nyx-space
Support estimation of ground station biases
High level description
Ground stations have biases, which are currently modeled as first order Gauss Markov processes. With a reasonably high bias, the OD results are quite noisy because we don't estimate that bias.
Requirements
- GM process should include a constant offset bias
- GM process constant bias should have a frequency mismatch initialization for Doppler, and somehow for range measurements too.
- Each GM of each Ground Station should be estimated independently.
- Allow enabling/disabling of biases per measurement type and per station
Test plans
- Set a high constant offset and ensure that with enough tracking data, the bias is correctly estimated
- Ensure that if the bias estimation is turned on for one type and one station, it is not necessarily the case for the other stations.
Design
What is estimated must be part of the state and the Jacobian. It isn't immediately clear how this fits in the statically sized vectors and matrices of the current architecture, but that needs to be figured out.
This work will also likely lead to a separation of a Ground Station and whatever can generate OD measurements. Right now, the Ground Station structure includes more information than needed to define a ground point.
In this work, I should also add the turn around time on the spacecraft side, using a fixed nanosecond offset. This work will require extending the state used in the estimation.