The performance of maximum likelihood over-the-horizon radar coordinate registration

Abstract

A well-known source of target localization errors in over-the-horizon radar is the uncertainty about downrange ionospheric conditions. Maximum likelihood (ML) coordinate registration, using statistical modeling of ionospheric parameters, has recently been proposed as a method which is robust to ionospheric variability. This paper reports ML performance results for real data from a known target using estimates of ionospheric statistics derived from ionosonde measurements. Bootstrap samples derived from these statistics are then used in a hidden Markov model approximation to the ground range likelihood function. Comparison of the ML and conventional methods for over 250 radar dwells indicates the new technique achieves better than a factor of two improvement in ground range accuracy.