# Measuring Uncertainty with Stochastic Probes

Jeremy Bradley Department of Computing Imperial College London 4pm 28 October 2003 Room 2511, JCMB, King's Buildings

Having constructed a process algebra model of a communicating system the next challenge is to ask useful questions of it. Usually one might use a logic to interrogate the model but unless the query is automatically generated this requires the user to learn another formalism.

Our interest is in stochastic process algebra models and in particular PEPA models. Here the questions that can be asked extend beyond the functional to ones involving time and probability as well. In some recent work, we've looked at using regular expressions to specify so-called stochastic probes, which take the form of PEPA components. These are then used to set up a variety of stochastic queries on a PEPA model, such as: will my system recover from a failure mode within 1.7 seconds with probability 0.9995 or better?

In the presentation, I'll look at the specification and construction of the stochastic probes, along with model/query analysis using the Imperial PEPA Compiler and the DNAmaca Markovian analyser.