Extraction of stability statistics from integrated rate data

A technique is described for estimating the values of white noise variance, per-unit-time random walk variance, and deterministic ramp which comprise a noise process of interest. Available measurements are assumed to be noisy observations of the time integral of the process itself. The measurement noise is assumed to be white and its variance is also estimated. Such measurements are taken from pulse-torqued inertial instruments. Derived equations provide the basis for a nonlinear estimation algorithm, coded in the program AUTOFIT. An example of AUTOFIT results based on state-of-the-art accelerometer data is included.