Compositional Approximately Bisimilar Abstractions of Stochastic Interconnected Systems

This paper addresses the challenge of constructing (alternating) bisimilar symbolic models for discrete-time, stochastic, interconnected systems. By leveraging a refined over-approximation of the stochastic reachable sets and the q-th moment input-to-state stability (delta-ISS-Mq) property, we establish an approximate bisimulation relation between stochastic systems and their deterministic (noise-free) counterparts. This approach significantly simplifies the analysis of the approximate (alternating) bisimulation between the global interconnected stochastic system and its symbolic model. We derive formal conditions under which local bisimulation relations are preserved when composing subsystems into a global interconnected system. The engineering relevance of the proposed compositional framework is demonstrated through temperature regulation case studies involving a fixed ten-room building and a circular cascade interconnection scaled from six to fourteen rooms. The results show that, beyond a small number of subsystems, the proposed method achieves substantially faster abstraction computation times than standard monolithic approaches, with the performance gap widening as the number of interconnected subsystems grows and the monolithic approach failing to converge beyond ten rooms.