PRIOR INFORMATION AND BIAS IN SEQUENTIAL ESTIMATION,
RAND CORP SANTA MONICA CALIF
Pagination or Media Count:
This article applies discrete sequential filtering to estimation of an unknown vector x imbedded in nonstationary uncorrelated noise, when observations depend linearly on x in a time-varying manner. Such a situation occurs in trajectory determination close to the earth. We solve the recursive filter equations to obtain the n-th estimate of x, xn, and its convariance matrix Pn, in terms of the observations zi, i1, ..., n and initial values xo, Po. These expressions illustrate how the a priori information can introduce bias into the sequential estimates.
- Statistics and Probability