AI 中文总结
本文提出混合STAR-RIS架构，采用PARAFAC-ALS张量分解方法，通过交替迭代结合多种算法优化，实现联合定位、通信与功率传输，能效较全无源/有源架构提升1.5至3倍。
AI 中文摘要
本文提出一种混合同时发射与反射可重构智能表面（STAR-RIS）架构，该架构具有动态切换的有源和无源元件，以支持联合定位、通信与无线功率传输（WPT）。我们首先采用基于平行因子分析与交替最小二乘（PARAFAC-ALS）的张量分解方法，该方法可解耦基站（BS）-可重构智能表面（RIS）以及RIS-用户信道，从而实现低开销的信道获取。基于此，我们构建系统能效（EE）最大化问题，约束条件包括通信用户的频谱效率（SE）要求、感知信号干扰加噪声比约束，以及能量收集用户的非线性能量收集要求。该优化问题为非凸问题，因为发射功率分配、STAR-RIS系数以及有源/无源模式分配在目标函数和约束中紧密耦合。我们通过在两个子问题间交替迭代解决该问题，并通过分数规划、逐次凸近似以及作为初始化步骤的多种子贪心策略求解这些子问题。数值结果表明，与全无源/有源架构相比，有选择性激活STAR-RIS的一小部分精心挑选的元件可实现1.5至3倍的能效提升，同时满足通信、感知和功率传输要求。
英文摘要
We propose a hybrid simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) architecture with dynamically switched active and passive elements to support joint localization, communication, and wireless power transfer (WPT). We first pursue a parallel factor analysis with the alternating least squares (PARAFAC-ALS)-based tensor decomposition approach that decouples the base station (BS)-reconfigurable intelligent surface (RIS) and RIS-user channels, thereby enabling low-overhead channel acquisition. Based on this, we formulate a system energy efficiency (EE) maximization problem, subject to the spectral efficiency (SE) requirements of communication users, sensing signal-to-interference-plus-noise ratio constraints, and the nonlinear energy harvesting requirements of energy-harvesting users. The optimization problem is nonconvex since the transmit power allocation, STAR-RIS coefficients, and active/passive mode assignments are tightly coupled in both the objective and constraints. We address this issue by alternating between two subproblems, and solving them via fractional programming, successive convex approximation and a multi-seed greedy strategy employed as an initialization step. Numerical results demonstrate that selectively activating a small, well-chosen subset of STAR-RIS elements achieves 1.5 to 3 times EE improvements compared with fully passive/active architectures, while satisfying communication, sensing, and power-transfer requirements.
CommentsAccepted for publication in IEEE Transactions on Wireless Communications, Aug. 2026
