On the Accuracy of Pairwise Time Synchronization

Time synchronization is a fundamental problem in any distributed system. In particular, wireless sensor networks require scalable time synchronization for implementing distributed tasks on multiple sensor nodes. Different synchronization schemes and hardware systems are able to achieve widely different levels of synchronization accuracy. Since synchronization accuracy is very sensitive to random delays and clock rate fluctuations, we model the pairwise synchronization error as a function of random delays and clock drifts to quantitatively model synchronization performance as a function of the relevant parameters, such as the beacon interval, the number of beacons per synchronization period, and the measurement point. The synchronization error is analyzed for different state-of-the-art pairwise synchronization schemes, based on one-way sender-receiver, two-way sender-receiver, and receiver-receiver approaches, using application layer or medium access control layer timestamping. The models are validated through numerical and experimental results.