Switched PIOA: Parallel composition via distributed scheduling

Abstract

This paper presents the framework of switched probabilistic input/output au tomata (or switched PIOA), augmenting the original PIOA framework with an explicit control exchange mechanism. Using this mechanism, we model a network of processes passing a single token among them, so that the location of this token determines which process is scheduled to make the next move. This token struc ture therefore implements a distributed scheduling scheme: scheduling decisions are always made by the (unique) active component. Distributed scheduling allows us to draw a clear line between local and global non-deterministic choices.We then require that local non-deterministic choices are resolved using strictly local information. This eliminates unrealistic schedules that arise under the more common centralized scheduling scheme. As a result, we are able to prove that our trace-style semantics is compositional.

On Model Checking Techniques for Randomized Distributed Systems

“…In the literature, several classes of schedulers have been considered that are more adequate for reasoning about distributed systems [12,18,1]. Partiallyobservable MDPs, briefly called POMDPs, can be seen as a simple variant that can serve to model the view of one process.…”

“…This problem can be avoided by introducing some appropriate concept of distributed scheduling [12]. But as the series of undecidability results for POMDPs shows, the price one has to pay when switching from the full class of schedulers to more realistic ones is the loss of model checking algorithms for the quantitative analysis against infinite-horizon path events, and partly also verification algorithms for qualitative properties.…”

Task-structured probabilistic I/O automata

“…However, for extra expressive power, we define a second mechanism called local schedulers, which uses local information only to resolve nondeterminism within system components. This mechanism is based on earlier work in [CLSV06].…”

Compositionality for Markov reward chains with fast and silent transitions

“…This includes stochastic process algebras [2,3], (generalized) stochastic Petri nets [4,5], probabilistic I/O automata [6,7], stochastic automata networks [8], etc. The compositional modeling enables composing a bigger system from several smaller components.…”

“…In the process of compositional modeling, performance evaluation techniques produce intermediate constructs that are typically extensions of Markov chains featuring transitions with communication labels [2][3][4][5][6][7][8]. In the final modeling phase, all labels are discarded and communication transitions are assigned instantaneous behavior.…”