Neural pub/sub

Neural 6G

The next-generation networks are expected to be optimized by AI processes through MLOps, and in-network AI is expected to provide significant benefits to applications. The neural publish/subscribe paradigm proposes weaving AI into the communication substrate and within the edge-cloud continuum decoupling communicating endpoints in space, time, and synchronization.

Funders

This figure illustrates the three key components of the Neural Pub/Sub system (Publisher, Broker, and Subscriber), as well as the testbed development plan for the implementation of our solution. Each component features its own functionality stack, underpinned by 5G-based (or semantic-based for 6G transition) communication and orchestration layers. The Broker’s central role in AI model management is evident, while the Publisher’s role includes AI services provision and application enablement.
Neural pub/sub realizes many-to-many deep learning-based information dissemination and distributed inference and learning

Project information

Project duration

-

Funded by

Business Finland

Project coordinator

University of Helsinki

Unit and faculty

Center for Ubiquitous Computing
Faculty of Information Technology and Electrical Engineering

Contact information

Project leader

  • Professor
    Jukka Riekki

Contact person

Project description

We propose the new neural publish/subscribe paradigm for weaving AI into the communication substrate and within the edge-cloud continuum. Neural pub/sub realizes many-to-many deep learning-based information dissemination and distributed inference and learning.

Research groups

Interactive Edge