Prognostics CoE Publications

We detail all of the facets of adapting classical model checking to a real aerospace system, in- cluding deriving the formal model and a set of specifications from natural language descriptions. To ensure the model checking results are meaningful, we have to ensure that both the model and specifications correctly reflect the intentions of the designers, thus we employ model validation and property debugging techniques. We demonstrate the utility of enhancing LTL satisfiability checking by taking the fairness constraints of the system model into consideration. We argue that specification debugging in real applications deserves more attention in future research efforts, and the utility of a system formalization, model and specification debugging, and verification trilogy for model checking real systems under development.
In this paper we assume there are no hardware failures or lost messages. As the AAC design develops and hardware details are decided by AAC designers, we plan to take the failure rates of the chosen components into consideration, i.e. by extending our work to probabilistic model checking using PRISM [19]. Previous work has reported on analyzing the safety of air traffic control systems using simulation [3] or fault trees [1]. By extending the model we designed in this paper, we can carry out safety analysis using PRISM to capture the dynamic interactions in the AAC.