Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RISs) are being explored for sixth-generation (6G) wireless networks. A promising configuration for their deployment is within cell-free massive multiple-input multiple-output (MIMO) systems. However, despite the advantages that STAR-RISs could bring, challenges such as electromagnetic interference (EMI) and phase errors may lead to significant performance degradation. In this paper, we investigate the impact of EMI and phase errors on STAR-RIS-assisted cell-free massive MIMO systems and propose techniques to mitigate these effects. We introduce a tailored projected gradient descent (GD) algorithm for STAR-RIS coefficient matrix design by minimizing the local channel estimation normalized mean square error (NMSE). We also derive the novel closed-form expressions of the uplink and downlink spectral efficiency (SE) to analyze system performance with EMI and phase errors, in which fractional power control methods are introduced for performance improvement. The results reveal that the projected GD algorithm can effectively tackle EMI and phase errors to improve estimation accuracy and compensate for performance degradation with nearly 30% NMSE improvement and over 10% SE improvement. Moreover, increasing the number of access points (APs), antennas per AP, and STAR-RIS elements can also improve SE performance. However, the advantages of employing STAR-RIS are reduced when EMI and phase errors are severe. Notably, compared to conventional RISs, the incorporation of STAR-RIS in the proposed system yields better performance and presents less performance degradation in highly impaired environments.

翻译：同时透射与反射可重构智能表面（STAR-RIS）正被探索用于第六代（6G）无线网络。其在无蜂窝大规模多输入多输出（MIMO）系统中的部署是一种前景广阔的配置。然而，尽管STAR-RIS可能带来诸多优势，但电磁干扰（EMI）和相位误差等挑战可能导致显著的性能下降。本文研究了EMI和相位误差对STAR-RIS辅助的无蜂窝大规模MIMO系统的影响，并提出了缓解这些影响的技术。我们通过最小化局部信道估计的归一化均方误差（NMSE），为STAR-RIS系数矩阵设计引入了一种定制的投影梯度下降（GD）算法。我们还推导了上行链路和下行链路频谱效率（SE）的新颖闭式表达式，以分析存在EMI和相位误差时的系统性能，其中引入了分数功率控制方法以提升性能。结果表明，投影GD算法能有效应对EMI和相位误差，提高估计精度并补偿性能下降，实现了近30%的NMSE提升和超过10%的SE提升。此外，增加接入点（AP）数量、每个AP的天线数以及STAR-RIS单元数量也能改善SE性能。然而，当EMI和相位误差严重时，采用STAR-RIS的优势会减弱。值得注意的是，与传统RIS相比，在所提出的系统中集成STAR-RIS能获得更好的性能，并且在高度受损环境中表现出更少的性能下降。