AI 中文总结
研究中高频段地面与卫星下行共存问题，提出将基于MDL的模型阶数选择与MUSIC相结合获取匿名方向和功率信息用于波束设计，经对零波束形成器减少干扰，分析表明估计精度随感知SNR提高，射线追踪结果显示能降低干扰噪声比。
AI 中文摘要
当中频段地面基站对非地面接收机及其上行传输的先验信息有限时，地面与卫星共存具有挑战性。本文研究非相干受害方感知和干扰对零，在发射波形未知时观察上行感知快照。对于单个受害方，广义似然比检验简化为样本协方差的主特征向量估计器。对于多个受害方，将基于MDL的模型阶数选择与MUSIC相结合，以恢复匿名方向和功率信息，用于波束设计。这些估计用于对零波束形成器，在减少对检测到的受害方泄漏的同时保留地面链路。在大矩阵情况下分析了单受害方估计器，表明估计精度随感知SNR提高。特定地点的射线追踪结果表明，非地面网络干扰噪声比显著降低，而地面网络信号与干扰加噪声比仅有适度下降。
英文摘要
Terrestrial--satellite coexistence in the upper mid-band is challenging when a terrestrial base station has limited prior information about non-terrestrial receivers and their uplink transmissions. This paper studies noncoherent victim sensing and interference nulling, where uplink sensing snapshots are observed while the transmitted waveform is treated as unknown. For a single victim, we show that the generalized likelihood-ratio test reduces to a principal-eigenvector estimator of the sample covariance. For multiple victims, we combine MDL-based model-order selection with MUSIC to recover anonymous direction and power information that is sufficient for beam design without pilot knowledge or user identities. These estimates are then used in a nulling beamformer that preserves the intended terrestrial link while reducing leakage toward detected victims. We further analyze the single-victim estimator in the large-matrix regime and show that estimation accuracy improves with sensing SNR, which reveals the observed interplay between path loss and estimation quality. Site-specific ray-tracing results show significant reduction of NTN INR with only modest degradation of TN SINR.
CommentsAccepted by IEEE SPAWC 2026
