Compositional Construction of Infinite Abstractions for Networks of Stochastic Control Systems [article]

Abolfazl Lavaei, Sadegh Soudjani, Majid Zamani
2019 arXiv   pre-print
This paper is concerned with a compositional approach for constructing infinite abstractions of interconnected discrete-time stochastic control systems. The proposed approach uses the interconnection matrix and joint dissipativity-type properties of subsystems and their abstractions described by a new notion of so-called stochastic storage functions. The interconnected abstraction framework is based on new notions of so-called stochastic simulation functions, using which one can quantify the
more » ... tance between original interconnected stochastic control systems and interconnected abstractions in the probabilistic setting. In the first part of the paper, we derive dissipativity-type compositional reasoning for the quantification of the distance in probability between the interconnection of stochastic control subsystems and that of their abstractions. Moreover, we focus on a class of discrete-time nonlinear stochastic control systems with independent noises in the abstract and concrete subsystems, and propose a computational scheme to construct abstractions together with their corresponding stochastic storage functions. In the second part of the paper, we consider specifications expressed as syntactically co-safe linear temporal logic formulae and show how a synthesized policy for the abstract system can be refined to a policy for the original system while providing guarantee on the probability of satisfaction. We demonstrate the effectiveness of the proposed results by constructing an abstraction (totally 3 dimensions) of the interconnection of three discrete-time nonlinear stochastic control subsystems (together 222 dimensions) in a compositional fashion. We also employ the abstraction as a substitute to synthesize a controller enforcing a syntactically co-safe linear temporal logic specification.
arXiv:1801.10505v2 fatcat:o3pdlu4u2nbs5h3esfpgoyelb4