A Deep Learning Approach for Target-oriented Communication Resource Allocation in Holographic MIMO

Apurba Adhikary, Md. Shirajum Munir, Avi Deb Raha, Min Seok Kim, Jong Won Choe, Choong Seon Hong

http://doi.org/10.5626/JOK.2023.50.5.441

In this paper, we propose a single-cell massive multiple-input multiple-output (mMIMO) system assisted with holography that performs target-oriented communication resource allocation for heterogeneous users. This paper proposes a technique that can minimize the number of active grids from holographic grid arrays (HGA) for confirming the requirement of lower power toward beamforming to serve target-oriented users. Therefore, we formulated a problem by maximizing the signal-to-interference-noise ratio (SINR), which, in turn, maximizes the efficient resource allocation for the users by generating effective beamforming and controlling the sum-power rule. Additionally, our holography-assisted mMIMO system is capable of serving heterogeneous user equipment simultaneously with a lower power budget. To devise the artificial intelligence (AI)-based solution, we developed a sequential neural network model for grid activation decisions with minimized power constraint. Finally, the simulation and performance evaluation results show that power was allocated efficiently, and effective beams were formed for serving the users with a lower RMSE score of 0.01.