Adaptive filtering of imprecisely time-stamped measurements with application to AIS networks

Driven by real-world issues in maritime surveillance, we consider the problem of estimating the target state from a sequence of observations that can be imprecisely time-stamped. That is, the time between two consecutive observations can be affected by an unknown error or delay. We propose an adaptive filtering strategy able to sequentially detect the time delays and correctly estimate the target state. Two decision statistics for the presence of delay are derived, the first is non-parametric while the second is based on the Generalized Likelihood Ratio Test (GLRT). When a delayed measurement is detected, the Maximum Likelihood (ML) estimate of the delay can be used to correct the timestamps of the target observation used in the filter. The validation of the proposed method is carried out using Monte Carlo computer simulations and analyzing real-world data collected by a global network of Automatic Identification System (AIS) receivers.