Predictability of Event Occurrences in Timed Systems

We address the problem of predicting events’ occurrences in partially observable timed systems modelled by timed automata. Our contribution is manyfold: 1) we give a definition of bounded predictability, namely k-predictability, that takes into account the minimum delay between the prediction and the actual event’s occurrence; 2) we show that 0-predictability is equivalent to the original notion of predictability of S. Genc and S. Lafortune; 3) we provide a necessary and sufficient condition for k-predictability (which is very similar to k-diagnosability) and give a simple algorithm to check k-predictability; 4) we address the problem of predictability of events’ occurrences in timed automata and show that the problem is PSPACE-complete.