Error‐preserving reductions on communication protocols

Abstract

Common logical design errors in communication protocols are typically detected through reachability analysis or state‐based model checking techniques. These techniques suffer from a state explosion problem and a variety of methods have been explored to deal with it. In this paper, two transformation rules are proposed to be applied specifically on the specifications of communication protocols to reduce their sizes while preserving common logical design errors. These rules are applied on the specification of each individual process in the protocols before state exploration and thus can be combined with most of the existing reachability analysis or model checking techniques applied during state exploration. Copyright © 2007 John Wiley & Sons, Ltd.

Error‐preserving local transformations on communication protocols

“…Both EPLTs, in the following called Θ 1 and Θ 2 , are very elementary, but when applied to a CSM in an adequate combination, they facilitate elimination of its globally insignificant states and exploitation of commutation of its mutually independent transitions . It is, hence, not surprising that, as demonstrated by Duan and Chen 3, their application to individual system components complements partial order reduction (POR) 5, 6, the most typical EPLT applied during reachability analysis, whose essence is exploitation of commutation of mutually independent transitions belonging to different CSMs of the considered system. Pre‐processing with Θ 1 and Θ 2 is helpful 3 also if the employed reachability analysis technique is simultaneous reachability analysis (SRA) 7, 8, another technique exploiting commutation of transitions belonging to different system components.…”

“…It is, hence, not surprising that, as demonstrated by Duan and Chen 3, their application to individual system components complements partial order reduction (POR) 5, 6, the most typical EPLT applied during reachability analysis, whose essence is exploitation of commutation of mutually independent transitions belonging to different CSMs of the considered system. Pre‐processing with Θ 1 and Θ 2 is helpful 3 also if the employed reachability analysis technique is simultaneous reachability analysis (SRA) 7, 8, another technique exploiting commutation of transitions belonging to different system components. After exhaustive pre‐processing with Θ 1 and Θ 2 , there is, however, no need to use the more advanced blocking SRA (BSRA) 4, whose advantage is in treating specific sequences of transitions of individual system components as compound transitions, where the pre‐processing already transforms every such sequence into one transition.…”

“…The paper was motivated by the work of Duan and Chen 3, who innovatively proposed two EPLTs for systems of CSMs differing from the classical CFSM definition 1 in that an individual transition could denote un‐interruptible and atomic execution of a sequence of multiple message receptions and/or transmissions . Their motivation for considering a more general kind of CSM was that in the reachability analysis, it is sometimes acceptable to treat specific sequences of events as one atomic action 4.…”

“…As different reachability analysis techniques rely on different EPLTs and new techniques are being invented, it is desirable to have both a variety of easily applicable EPLTs to choose from for the pre‐processing and a formal setting for easy development of further useful EPLTs. The paper drops the assumption of Duan and Chen 3 that every system component is deterministic and minimal and has only a finite number of specified states and transitions, at most one outgoing channel per destination, at most one incoming channel per source and no private channels. It proves that Θ 1 and Θ 2 nevertheless remain error‐preserving.…”

“…Furthermore, it proposes four new simple EPLTs , in the following called Θ 3 to Θ 6 , and a generic EPLT Θ which strongly generalizes Θ 1 to Θ 6 and from which further easily applicable EPLTs can be derived simply by specialization . For all the transformations, the paper, like Duan and Chen 3, also discusses how (non‐)executability of CSM transitions in the new system version reflects (non‐)executability of those in the old one.…”