AIDA: A holistic AI-driven networking and processing framework for industrial IoT applications
Session Recording
Click here to access slides from the talk
Join Red Hat Research for the next Research Days event, “AIDA: A holistic AI-driven networking and processing framework for industrial IoT applications” on September 21, 2023 from 3PM to 4:30PM CEST (9AM EDT, 4PM IDT).
Recently, we’ve seen AI used for a plethora of use cases. But is it already mature enough for real-time industrial applications? Can it work under tight time constraints in large IoT applications? In this talk, the team from Karlstad University, led by Professor Anna Brunstrom, will introduce their research on a holistic AI-driven framework for reliable data-driven real-time industrial IoT applications. Red Hatters and project participants Simone Ferlin-Reiter, Senior Performance Engineer, and Toke Høiland-Jørgensen, Principal Engineer, will lead the conversation.
Abstract
Ongoing efforts in digitalization are leveraging the development of Internet-connected smart sensors and the processing capabilities of cloud and edge based systems. Such processing requires trustworthy data gathering and machine learning (ML) algorithms in order to improve production processes, predict machine failures, etc. These algorithms need to be integrated with networked real-time decision systems in order to react under tight timing constraints. Such integration requires frameworks that flexibly integrate and program networking and computing infrastructures, while allowing ML pipelines to ingest collected data and make trustworthy decisions in real time.
In this talk, we present AIDA – a novel holistic AI-driven network and processing framework for reliable data-driven real-time industrial IoT applications. The AIDA architecture aims to integrate three pillars required to enable reliable industrial IoT applications: (1) a converged real-time network infrastructure, which supports flexibility and dynamic reconfiguration using a software-defined approach for Time-Sensitive Network (TSN) configuration management; (2) an industrial edge–cloud continuum that integrates flexible observability components using data-driven algorithms; and (3) agile and trustworthy ML/AI algorithms that are embedded in the real-time industrial data stream processing infrastructure deployed in the edge/cloud continuum. We provide an overview of the AIDA architecture and its key building block and take a closer look at its distributed observability framework and ML pipeline.
The AIDA framework is developed within a research project run by Karlstad University in collaboration with six industry partners, including Red Hat. The project is funded by the Knowledge Foundation of Sweden.
See the Red Hat Research project page for AIDA – A holistic AI-Driven networking and processing framework for Industrial IoT
Speakers
Anna Brunstrom, Professor, Karlstad University
Muhammad Usman, Associate Senior Lecturer, Karlstad University
Bestoun Ahmed, Professor, Karlstad University
Conversation Leaders
Simone Ferlin-Reiter, Senior Performance Engineer, Red Hat
Toke Høiland-Jørgensen, Principal Engineer, Red Hat
The recording and materials will be available following the talk. Learn more about Red Hat Research Days Events and watch the recordings from previous events at research.redhat.com/research-talks