AI 中文总结
研究中高频段ISAC Systems多频段多基地目标定位问题，提出SCOPE算法，利用多视图一致性抑制栅瓣模糊，发射机抑制旁瓣，接收机采用特定峰值选择和离网格细化，仿真显示该算法在不同SNR下有良好定位精度。
AI 中文摘要
在7-24GHz的FR3频段中的多频段多基地综合感知与通信（ISAC），可通过虚拟宽带和空间分集实现高分辨率定位。然而，频率各向异性会使目标在非连续频段上的散射去相关，而大的频段间频率间隔会产生严重的栅瓣，导致持续的虚假峰值。我们提出了旁瓣控制离网格轮廓估计（SCOPE），一种稳健的定位算法，利用分布式接收器和频段之间的多视图一致性来抑制栅瓣模糊性。在发射机处，迭代极小极大预编码器抑制区域外旁瓣，以减少粗略似然图中的虚假峰值。在接收机处，SCOPE采用基于Top-K抑制的峰值选择的轮廓似然性，以避免陷入虚假盆地，随后进行无导数离网格细化。仿真表明，SCOPE在-5dB SNR时以90%的概率实现亚米级定位，在25dB SNR时实现3mm均方根误差（RMSE）。
英文摘要
Multiband multistatic integrated sensing and communication (ISAC) in fragmented FR3 bands (7-24 GHz) enables high resolution localization via virtual wideband and spatial diversity. However, frequency anisotropy decorrelates target scattering across non-contiguous bands, while large inter-band frequency gaps generate severe grating lobes that create persistent ghost peaks. We propose sidelobe-controlled off-grid profile estimation (SCOPE), a robust localization algorithm that exploits multi-view consistency across distributed receivers and frequency bands to suppress grating-lobe ambiguities. At the transmitter, an iterative minimax precoder suppresses out-of-region sidelobes to reduce false peaks in the coarse likelihood map. At the receiver, SCOPE employs profile likelihood with Top-K inhibition-based peak selection to avoid trapping in ghost basins, followed by derivative-free off-grid refinement. Simulations demonstrate that SCOPE achieves sub-meter localization with 90% probability at -5 dB SNR and 3 mm root mean square error (RMSE) at 25 dB SNR.
CommentsAccepted for presentation at the 2026 IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC)
