Combination of adaptive filters for relative navigation

Relative navigation of spacecrafts may be accomplished via time delay estimates. In this work an adaptive filtering approach is employed, which involves an estimation and a detection step. By formally posing the detection problem, a more meaningful detector, that embeds a reliability measure into the delay estimates, is proposed. The estimation step is enhanced via convex combination schemes, that address the Poisson distributed signals, sparse channel and low signal-to-noise ratio. To evaluate time delay estimation techniques, different criteria based on probability of detection are studied, leading to a new figure of merit. The resulting solution outperforms the existing adaptive filters techniques under the new criterion, as shown by simulations.