Let a sensor (e.g. accelerometer) with a combination of white gaussian noise and random walk noise. The loglog plot that is usually shown as function of a time parameter, affine decreasing for a white gaussian noise and affine increasing for a random walk. What does it mean? Can we read from the graph the variance of the gaussian noise? I cannot really grasp the intuition behind it: how does the Allan variance help to get more information about the noise than a PSD or some cross-correlation?


Browse other questions tagged or ask your own question.