Formal verification tools are useful in distinguishing faulty from correct implementations of the same protocol, which are both supposed to follow a commonly agreed upon specification.
This paper describes a new method for protocol conformance testing that enhances the current practice: the ISO-9646/1992 standard. Enhancements relate to the way a normal finite state machine (FSM) is used for testing: all transitions following failing transitions are themselves invalid, and are not worth exploration. The dynamic conformance test method (DCTM) performs an analysis of the state set in the FSM at runtime, and dynamically adjust behavior depending on whether the state fails or not (preamble and postamble sets of states are used). The algorithm is presented with a fair amount of detail, but the presentation is sometimes careless. Comparing DCTM to conventional testing on an analytical state/transition representation of the transmission control protocol (TCP) yields a better behavior for DCTM (the number of needlessly considered faulty transitions is lower).
Graph analysis of FSMs is sometimes weak; even if the examples given are only for explanatory purposes, some of the choices lack explanation (for example, transforming a nonloop-free graph into a tree involved unclear decisions).
A more realistic analysis of the usefulness of the DCTM method could be achieved by considering an actual TCP implementation (and not only its representation). Actual TCP behavior also depends on the state sequence followed by the other end’s implementation.
The paper is well structured and easily readable, even though a picky reader might find some spelling errors. The comparison of DCTM to the ISO standard is consistently intertwined along sections.