🤖 AI Summary
To address the high computational complexity of joint base station precoding and RIS phase optimization in RIS-assisted wireless systems, this paper proposes a Riemannian manifold-based co-optimization framework in the complex domain. Our method innovatively integrates complex-circle geometry—modeling RIS phase shifts—with spherical geometry—modeling unit-norm precoding vectors—to establish the first meta-learning architecture tailored for RIS optimization. We further introduce an Euler-angle-parameterized complex-valued neural network, coupled with an Euler-inspired update rule, to accelerate convergence and enhance cross-channel generalization. Optimizing for weighted sum rate maximization, the proposed approach achieves approximately 100-epoch faster convergence than baseline algorithms, improves weighted sum rate by 0.7 bps/Hz, and reduces power consumption by 1.8 dBm.
📝 Abstract
In reconfigurable intelligent surface (RIS) aided systems, the joint optimization of the precoder matrix at the base station and the phase shifts of the RIS elements involves significant complexity. In this paper, we propose a complex-valued, geometry aware meta-learning neural network that maximizes the weighted sum rate in a multi-user multiple input single output system. By leveraging the complex circle geometry for phase shifts and spherical geometry for the precoder, the optimization occurs on Riemannian manifolds, leading to faster convergence. We use a complex-valued neural network for phase shifts and an Euler inspired update for the precoder network. Our approach outperforms existing neural network-based algorithms, offering higher weighted sum rates, lower power consumption, and significantly faster convergence. Specifically, it converges faster by nearly 100 epochs, with a 0.7 bps improvement in weighted sum rate and a 1.8 dBm power gain when compared with existing work.