MATRIX: A message passing interface for MPMD (Multiple Program Multiple Data) applications on heterogeneous distributed network

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

This paper presents a message passing interface called MATRIX for the MPMD (Multiple Program Multiple Data) type parallel application running on a distributed network. The proposed interface provides network abstraction and peer-to-peer message transfer method to the application layer. The interface is designed to have scalability for a large-scale application. The concurrent application established on the proposed interface can be applied to a heterogeneous distributed network composed of multi-architectures and multi-OS machines. This paper describes the concepts, protocols and implementations of the message passing interface. Additionally, we present the remote file transfer and execution protocol called FTXP for large-scale applications on the message passing interface. For the verification of the proposed methods, we present an example of parallel application applied to a surveillance robot system composed of multi-architecture and multi-OS.

Original languageEnglish
Title of host publication2012 9th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2012
Pages310-315
Number of pages6
DOIs
StatePublished - 2012
Event2012 9th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2012 - Daejeon, Korea, Republic of
Duration: 26 Nov 201229 Nov 2012

Publication series

Name2012 9th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2012

Conference

Conference2012 9th International Conference on Ubiquitous Robots and Ambient Intelligence, URAI 2012
Country/TerritoryKorea, Republic of
CityDaejeon
Period26/11/1229/11/12

Keywords

  • Distributed robotics
  • Message passing interface

Fingerprint

Dive into the research topics of 'MATRIX: A message passing interface for MPMD (Multiple Program Multiple Data) applications on heterogeneous distributed network'. Together they form a unique fingerprint.

Cite this