Type: Book Chapter

Lossy Channels in a Dataflow Model of Computation

Journal: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (03029743)Year: 2018/01/01Volume: Issue:

Fradet, PascalGirault, AlainJamshidian, LeilaNicollin, XavierShafiei A.^a

a :University of Isfahan - IRAN(IR) - Isfahan

DOI:10.1007/978-3-319-95246-8_15Language: English

Abstract

In this paper, we take into account lossy channels and retransmission protocols in dataflow models of computation (MoCs).Traditional dataflow MoCs cannot easily cope with lossy channels, due to the strict notion of iteration that does not allow the re-emission of lost or damaged tokens. A general dataflow graph with several lossy channels will indeed require several phases, each of them corresponding to a portion of the initial graph’s schedule. Correctly identifying and sequencing these phases is a challenge. We present a translation of a dataflow graph, written in the well-known Synchronous DataFlow (SDF) MoC of Lee and Messerschmitt, but where some channels may be lossy, into the Boolean Parametric DataFlow (BPDF) MoC. © 2018, Springer International Publishing AG, part of Springer Nature.

Author Keywords

Computation theoryResponse time (computer systems)