An Expressive Language for Statistical Verification of Stochastic Models

old_uid9467
titleAn Expressive Language for Statistical Verification of Stochastic Models
start_date2010/12/17
schedule15h-15h30
onlineno
location_infobatiment d’Alembert, Amphithéâtre Fonteneau
detailsjoint work with P. Ballarini, M. Duflot, S. Haddad and N. Pekergin
summaryBeing inspired by existing stochastic logic, such as, the Continuous Stochastic Logic CSL and its following action-state asCSL and timed automata CSLTA evolutions, SHAL extends them in two respects: firstly it targets a much broader class of stochastic models (i.e. DESP rather than Markov chains); secondly it increases the expressiveness of verification by allowing to capture sophisticated dynamics expressed in terms of multivariate conditions. A formula of SHAL consists of a Linear Hybrid Automaton (LHA) and an expression. The LHA selects prefixes of relevant execution paths of a DESP and collects information along the paths. The result of the formula verification is the evaluation of an expression of the considered path random variable. A statistical engine is then employed for the confidence-interval estimation of the expected value of the expression associated to an LHA. We illustrate the SHAL approach by means of some examples and experimental results obtained through a prototype software tool for SHAL verification.
responsiblesBaerecke