Verifying Probabilistic Timed Automata Against Omega-Regular Dense-Time Properties

Probabilistic timed automata (PTAs) are timed automata (TAs) extended with discrete probability distributions.They serve as a mathematical model for a wide range of applications that involve both stochastic and timed behaviours. In this work, we consider the problem of model-checking linear \emph{dense-time} properties over {PTAs}. In particular, we study linear dense-time properties that can be encoded by TAs with infinite acceptance criterion.First, we show that the problem of model-checking PTAs against deterministic-TA specifications can be solved through a product construction. Based on the product construction, we prove that the computational complexity of the problem with deterministic-TA specifications is EXPTIME-complete. Then we show that when relaxed to general (nondeterministic) TAs, the model-checking problem becomes undecidable.Our results substantially extend state of the art with both the dense-time feature and the nondeterminism in TAs.

💡 Research Summary

This paper investigates the model‑checking problem for Probabilistic Timed Automata (PTA) against linear dense‑time specifications that are expressed by Timed Automata (TA) equipped with an infinite acceptance condition, specifically Rabin acceptance. The authors distinguish two classes of specifications: deterministic timed automata (DTA) and general nondeterministic timed automata (NTA).

For deterministic specifications, they construct a product automaton that combines a PTA with a deterministic timed Rabin automaton (DTRA). The product’s state space consists of a PTA location together with its clock valuation and a DTRA mode together with its clock valuation. Time‑delay transitions are synchronized so that both components evolve under the same elapsed time, while discrete actions are synchronized via the labeling function of the PTA and the input alphabet of the DTRA. The Rabin condition of the DTRA is lifted to the product, turning the product into a Markov Decision Process (MDP) with an ω‑regular acceptance condition. Standard MDP analysis techniques (value iteration, strategy synthesis) can then compute the minimum and maximum probabilities that a run of the PTA satisfies the DTRA specification.

A detailed complexity analysis shows that the product construction incurs an exponential blow‑up in the number of clocks, but the overall model‑checking problem lies in EXPTIME. Moreover, the authors prove EXPTIME‑hardness by reduction from known hard problems for timed automata, establishing EXPTIME‑completeness for the deterministic case.

When the specification automaton is allowed to be nondeterministic, the product no longer yields a pure MDP; instead, it becomes a stochastic game where the nondeterminism of the TA introduces an adversarial player. By encoding the halting problem of a Turing machine into such a nondeterministic TA, the authors demonstrate that the model‑checking problem becomes undecidable. Consequently, there is no algorithm that can compute the extremal probabilities for arbitrary TA specifications.

The paper includes a concrete example: a simple task‑processing PTA that models a machine handling two job types (α and β) with bounded processing times and failure probabilities. A DTRA specification captures a property such as “the machine eventually processes an infinite sequence of jobs without exceeding given time bounds.” The product construction is applied, and the extremal probabilities are computed, illustrating the practical applicability of the theoretical results.

In summary, the contributions are: (1) a product‑based framework for verifying PTAs against dense‑time ω‑regular properties expressed by deterministic timed automata; (2) an EXPTIME‑complete complexity result for the deterministic case; (3) a proof of undecidability for the general nondeterministic case; and (4) an illustrative example that bridges theory and practice. The work extends the state of the art by handling continuous‑time linear properties and clarifying the limits of decidability when nondeterminism is present in the specification automaton. Future directions suggested include extending the approach to priced PTAs, incorporating reward structures, and developing tool support for large‑scale case studies.