Adaptive Beamforming and Optimization of Near-field Ultra Large Scale MIMO Systems Using ADMM and RL Algorithms

In the near field wireless communication system, in order to improve the performance and efficiency of the system, the ultra-large scale MIMO (Multiple Input Multiple Output) technique is usually used. The traditional methods have problems such as low efficiency of beam adjustment and inability to adapt to real-time channel state changes, etc. In this paper, we study the adaptive beam adjustment and optimization method based on ADMM (Alternating Direction Multiplier Method) and RL (Reinforcement Learning) algorithms in order to improve the performance and efficiency of the system. The paper is based on ADMM (Alternating Direction Multiplier Method) and RL (Reinforcement Learning) algorithms. In this paper, adaptive beam tuning and optimization of the system is achieved by decomposing the beam tuning problem into sub-problems and using ADMM algorithm for alternate optimization, which is combined with RL algorithm to realize the learning of intelligent beam tuning strategy. The experimental results show that the investigated method effectively improves the communication performance and system efficiency in ultra-large-scale MIMO systems, adapts to the changes of different channel states and communication demands, and combines the RL algorithm and the ADMM algorithm with the lowest system packet loss rate of only 2.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>, the reliability of the system reaches up to 94.9%, and the system efficiency of the system reaches up to 91.9%. This study provides a new solution idea for beam tuning and optimization of ultra-large-scale MIMO systems, which has certain practicality and promotion value.