Estimation of Vector Field of Systematic Errors of Radars Based on Multi-Tracking Data

Abstract

The problem of identification of systematic errors of several radars based on multi-
tracking data from moving objects (aircrafts) is considered. In case the model of spatial
dependence of systematic errors is not completely known the identification leads to an ill-
posed estimation problem. The author suggests an approach which provides a good estimate
under these conditions. The basis of the approach is the local approximation of the unknown
systematic errors as a function of geometric position. Position space is divided into the
system of sufficiently small parts. In each part the vector of the local value of the systematic
errors is estimated. Due to the ill-posedness of the problem only an uncertainty set can be
identified; this set contains all possible vectors that can provide identical measurements.
These uncertainty sets can be considered as a multivalued function of geometric position.
Then the selection of a single-valued function of systematic errors out of a multivalued
function is done on the basis of criterion the minimization of which enables to define the
most "flat" function. The algorithm was tested on real trajectory tracking data.