Extended Finite State Machine based Test Derivation Strategies for Telecommunication Protocols

Abstract

Abstract-In this paper, we consider the problem of test derivation based on an Extended Finite State Machine (EFSM) that is widely used for describing the behavior of telecommunication protocols and software. An EFSM augments a classical Finite State Machine (FSM) with context variables, input/output parameters and predicates. Tests based on various coverage criteria for EFSMs do not capture many functional faults and thus, there is a strong need for tests checking functional properties. Moreover, since there are no constructive necessary and sufficient conditions for checking whether two arbitrary EFSMs are equivalent, most methods are based on some kind of a transition tour, despite of the fact that such methods do not provide test suites with the guaranteed fault coverage. Given possibly nondeterministic and partial EFSM, we consider a transition tour of an FSM obtained by the simulation of the initial EFSM and provide some experimental results that such a test suite detects a number of inconsistencies in available protocol implementations with respect to protocol specifications. Since a transition tour augmented with state identifiers is known to have the higher fault coverage, we also discuss how state identifiers can be generated without facing the state explosion problem. Correspondingly, we consider FSM slices that are obtained by deleting from the initial EFSM all the context variables and possibly, input and output parameters. As the obtained FSM can be nondeterministic, a state identifier should contain separating sequences for pairs of states and we adapt the known techniques for deriving separating sequences for nonobservable partial FSMs.

Increasing the Fault Coverage of Tests Derived Against Extended Finite State Machines

“…Distinguishing sequences for two FSMs are constructed based on the product of these machines [11]; for EFSMs it is a bit more complex, since appropriate FSM abstractions are derived first [16,17,18]. Such abstractions can be derived in various ways, for example, we can simply delete all predicates, context variables, input and output parameters and updating functions.…”