Probabilistic Automata for Safety LTL Specifications

  • Dileep Kini
  • Mahesh Viswanathan
Conference paper

DOI: 10.1007/978-3-642-54013-4_7

Part of the Lecture Notes in Computer Science book series (LNCS, volume 8318)
Cite this paper as:
Kini D., Viswanathan M. (2014) Probabilistic Automata for Safety LTL Specifications. In: McMillan K.L., Rival X. (eds) Verification, Model Checking, and Abstract Interpretation. VMCAI 2014. Lecture Notes in Computer Science, vol 8318. Springer, Berlin, Heidelberg

Abstract

Automata constructions for logical properties play an important role in the formal analysis of the system both statically and dynamically. In this paper, we present constructions of finite-state probabilistic monitors (FPM) for safety properties expressed in LTL. FPMs are probabilistic automata on infinite words that have a special, absorbing reject state, and given a cut-point λ ∈ [0,1], accept all words whose probability of reaching the reject state is at most 1 − λ. We consider Safe-LTL, the collection of LTL formulas built using conjunction, disjunction, next, and release operators, and show that (a) for any formula ϕ, there is an FPM with cut-point 1 of exponential size that recognizes the models of ϕ, and (b) there is a family of Safe-LTL formulas, such that the smallest FPM with cut-point 0 for this family is of doubly exponential size. Next, we consider the fragment LTL(G) of Safe-LTL wherein always operator is used instead of release operator and show that for any formula ϕ ∈ LTL(G) (c) there is an FPM with cut-point 0 of exponential size for ϕ, and (d) there is a robust FPM of exponential size for ϕ, where a robust FPM is one in which the acceptance probability of any word is bounded away from the cut-point. We also show that these constructions are optimal.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • Dileep Kini
    • 1
  • Mahesh Viswanathan
    • 1
  1. 1.Department of Computer ScienceUniversity of Illinois at Urbana-ChampaignUSA

Personalised recommendations