Controller Synthesis for Parametric Timed Games

📅 2025-06-18
📈 Citations: 0
Influential: 0
📄 PDF
🤖 AI Summary
This paper addresses the computational and deployability challenges of synthesizing winning strategies for parametric timed game systems. Methodologically, it introduces a symbolic strategy synthesis algorithm that—uniquely—constructs explicit winning strategies (rather than merely deriving parameter constraints), enabling direct generation of executable controllers and automated translation into parametric timed automata for practical deployment. The technical framework integrates parametric timed automaton modeling, game-theoretic semantics, and efficient symbolic solving. Evaluated on the Production Cell benchmark, the approach successfully automates controller synthesis and formal verification. It significantly enhances strategy interpretability, executability, and practical utility. By bridging high-level strategic reasoning with low-level implementation, this work establishes a novel paradigm for controller synthesis in parametric real-time systems.

Technology Category

Application Category

📝 Abstract
We present a (semi)-algorithm to compute winning strategies for parametric timed games. Previous algorithms only synthesized constraints on the clock parameters for which the game is winning. A new definition of (winning) strategies is proposed, and ways to compute them. A transformation of these strategies to (parametric) timed automata allows for building a controller enforcing them. The feasibility of the method is demonstrated by an implementation and experiments for the Production Cell case study.
Problem

Research questions and friction points this paper is trying to address.

Compute winning strategies for parametric timed games
Synthesize constraints on clock parameters for winning
Transform strategies to timed automata for controllers
Innovation

Methods, ideas, or system contributions that make the work stand out.

Semi-algorithm for parametric timed games
New winning strategies definition
Transform strategies to timed automata
🔎 Similar Papers
No similar papers found.
M
Mikael Bisgaard Dahlsen-Jensen
Aarhus University, Aarhus, Denmark
B
Baptiste Fievet
CNRS, Universit'e Sorbonne Paris Nord, LIPN, F-93430 Villetaneuse, France
Laure Petrucci
Laure Petrucci
LIPN, CNRS UMR 7030, Université Paris 13
Jaco van de Pol
Jaco van de Pol
Professor of Computer Science, Aarhus University
Formal Methodsverification and validationhigh-performance model checking