Efficient Temporal Logic Verification by Incremental Abstraction

Abstract

A recently formulated incremental abstraction method that preserves $\mathrm{CTL}^{*}$-X is further developed and evaluated in this paper. The temporal logic $\mathrm{CTL}^{*}$-X includes both linear temporal logic (LTL) and computational tree logic (CTL), except for the next operator (often denoted by X). For a modular transition system, the incremental abstraction means that any local events, only included in one of the subsystems, are first abstracted. After synchronization more events become local. Repeated abstraction then means that further reduction of local transitions is achieved when more submodels are synchronized. This incremental abstraction is evaluated as a tool for temporal logic verification, and compared with a couple of competitive alternatives. In a preliminary evaluation, the proposed incremental abstraction is often shown to be the most efficient choice for verification of temporal logic properties, especially when some individual submodels are comparatively large.