Wrong handling of indexes when constructing the state transition matrix for multi satellites obit determination based on Kalman Filter
I performed some tests of multi satellites orbit determinations using the Kalman Filter.
I considered 3 simulations:
- 2 by estimating independently 2 satellite orbits
- 1 by estimating simultaneously the two previous orbits
For the multi satellites case, there is no relation between the estimated parameters of the first and the second satellite. Therefore, I expect the final result to be the combination of the two "independent" simulations. However, it is not. The accuracy is considerably degraded.
After investigation, I found the errors are related to the construction of the state transition matrix for multi-satellite applications. For instance, the initial state transition matrix is not equal to the identity matrix.
Edited by Bryan Cazabonne