Joint Beamforming Design Based on Graph Neural Convection Diffusion Network

Intelligent Reflecting Surface (IRS), as an emerging green communication technology, improves the communication quality of wireless communication networks by reconfiguring the wireless propagation environment. This paper focuses on IRS-assisted multiple input single output wireless communication systems, considering the j oint optimization of transmission beamforming at the Access Point (AP) and reflection beamforming of passive phase shifters at the IRS under the constraint of total transmission power at the AP to maximize the sum rate of the users. To solve the nonconvex optimization problem of multivariable coupling, this paper develops a graph neural convection diffusion model based on the convection diffusion equation to solve the joint beamforming design. The experimental results show that under the same conditions, the proposed algorithm achieves a larger sum rate of users compared to other algorithms, demonstrating its effectiveness.