Specification formalisms for component-based concurrent systems

Abstract

This project builds on my ongoing research into design formalisms for, and the automatic verification of, concurrent systems. The difficulties such systems pose for system engineers are well-known and result in large part from the the complexities of process interaction and the possibilities for nondeterminism. My work is motivated by a belief that mathematically rigorous specification and verification techniques will ultimately lead to better and easier-to-build concurrent systems.My specific research interests lie in the development of fully automatic analysis methods and process-algebraic design formalisms for modeling system behavior. I have worked on algorithms for checking properties of, and refinement relations between, system descriptions [CH93, CS93]; the implementation and release of a verification tool, the CWB- NC [CS96] (see http://www.cs.sunysb.edu/~rance to obtain the distribution); case studies [BCL99, ECB97]; and the formalization of system features, such as real time, probability, and priority, in process algebra [BCL99, CDSYar].The aims of this project include the development of expressive and usable formalisms for specifying and reasoning about properties of open, component-based concurrent systems. More specifically, my colleagues and I have been investigating new approaches for describing component requirements and automated techniques for determining when finite-state components meet their requirements. The key topics under study include the following. A temporal logic for open systems. We are working on a notation for conveniently expressing properties constraining the behavior of open systems. Implicit specifications. Implicit specifications use system contexts, or "test harness," to define requirements for open systems. We are studying expressiveness issues and model-checking algorithms for such specifications. Automatic model-checker generation. We have been developing a model-checker generator that, given a temporal logic and "proof rules" for the logic, automatically produces an efficient model checker.