No abstract

Many network applications like motion control or precise monitoring of machines need precise knowledge about the time synchronization accuracy. But time synchronization accuracy depends on the performance of PTP nodes, network topology, environmental conditions and various other factors. This makes the determination of synchronization accuracy a complex task. A mechanism for determining the worst case synchronization accuracy is defined in the PTP Power Profile. A TLV is used for accumulating a vendor defined worst case inaccuracy. However in practice when using this approach the inaccuracy values are often much higher than the real synchronization accuracy. In this paper, a technique for automatic determination of synchronization path quality is investigated. It utilizes PTP Bridges with an inaccuracy estimation performed using an inaccuracy model that is separated into static and dynamic inaccuracy contributors. One assumption is that the inaccuracy in the PTP time observed by a bridge, depends on the position of the bridge in the sync path. The latter is due to accumulation of more, possibly independent, random contributors. The effect of increasing deviation is modeled and verified using an experimental setup. The paper concludes that detailed knowledge about the PTP network (cable lengths, inaccuracy contribution metrics for the specific nodes within the sync path, etc.) are useful for automatic determination of the synchronization path quality, synchronization monitoring, system configuration and diagnosis. We suggest an enhancement to the TimeInaccuracy TLV that may possibly be incorporated within the next PTP revision, to better facilitate the protocol support for the above functions

No abstract