Performability Measure Specification: Combining CSRL and MSL

An integral part of the performance modeling process is the specification of the performability measures of interest. The notations proposed for this purpose can be grouped into classes that differ from each other in their expressiveness and usability. Two representative notations are the continuous stochastic reward logic CSRL and the measure specification language MSL. The former is a stochastic temporal logic formulating quantitative properties about states and paths, while the latter is a component-oriented specification language relying on a first-order logic for defining reward-based measures. In this paper, we combine CSRL and MSL in order to take advantage of the expressiveness of the former and the usability of the latter. To this aim, we develop a unified notation in which the core logic of MSL is employed to set up the reward structures needed in CSRL, whereas the measure definition mechanism of MSL is exploited to formalize measure and property specification patterns in a component-oriented fashion.