I want to predict the beamforming matrix for BS (Base Station) and the phase shift matrix for RIS (Reconfigurable Intelligent Surface). How can I obtain a large number of past successful optimization results of beamforming matrices and phase shift matrices to use as labels for supervised pre-training?
Emil Björnson
The most straightforward approach is to solve the problem many times using optimization theory, to create your dataset. The purpose of using ML shouldn’t be to improve performance but reduce the signaling, or something like that.
One example is the following paper:
https://arxiv.org/pdf/2009.13988.pdf
I believe there are more recent and better performing examples.
3 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Writing
More Chris Lyu's questions See All
Similar questions and discussions