Online Model Checking for Dependable Real-Time Systems
Yuhong Zhao,Franz J. Rammig
DOI: https://doi.org/10.1109/isorc.2012.28
2012-01-01
Abstract:This paper presents a lightweight verification technique, which is applicable to dependable real-time systems, provided that the (abstract) model and the (concrete) implementation of the system under test are given in advance. In addition to the usual quality assurance techniques at design time (e.g., formal verification) and at implementation time (e.g., testing), we provide a special form of model checking at run time. That is, we check the correctness of an actual system execution by means of exploring a partial model space covering the current execution trace. In doing so, concrete state information is observed from time to time while the system to be checked is running. This runtime information is used to guide model checking to reduce the model space to be explored. In this sense, we call this method online model checking. Since we do not directly check the execution trace itself, our online checking at model level is capable of checking a running system some steps ahead of the actual state of execution. In this paper, we describe online model checking as well as the underlying system architecture in general, explain the basic algorithm and its extension to improve performance, and provide experimental results.