Joint Beamforming and Reflecting Elements Optimization for Segmented RIS Assisted Multi-User Wireless Networks

The emerging reconfigurable intelligent surface (RIS) is a prospective technique to modulate the wireless channel and improve performance, in which large amounts of passive elements manipulate independently, inevitably resulting in a high-dimensional optimization problem that is intractable to solve. With the aim to strike a balance between optimality and complexity for RIS assisted multi-user systems, in this article, we formulate the achievable sum rate maximization problem under a novel RIS segmentation structure, where the distributions and sizes of each segmentation can be adaptively adjusted. Since the formulated optimization problem considering the quality of service (QoS) requirements for the users is non-convex, we suggest a computationally-efficient approach to derive an optimal solution by exploiting fractional programming, successive convex approximation (SCA), greedy algorithm, and alternating optimization. Finally, numerical simulations reveal that the proposed optimization design enables RIS to configure by grouping elements into some sub-surfaces without significant performance degradation while with much lower computational complexity than conventional element-wise optimization RIS. Moreover, our proposed adjustable segmentation outperforms the fixed one employing the determined positions and equal number of reflecting elements in each sub-surface. Additionally, the results demonstrate that the optimization of segmentation is much more significant than the phase shift optimization, showing the superiority and practical significance of the sub-surface segmentation strategy.