Non-orthogonal multiple access (NOMA) has been proposed to raise the spectral efficiency, which also brought new security threats. This letter studies the channel gain feedback falsification (CGFF) attacks of malicious users in NOMA. Since a slight falsification on channel gain feedback can still seriously damage the efficiency and security of NOMA, the detection of malicious users in NOMA is an important problem that is difficult to be solved by the traditional methods. Specific to the sensitivity of the eigenvalues to the sparse and slight anomaly, this letter analyzes the empirical spectral distribution (e.s.d.) of eigenvalues of channel gains feedback with and without CGFF attacks based on random matrix theory. On this basis, two lightweight detection schemes are proposed to detect CGFF attacks. The simulations prove the effectiveness of our proposed methods.

中文翻译：

---

基于频谱分析的非正交多址恶意用户检测

非正交多址（NOMA）被提出来提高频谱效率，这也带来了新的安全威胁。这封信研究了 NOMA 中恶意用户的信道增益反馈伪造（CGFF）攻击。由于对信道增益反馈的轻微篡改仍会严重损害 NOMA 的效率和安全性，因此 NOMA 中恶意用户的检测是传统方法难以解决的重要问题。具体到特征值对稀疏和轻微异常的敏感性，本文基于随机矩阵理论分析了有和没有CGFF攻击的信道增益反馈特征值的经验谱分布（esd）。在此基础上，提出了两种轻量级检测方案来检测CGFF攻击。