天基综合信息系统全国重点实验室鲁班青年论坛2026年度第3期

时间：2026年07月07日（周二）10:00-11:00

地点：中国科学院软件研究所5号楼4层第二会议室

主讲人：Julian Klein. Technical University of Berlin (TUB)

报告摘要：

Timed automata are a standard formalism for modeling real-time systems with timing-dependent behavior. In partially observable settings, an external observer sees only visible events and their timing information, giving rise to the current-state estimation problem: given an observation, determine the set of states in which the system may currently be. While exact current-state estimation is undecidable for dense-time timed automata, it becomes tractable for discrete-timed automata, where time is represented by discrete tick events. However, existing constructions scale exponentially with the largest timing constant M of the input system, since they require explicit enumeration of all timesteps up to M. To address this limitation, we have proposed threshold estimators, which store and propagate symbolic threshold timestamps instead of constructing all time-successor states explicitly. Recently, we have identified structural conditions under which threshold estimators scale independently of M. Instead, our construction scales only with the total number of timing constraints. This yields a more scalable framework for current-state estimation in discrete-timed automata with large timing constants and enables our method for real-world applications that satisfy our structural conditions.

报告人介绍：

Julian Klein is a PhD student at the Technical University of Berlin (TUB) in Germany in the Software and Embedded Systems Engineering group, lead by Prof. Dr. Sabine Glesner.