Computationally simple adaptive tracking filters based on weighted averages

A new algorithm for state estimation, based on certain fading-memory weighted averages, is introduced. The state estimates are obtained through linear combinations of the weighted averages. In their basic general forms, the second-order and third-order Weighted Average Filters are equivalent to over-damped alpha-beta and alha-beta-gamma filters, respectively. Different linear combinations of the weighted averages give state estimates with different band-width characteristics simultaneously. This property leads to the decision-directed, multi-level, switching type adaptive filter described. In the swinging type adaptive filter described later, simple maneuver dependent modifications to the weighted averages make the filter continually self-adaptive to changing situations, with optimized parameters. The computational simplicity of this type is very attractive for tracking maneuvering targets in a real-time, multi-target environment.