Particle estimation, basic theory: an introduction
Publish date: 2002-01-01
Report number: FOI-R--0473--SE
Written in: English
In the project "Jamming of Navigation Systems" different combinations of inertial and GPS-based navigation systems are investigated. A topic of special interest in this connection is the so called tightly coupled integration of the two types of systems. One reason for this is its importance as a prerequisite for the introduction of electrically steerable antennas, by means of which external disturbances can be efficiently suppressed, thereby increasing the robustness of the navigation system. However, the tightly coupled integration gives rise to nonlinear estimation problems. For this reason a Monte Carlo-based estimator called "Particle Filter" has been investigated in an earlier report and found to give a better performance in certain respects. It does not, as opposed to the Kalman estimator, require that the system noise is gaussian in order to work optimally. This report consists of a derivation of the Particle Filter algoritm.