Formal Model Simulation: Can It be Guided?

In this work we present an approach to system development that bridges the complementary worlds of model checking and simulation. This approach, which we refer to as formal model simulation, uses the same formal model as model checking, but instead of using a total search of the state space, we use a guided random-walk based search to find errors. The guide we use is a heuristic that can be derived from an abstract model of the system. We have implemented the technique in a tool called GRANSPIN, which is derived from the popular model checker SPIN. A series of experiments is outlined in this work using different selection strategies and different heuristics. We compare the performance of these different strategies and heuristics in GRANSPIN and SPIN on a buggy Peterson protocol.