servers on the network.
Q7. What is the place to capture a timetamp?
By far the best place to capture a serial-port timestamp is right in the kernel interrupt routine, but this generally requires intruding in the kernel code itself, which can be intricate and architecture dependent.
Q8. How can a kernel be modified to achieve submillisecond accuracy?
Most of the latency burden can be avoided without kernel modifications, but some workstations will require additional hardware or kernel software to achieve submillisecond accuracy.
Q9. What is the x and y axes of the characteristic shown in Figure 14 scale?
Analysis confirms the x and y axes of the characteristic shown in Figure 14 scale directly as τ, which means the timing errors will scale as well.
Q10. How fast is the new microtime() routine?
A replacement microtime() routine coded in assembler language is available in the NTP Version 3 distribution and is much faster at about 3 µs per call.
Q11. What is the way to know the maximum error of the reported time?
For some applications it is useful to know the maximum error of the reported time due to all causes, including clock reading errors, oscillator frequency errors and accumulated latencies on the path to a primary reference source.
Q12. What is the probability distribution for the gettimeofday() latency?
The graph shows the cumulative probability distribution for the gettimeofday() latency over one full day, from which a conclusion can be drawn that the probability of exceeding even a threshold as low as about 60 µs is about 0.5 percent, or about the probability of colliding with a timer interrupt on a random request.
Q13. What is the new microtime() routine for the OSF/1 kernel?
In both the Ultrix and OSF/1 kernels the gettimeofday() system call uses the new microtime() routine, which returns the actual interpolated value.
Q14. How can a NTP-synchronized host remove residual jitter?
It is possible for an NTP-synchronized host to derive the latter information using other NTP peers, presumably properly synchronized within ±0.5 second, and to remove residual jitter using the PPS signal.
Q15. What is the frequency error for the update vs(i)?
When the update vs(i) is received, the frequency error f(i + 1) and phase error g(i + 1) are computed:f(i + 1) = f(i) + µ(i)vs(i) τ2 , g(i + 1) = vs(i)τ