IMITATOR4AMAS: Strategy Synthesis for STCTL

IMITATOR4AMAS supports model checking and synthesis of memoryless imperfect information strategies for STCTL, interpreted over networks of parametric timed automata with asynchronous execution. While extending the verifier IMITATOR, IMITATOR4AMAS is the first tool for strategy synthesis in this setting. Our experimental results show a substantial speedup over previous approaches.

💡 Research Summary

The paper presents IMITATOR4AMAS, the first tool that can both model‑check and synthesize memoryless imperfect‑information strategies for Strategic Timed Computation Tree Logic (STCTL) over networks of parametric timed automata (PTA) with asynchronous execution. The authors build on the existing parametric model checker IMITATOR, extending its breadth‑first search (BFS) state‑space exploration to incorporate strategy information directly into each global state. A global state is represented as a tuple of automata locations, clock and parameter constraints, and the current (partial) strategy. During BFS, a successor is added only if its transition label is compatible with the already chosen strategy for the source state, or if no strategy has yet been fixed (in which case the transition label becomes part of the target’s strategy). This ensures that the explored state space consists solely of outcome paths consistent with the constructed strategy, and STCTL sub‑formulas inside strategic operators are evaluated on‑the‑fly.

The logic STCTL extends CTL by adding time intervals to temporal operators and a coalition operator ⟨⟨A⟩⟩γ, which states that a set of agents A has a strategy to enforce property γ within the given time bounds. The paper focuses on ir‑strategies (memoryless: decisions depend only on the current local state) under imperfect information (agents observe only a projection of the global state). This modeling choice reflects realistic scenarios in safety‑critical domains such as industrial automation, avionics, and distributed control, where agents have limited sensing and communication capabilities.

Implementation details: IMITATOR4AMAS is written in OCaml, reusing IMITATOR’s parametric verification engine. The only user input required is the PTA model and an STCTL property, optionally annotated with the “#synth” keyword to trigger strategy synthesis. The tool automatically performs timing‑parameter synthesis together with strategy synthesis, and partial results are sound even if the exploration does not terminate.

Experimental evaluation compares IMITATOR4AMAS against the earlier IMITATOR‑based approach (which encoded strategies as parameters) and against a Maude rewriting‑logic implementation. The benchmark is a voting scenario where each voter Vi must have a strategy to cast a vote for the first candidate within 8 time units, expressed as ⟨⟨Vi⟩⟩∃F