Abstract
Digital instrumentation and control (I&C) systems are
increasingly used in the nuclear engineering domain. The
exhaustive verification of these systems is challenging,
and the usual verification methods such as testing and
simulation are typically insufficient. Model checking is
a formal method that is able to exhaustively analyse the
behaviour of a model against a formally written
specification. If the model checking tool detects a
violation of the specification, it will give out a
counter-example that demonstrates how the specification
is violated in the system. Unfortunately, sometimes real
life system designs are too big to be directly analysed
by traditional model checking techniques. We have
developed an iterative technique for model checking large
modular systems. The technique uses abstraction based
over-approximations of the model behaviour, combined with
iterative refinement. The main contribution of the work
is the concrete abstraction refinement technique based on
the modular structure of the model, the dependency graph
of the model, and a refinement sampling heuristic similar
to delta debugging. The technique is geared towards
proving properties, and outperforms BDD-based model
checking, the k-induction technique, and the property
directed reachability algorithm (PDR) in our experiments.
Original language | English |
---|---|
Pages (from-to) | 120-130 |
Journal | Reliability Engineering and System Safety |
Volume | 139 |
DOIs | |
Publication status | Published - 2015 |
MoE publication type | A1 Journal article-refereed |
Keywords
- model checking
- verification
- validation
- iterative abstraction refinement