Executing Model Checking Counterexamples in Simulink

Abstract

Verification of embedded systems has become increasingly important in many industrial domains. Safetycritical embedded systems, such as those developed in aerospace industry, are regularly subject to automated formal verification process. In this paper we extend our tool integration chain of parallel, explicit-state LTL model checker DIVINE and Matlab Simulink tool suit with an improved support of counterexample simulation. In particular, we show how to provide the verification engineer with a direct connection between the error discovered by the model checker and the simulation in Matlab Simulink. This work has been conducted within the Artemis project industrial Framework for Embedded Systems Tools (iFEST).

Automated test suite generation for time-continuous simulink models

“…It has been previously used to detect faults in Simulink models [14,21,5,31] by showing that a path leading to an error (e.g., an assertion or a runtime error) is reachable, or by maximizing structural coverage (e.g., by executing as many paths as possible in a model). To solve the reachability problem or to achieve high coverage, these techniques often extract constraints from the underlying Simulink models and feed the constraints into some constraint solver or SAT solver.…”

Requirements simulation for early validation using Behavior Trees and Datalog

“…Bus protocol specifications are used for simulation to automatically generate the verification aids in [39]. Counterexamples of the Simulink models have been simulated to visually interpret them, which not only accelerates the process of requirements error detection but also discover misinterpreted requirements [40]. A unified environment SQUANDER has been proposed in [41] that can execute specifications written in the JForge specification language (JFSL) along with imperative code.…”

Analysis and testing of matlab simulink models: a systematic mapping study

“…Three of them [1,14,39] discuss combinations of analysis and simulation (which we consider here as part of testing, as the articles do as well). Garro and Tundis [1] present a model-based method for checking system reliability, which is combined in a unified framework with the Mathworks simulation and analysis environments.…”

“…Garro and Tundis [1] present a model-based method for checking system reliability, which is combined in a unified framework with the Mathworks simulation and analysis environments. A direct connection between defects discovered by the model checker and the simulation environment in Matlab Simulink is provided in [14]. In the last article of this subcategory, combinations of numerical simulation and symbolic analysis are presented [39].…”

“…own example literature example [2,17,18,21,32,35,38,42] 5 10 15 20 [23,39] 25 [5,11,12,14,15,20,22,25,30,33,34] none Matlab example industry example [1, 2, 4, 6, 7, 8, 9, 10, 13, 16, 17, 18 ,19, 23, 24, 26, 27, 28, 29, 31, 35, 36, 37, 38, 42, 44] [6] use 20 different industrial models to evaluate their approach; Böhr [26] uses only one example. However, even in these cases, the models are of different sizes.…”