This Diploma Thesis explores the possibilities of formal modeling and analysis of the airbag control unit, and has been carried out in close collaboration with Robert Bosch GmbH, CR/AEA in Frankfurt. The on-hand work strives for an assessment of the dependability of the air bag control unit manufactured by Bosch. Image

Based on a formal concept of synchronization concepts, the modelling languages of Uppaal [LAR97] and MoDeST [ARG01] are exemplified that prepare grounds for later modelling of the airbag's control unit. Using simulation in MoDeST the former defined LTL properties e.g. "Does the airbag deploy when it should?" - are proven to assure the proper behavior and reliable function of the underlying device. In the succeeding, Matlab Simulink is used to analyze the ECU model and derive a statistical figure which reveals the deployment probability over a time horizon of 12.000 time units. Security requirements like MTTF et cetera are evaluated using a Markov chain model, reflecting the airbag controllers life cycle by simulation in MoDeST, testified at an analytical level by exponentiating the transition matrix. A feasibility analysis of simulating parts of the ECUs failure tree is executed in MoDeST where results are confered with outcomes of commercial tools for Fault Tree Analysis (FTA).The automated fault tree generation is applied for a simple airbag controller model exhibiting a way, how fault trees can be extracted out of the ECUs behavior model. Based on a pre-processor grammar, artificial failures are inserted like noise and the delay of signals that contribute to a failure of malfunction of the device under consideration.

The general principles of Importance Analysis - especially the structural, marginal, Barlow-Proschan, and Fussell-Vesely measure are treated in the following, giving examples and comparing the advantages and cons of each over the others. In the end, requirements of the visionary Single Source Model that can be used for simulation, verification, dynamic- and static failure analysis are discussed and the advantages and respective disadvantages are pointed out.


The thesis is available here.


[LAR97] K. G. Larsen, P. Pettersson,W. Yi. Uppaal in a Nutshell. Int. Journal on Software Tools for Technology Transfer, October 1997.
[ARG01] P. D'Argenio, H. Hermanns, J.-P. Katoen, R. Klaren. MoDeST -- A Modelling and Description Language for Stochastic Timed Systems. PAPM-PROBMIV 2001, LNCS 2165, 2001.