Overlapping Variance

Sample Density

Data reuse in the calculation allows for a higher number of degrees of freedom in the resulting estimate. When computing overlapping variance, the windows of observation are shifted by a single time increment rather than by the full width of the window. This approach reduces the uncertainty of the measurement without requiring a longer total duration of data collection.

Monitoring the overlap ensures that no phase transitions are missed during long term testing. It is particularly useful for detecting low frequency drift in a stable reference signal.

Noise Identification

Different noise processes leave distinct signatures on the plot of stability versus time. Analysts use overlapping variance to distinguish between white frequency noise or random walk frequency noise. The slope of the resulting curve indicates the dominant error source in the timing system.

This information allows engineers to compensate for environmental interferences such as temperature or vibration.

Confidence Level

Error bars for the stability measurement depend on the number of independent samples captured during the test. Because overlapping variance uses correlated data, the calculation of the confidence interval requires a complex mathematical approach. High quality timing equipment undergoes this analysis during the final calibration to verify that it meets the specification for long term jitter.

This ensures that the clock remains synchronized.