Universality Analysis for One-Clock Timed Automata

  • Authors:
  • Parosh Aziz Abdulla;Johann Deneux;Joë/l Ouaknine;Karin Quaas;James Worrell

  • Affiliations:
  • (Correspd.) Uppsala University, Uppsala, Sweden. parosh@it.uu.se/ Johann.Deneux@it.uu.se;Uppsala University, Uppsala, Sweden. parosh@it.uu.se/ Johann.Deneux@it.uu.se;Oxford University, Wellington Square Oxford, United Kingdom. joel@comlab.ox.ac.uk;Leipzig University, Leipzig, Germany. quaas@informatik.uni-leipzig.de;Oxford University, Wellington Square Oxford, United Kingdom. jbw@comlab.ox.ac.uk

  • Venue:
  • Fundamenta Informaticae - Fundamentals of Software Engineering 2007: Selected Contributions
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper is concerned with the universality problem for timed automata: given a timed automaton A, does A accept all timed words? Alur and Dill have shown that the universality problem is undecidable if A has two clocks, but they left open the status of the problem when A has a single clock. In this paper we close this gap for timed automata over infinite words by showing that the one-clock universality problem is undecidable. For timed automata over finite words we show that the one-clock universality problem is decidable with non-primitive recursive complexity. This reveals a surprising divergence between the theory of timed automata over finite words and over infinite words. We also show that if ε-transitions or non-singular postconditions are allowed, then the one-clock universality problem is undecidable over both finite and infinite words. Furthermore, we present a zone-based algorithm for solving the universality problem for single-clock timed automata. We apply the theory of better quasi-orderings, a refinement of the theory of well quasi-orderings, to prove termination of the algorithm. We have implemented a prototype tool based on our method, and checked universality for a number of timed automata. Comparisons with a region-based prototype tool confirm that zones are a more succinct representation, and hence allow a much more efficient implementation of the universality algorithm.