Equilibrium-Equation-Based Fast Recursive Principal Component Tracking with an Adaptive Forgetting Factor for Joint Spatial Division and Multiplexing Systems
IEEE Transactions on Wireless Communications（2024）
摘要
In massive
multiple-input multiple-output
(MIMO) systems, block diagonalization-based precoding methods are employed to mitigate interference among users by relying on channel state information. However, as the number of antennas increases, the task of eigenvalue decomposition or matrix inversion for the channel covariance matrix becomes progressively challenging. In this paper, we propose an equilibrium-equation-based recursive channel principal component tracking algorithm specifically designed for linear precoding in
Joint Spatial Division and Multiplexing
(JSDM) systems. Unlike many existing algorithms that depend on gradient formulations and the choice of step size, our proposed recursive tracking algorithm operates without the need for a step size, significantly enhancing convergence speed and learning performance. We also derive the adaptive forgetting factor, which improves the convergence capability. Additionally, we provide a mathematical analysis of the algorithm’s convergence performance, mean deviation, and learning curve. Finally, we implement various precoding strategies to a downlink channel in a massive MIMO JSDM system, leveraging the channel’s principal components. Our simulations conclusively demonstrate that the proposed algorithm outperforms traditional tracking algorithms, while principal component-based precoding effectively enhances spectral efficiency.
更多查看译文
关键词
Massive MIMO systems,principal component analysis,weighted subspace principle,forgetting factor,hybrid digital-analog precoding
AI 理解论文
溯源树
样例
生成溯源树，研究论文发展脉络
Chat Paper
正在生成论文摘要