Modeling and Verifying Real-Time Properties of Reactive Systems

Abstract

SPACE is a model-driven engineering technique for reactive distributed systems. It enables to develop system models from reusable building blocks, formal analysis by model checking as well as automated transformation to executable code. In this paper, we describe an extension of the SPACE formalism which allows to model and verify also real-time behavior. In particular, one specifies real-time constraints in the interface descriptions of the building blocks, so-called Real-Time External State-Machines (RTESMs). The RTESMs are translated to guards, clocks and invariants of Timed Automata which can be analyzed by means of the model checker UPPAAL. The approach is explained by a component protecting an electrical motor controller system against overspeed. In particular, we prove that by keeping certain maximum response times, this system guarantees that the speed of the motor stays within certain limits.

Towards Verifying Safety Properties of Real-Time Probabilistic Systems

“…Following the concept of Timed Automata [1], we extended our external state machines (ESM) to Real-Time ESMs (RTESM) in [12,13]. RTESMs allow the specification of deadlines for the time a building block may rest in a certain RTESM state.…”

“…The values in the table do not correspond to an existing system, but rather represent typical values one might expect in some field-bus based systems. Following the concept of Timed Automata [1], we extended our external state machines (ESM) to Real-Time ESMs (RTESM) in [12,13]. RTESMs allow the specification of deadlines for the time a building block may rest in a certain RTESM state.…”

“…The composed system model is automatically transformed into executable Java code [17]. Further, various formal verification methods ensure functional correctness [17] as well as reliability [27], security [9] and safety [12,13] of targeted systems.…”

“…Here, we propose a framework for integrating probabilistic real-time verification and performance prediction with system development. This approach extends our existing model-driven engineering framework SPACE and its tool-set Reactive Blocks 1 [17] with real-time system behavior verification and schedulability analysis [12,13]. Reactive Blocks enables the model-based engineering of reactive systems by composing reusable building blocks each describing a certain sub-functionality of a system.…”

“…PRISM [19] is a probabilistic model checker for formal analysis of systems that exhibits stochastic behaviors. It supports multiple-formalisms, including discrete-time Markov chains, continuous-time Markov chains, Markov decision processes and PTA making it possible to capture the random behavior of our real-time system model [12], for example random aspects of failure or uncertain inputs, loads or timing. We choose PTA for the stochastic behavior since it integrates well with our existing timed-automata based real-time system model.…”