Critical states detection with bounded probability of false alarm and application to air traffic management

The analysis of error propagation in an Air Tra±c Management (ATM) environment is addressed. The theory of Hybrid Systems is used to model the error evolution, an observability problem for a Markov Chain with discrete output symbols associated to the transitions is stated, and a runtime observer is proposed for estimating the probability of a given discrete state to be active. Sufficient conditions are given for characterizing the decidability of the addressed observability problem. The results are related to previous works on location observability of deterministic hybrid systems, and are used to analyze an ATM case study, the "clearance to change the flight plan". Copyright © 2006 IFAC.