HASL: an expressive language for statistical verification of stochastic models

  • Authors:
  • Paolo Ballarini;Hilal Djafri;Marie Duflot;Serge Haddad;Nihal Pekergin

  • Affiliations:
  • LACL, Univ. Paris-Est Créteil, France;LSV, ENS-Cachan, France;LACL, Univ. Paris-Est Créteil, France;LSV, ENS-Cachan, France;LACL, Univ. Paris-Est Créteil

  • Venue:
  • Proceedings of the 5th International ICST Conference on Performance Evaluation Methodologies and Tools
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

We introduce the Hybrid Automata Stochastic Logic (HASL), a new temporal logic formalism for the verification of discrete event stochastic processes (DESP). HASL employs Linear Hybrid Automata (LHA) as machineries to select prefixes of relevant execution paths of a DESP. The advantage with LHA is that rather elaborate information can be collected on-the-fly during path selection, providing the user with a powerful means to express sophisticated measures. A formula of HASL consists of an LHA and an expression Z referring to moments of path random variables. A simulation-based statistical engine is employed to obtain a confidence-interval estimate of the expected value of Z. In essence HASL provides a unifying verification framework where temporal reasoning is naturally blended with elaborate reward-based analysis. We illustrate the HASL approach by means of some examples and a discussion about its expressivity. We also provide empirical evidence obtained through Cosmos, a prototype software tool for HASL verification.