Arrow Research search
Back to AAMAS

AAMAS 2026

A Verification Framework for Obstruction, Probability, and Time

Conference Paper Research Paper Track Autonomous Agents and Multiagent Systems

Abstract

Verifying strategic behaviour in real-time multi-agent systems under uncertainty is vital for safety- and security-critical domains. Existing obstruction logics treat either adversarial timing (TOL) or probabilistic risk (POTL), but real scenarios require both. We introduceProbabilisticTimedObstructionTemporalLogic (PTOTL), which unifies dense time, probabilities, and cost-bounded obstruction for real-time security games. Interpreted over Weighted Probabilistic Timed Automaton (WPTA), PTOTL models attacker–defender interactions where discrete actions evolve and time elapses, and the defender may disable transitions under a per-step budget. We give syntax and semantics and a symbolic model-checking procedure on a probabilistic zone graph. Despite the added strategic and probabilistic features, verification remains PSPACE, not higher than PTCTL or PTATL, while offering greater temporal expressiveness. An automotive Moving Target Defense (MTD) case study demonstrates practicality as a specification and verification language.

Authors

Keywords

  • Strategic reasoning
  • Cybersecurity
  • Model checking
  • Quantitative verification
  • Security games
  • Moving Target Defense mechanisms

Context

Venue
International Conference on Autonomous Agents and Multiagent Systems
Archive span
2002-2026
Indexed papers
8043
Paper id
328612415857305664
v2026.09.13