ABSTRACT With the upgrading of mobile communication technology and the rise of the mobile communications industry, the mobile Internet is booming. However, mobile Internet networks are facing the next technological revolution due to the explosive growth of mobile devices, the expanding scale of networks, and the increasing demand of users for quality of service. To overcome the problems of high computational overhead and ineffective use of network history data information in traditional optimization algorithms, a graph neural network‐based joint user scheduling and power allocation model is proposed, combined with an analytical formulation of beam vectors, to achieve joint user scheduling and beamforming optimization. Simulation analysis shows that the proposed algorithm improves the convergence speed by nearly 20% compared with the traditional scheduling algorithm and saves 4.8% energy consumption compared with the widely used base station always‐on strategy.
Jingya Zhang (Fri,) studied this question.