The thesis proposes to combine the IEEE 1588 protocol and the adaptive
oscillator correction method (AOCM) to improve master-slave synchronization.
This is done by extending the AOCM to a slave clock to train a model using the
IEEE 1588 synchronization updates from a master clock. The model corrections
are applied to the slave clock during the waiting period in between the
synchronization updates and when the slave-master network experiences an
outage. The NS-2 results indicate that the slave accuracy improves up to 10
times for no traffic networks depending on the slave clock training period.
In traffic scenarios, the slave accuracy is affected by the asymmetric
delays such that the degree of accuracy is one half of
the asymmetric latencies. During network congestion, the slave accuracy
improves because the delays are less asymmetric. During an outage in a traffic
scenario, the solution only improves the slave clock stability.