A formal approach to heuristically test restorable systems

Abstract

In order to test a Finite State Machine (FSM), first we typically have to identify some short interaction sequences allowing to reach those states or transitions considered as critical. If these sequences are applied to an implementation under test (IUT), then equivalent states or transitions would be reached and observed in the implementation - provided that the implementation were actually defined as the specification. In this paper we study how to obtain such sequences in a scenario where previous configurations can be restored at any time. In general, this feature enables sequences to reach the required parts of the machine in less time, because some repetitions can be avoided. However, finding optimal sequences is NP-hard when configurations can be restored. We use an evolutionary method, River Formation Dynamics, to heuristically solve this problem. © 2009 Springer Berlin Heidelberg.