Metrology and 1/f noise: linear regressions and confidence intervals in flicker noise context

Abstract

1/f noise is very common but is difficult to handle in a metrological way. After having recalled the main characteristics of stongly correlated noise, this paper will determine relationships giving confidence intervals over the arithmetic mean and the linear drift parameters. A complete example of processing of an actual measurement sequence affected by 1/f noise will be given.