3DML: Research

3DML - Data:

  • Enable the automated collection of user-defined datasets labeled by design, with the ability to operate in networks at different scales.

  • Take advantage of the fully observable stack being developed by the POWDER-RENEW platform to collect rich datasets that capture diverse network operating states and configurations.

3DML - Client:

  • An automated model-builder to generate efficient ML models and deployment strategies given an application (i.e., dataset), performance specification (e.g., task accuracy and energy efficiency), and target deployment device (e.g., drones).

  • A model compression library of efficient ML modules and functions to be readily available to the model-builder and non-expert developers for fast development.

3DML - Infrastructure:

  • Make use of 3DML-Client with data collected from 3DML-Data to generate ML models for wireless infrastructure.

  • A methodology to customize key modules for Massive MIMO channel estimation, detection, decoding, beamforming, and spectrum sharing.