Filtering and Implementation for Air-to-Air Target Tracking

This paper discusses some aspects of the design problem involved in the choice of a realization or approximation of a desired system behavior (as for instance dictated by the analytical solutions to a filtering problem) by parameters that can only be approximately adjusted, e.g., due to quantization, component tolerances (analog case) and finite wordlength (discrete case). The paper first addresses the mathematical characterization of this robustness problem, and its solutions under various criteria of optimality. Earlier results are here extended to multi-mode systems which can arise in non-linear approximation problems. The feasibility of this approach in multi-mode filtering is shown, and is illustrated by an air-to-air tracking example.