🤖 AI Summary
To address the challenges of frequent blockage of line-of-sight (LoS) paths and low communication and energy efficiency in millimeter-wave device-to-device (D2D) networks, this paper proposes a joint reconfigurable intelligent surface (RIS) deployment and user grouping optimization framework based on set cover theory. We theoretically prove and empirically validate— for the first time—that in typical non-line-of-sight (NLoS) scenarios, a two-RIS cascaded reflection link outperforms conventional single-RIS relaying, thereby breaking the single-hop reflection paradigm. Leveraging this insight, we design a geometric ray-tracing-assisted RIS placement strategy and an energy-efficiency-driven dynamic user grouping mechanism. Experimental results demonstrate that, compared to random and state-of-the-art (SOTA) deployment schemes, the proposed approach reduces RIS deployment density by approximately 40%, increases the number of successfully served blocked D2D pairs by 65%, and improves end-to-end energy efficiency by 32%.
📝 Abstract
Reconfigurable intelligent surfaces (RISs) offer a viable way to improve the performance of multi-hop device-to-device (D2D) communication. However, due to the substantial propagation and penetration losses of the millimeter waves (mmWaves), a direct line of sight (LoS) link and close proximity of a device pair are required for a high data rate. Static obstacles like trees and buildings can easily impede the direct LoS connectivity between a device pair. Hence, RIS placement plays a crucial role in establishing an indirect LoS link between them. Therefore, in this work, we propose a set cover-based RIS deployment strategy for both single and double RIS-assisted D2D communication. In particular, we have demonstrated that permitting reflections via two consecutive RISs can greatly lower the RIS density in the environment, preventing resource waste and enabling the service of more obstructed device pairs. After the RIS deployment, for information transfer, we also propose an energy-efficient group selection criteria. Moreover, we prove that sometimes double reflections are more beneficial than single reflection, which is counter-intuitive. Numerical results show that our approach outperforms a random and a recent deployment strategy.