在trimble手册中找到了一些：
5.1.2 kalman filtering
kalman filtering is a technique that improves the performance of a gps disciplined clock when gps drops out. this state is called holdover. during holdover the clock relies solely on the oscillator. oscillator performance is subject to two basic effects. first, changes in environmental temperature can cause the oscillator to speed up to slow down. second, the oscillator has a natural tendency to drift over time. this is called aging. both temperature and aging can be mathematically predicted. however, the characteristics vary from crystal to crystal. the kalman filtering monitors the unique oscillator performance over time and temperature and records this behavior. then when the clock goes into holdover this filtering corrects for these effects producing a more accurate clock. the longer a clock has to ’train’ the better the kalman filtering performance will be. 24 hours is considered the minimum necessary for good performance
kalman filter
a numerical method used to track a time-varying signal in the presence of noise. if the signal can be characterized by some number of parameters that vary slowly with time, then kalman filtering can be used to tell how incoming raw measurements should be processed to best estimate those parameters as a function of time.
