A calibration of timing accuracy in the NIST cyber-physical systems testbed

Abstract

We propose a general methodology for assessing the time accuracy and uncertainties, and report results from a project to calibrate timing in the NIST Cyber-Physical System (CPS) and Smart Grid Testbeds. We measured clock synchronization accuracy and stability as well as latencies for potential experiments in the testbeds. We determined calibrations of GPS receivers to UTC(NIST) with an uncertainty of 16 ns. However, an anomaly occurred coincident with a power shutdown, which resulted in a total uncertainty of receiver calibrations against UTC(NIST) of 100 ns. Synchronization at testbed locations relative to an IEEE 1588 Precision Time Protocol (PTP) grandmaster was found to have a max offset of 36 ns ± 6 ns one sigma from the grandmaster through two transparent clocks (TCs). Finally, we measured the time error relative to the grandmaster of an embedded device attached to a switch without PTP support with a mean offset of 50 µs ± 10 µs, and at 8 ms ± 500 µs for timestamping at the general-purpose input/output (GPIO). We report the methodology used, as well as some of the challenges encountered and solutions developed in the process.