eduzhai > Applied Sciences > Engineering >

Optimization-driven Machine Learning for Intelligent Reflecting Surfaces Assisted Wireless Networks

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 16 pages

Abstract: Intelligent reflecting surface (IRS) has been recently employed to reshapethe wireless channels by controlling individual scattering elements phaseshifts, namely, passive beamforming. Due to the large size of scatteringelements, the passive beamforming is typically challenged by the highcomputational complexity and inexact channel information. In this article, wefocus on machine learning (ML) approaches for performance maximization inIRS-assisted wireless networks. In general, ML approaches provide enhancedflexibility and robustness against uncertain information and imprecisemodeling. Practical challenges still remain mainly due to the demand for alarge dataset in offline training and slow convergence in online learning.These observations motivate us to design a novel optimization-driven MLframework for IRS-assisted wireless networks, which takes both advantages ofthe efficiency in model-based optimization and the robustness in model-free MLapproaches. By splitting the decision variables into two parts, one part isobtained by the outer-loop ML approach, while the other part is optimizedefficiently by solving an approximate problem. Numerical results verify thatthe optimization-driven ML approach can improve both the convergence and thereward performance compared to conventional model-free learning approaches.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×