DUST, the Distributed Uniform STreaming library for distributed artificial intelligence

Visual of dust
Description

DUST, the Distributed Uniform STreaming library, is a library/framework currently being developed internally by IDLab, with the collaboration of University of Antwerp and imec. DUST is intended to simplify the creation and maintenance of distributed software. It aids developers by abstracting away the communication layer, allowing them to modularly create their communication stack using modules and add-ons. DUST provides several other benefits as well, such as ease of use, extendable, multi-platform, transparency, and support for multiple programming languages. Due to its layered architecture, developers only need to know about the layer they want to work with and can abstract away the rest of the knowledge.

Contributions from AI Flanders program: a configurable communication layer for the coordinator, allowing to distribute tasks across a network of devices based on optimal allocation strategies that are developed within the Grand Challenge 2 "Edge AI". The framework can be used for rapid prototyping and designing distributed AI applications.  

More information
  • Grand Challenge/Workpackage(s): GC2 – WP2
  • Responsible Research Lead: Peter Hellinckx 
  • Authors: Simon Vanneste, Jens de Hoog, Thomas Huybrechts, Stig Bosmans, Reinout Eyckerman, Muddsair Sharif, Siegfried Mercelis, and Peter Hellinckx.
  • Contact: Christiaan Leysen 
  • Programming language: Python
  • Link to the website 
  • Category/Categories: Distributed intelligence
Publications
  • Simon Vanneste, Jens de Hoog, Thomas Huybrechts, Stig Bosmans, Reinout Eyckerman, Muddsair Sharif, Siegfried Mercelis, and Peter Hellinckx. "Distributed Uniform Streaming Framework: An Elastic Fog Computing Platform for Event Stream Processing and Platform Transparency" Future Internet 11, no. 7: 158, 2019. (https://doi.org/10.3390/fi110701) 
  • Simon Vanneste, Jens de Hoog, Thomas Huybrechts, Stig Bosmans, Reinout Eyckerman, Muddsair Sharif, Siegfried Mercelis, and Peter Hellinckx. " Distributed uniform streaming framework: towards an elastic fog computing platform for event stream processing". International Conference on P2P, Parallel, Grid, Cloud and Internet Computing, p426-436, 2018. (https://www.springerprofessional.de/en/distributed-uniform-streaming-framework-towards-an-elastic-fog-c/16204432)
  • Thomas Huybrechts, Reinout Eyckerman, Raf Van den Langenbergh, Simon Vanneste, Siegfried Mercelis, and PeterHellinckx. “DUST Initializr - Graph-based platform for designing modules and applications in the revised DUST framework”. In Internet of Thing, 2020. (https://doi.org/10.1016/j.iot.2020.100229)
  • Thomas Huybrechts, Simon Vanneste, Reinout Eyckerman, Jens de Hoog, Siegfried Mercelis, and Peter Hellinckx. “DUST Initializr - CAD Drawing Platform for Designing Modules and Applications in the DUST Framework”. International Conference on P2P, Parallel, Grid, Cloud and Internet Computing, p661-670, 2019. (https://link.springer.com/chapter/10.1007%2F978-3-030-33509-0_62)