Designing distributed systems w.r.t conformance

Abstract

This thesis is about revisiting an old yet classic problem - From a labeled transition system (LTS), can a distributed labeled transition system (DLTS) be synthesized such that the behavior of both systems are equivalent. This problem has been addressed in various papers using behavioral equivalence classes which were language equivalence, bisimulation and isomorphism where the strictness increased from former to latter. It was discovered that not all LTS are distributable even if the LTS is acyclic. Then, another term was coined as Conf. which was a preorder relation and was derived from Conformance testing. The general concept of Conformance was to inculcate extra behavior when not specified. Advantage of using this relation was that now every acyclic LTS is distributable. This thesis walks through all the survey done over the time and after understanding the advantages and disadvantages of each kind of behavioral equivalence, we finally implement an algorithm which when given any acyclic transition system as an input will generate a synchronous product of the distributed transition systems as output.