In this paper, we consider a semantic-aware reconfigurable intelligent surface (RIS)-assisted wireless network, where multiple semantic users (SUs) simultaneously transmit semantic information to an access point (AP) by using the non-orthogonal multiple access (NOMA) method. The SUs can reshape their traffic demands by modifying the semantic extraction factor, while the RIS can reconfigure the channel conditions via the passive beamforming. This provides the AP with greater flexibility to decode the superimposed signals from the SUs. We aim to minimize the system's overall energy consumption, while ensuring that each SU's traffic demand is satisfied. Hence, we formulate a joint optimization problem of the SUs' decoding order and semantic control, as well as the RIS's passive beamforming strategy. This problem is intractable due to the complicated coupling in constraints. To solve this, we decompose the original problem into two subproblems and solve them by using a series of approximate methods. Numerical results show that the joint traffic reshaping and channel reconfiguration scheme significantly improves the energy saving performance of the NOMA transmissions compared to the benchmark methods.

翻译：本文研究一种语义感知的可重构智能表面(RIS)辅助无线网络，其中多个语义用户(SU)通过非正交多址(NOMA)方式同时向接入点(AP)传输语义信息。SU可通过调整语义提取因子来重塑其流量需求，而RIS则能通过无源波束赋形重构信道条件。这为AP解码来自SU的叠加信号提供了更大的灵活性。我们的目标是在满足每个SU流量需求的前提下，最小化系统总能耗。为此，我们构建了关于SU解码顺序与语义控制以及RIS无源波束赋形策略的联合优化问题。该问题因约束条件间的复杂耦合而难以直接求解。为解决此问题，我们将原问题分解为两个子问题，并采用一系列近似方法进行求解。数值结果表明，与基准方法相比，联合流量重塑与信道重构方案能显著提升NOMA传输的节能性能。