To detect message collisions in multichannel spaced‐based automatic identification system (AIS), a dynamic complex symmetric fast independent component analysis (CSFICA) and feature‐based collision detection method is proposed in this paper. A fast and stable blind source separation algorithm, dynamic CSFICA, is utilized to separate signals dynamically and improve the signal‐to‐interference ratio (SIR) in each channel. A frequency and phase offset insensitive feature detection algorithm is used to calculate the test statistics in each channel. The false alarm is suppressed by applying the arithmetic to geometric mean (AGM) method, and test statistics of channels with sufficient signal quality are extracted to detect the preamble. Simulation results show that the proposed algorithm outperforms the reference feature detection algorithm under collision conditions and is insensitive to the SIR. The proposed algorithm is more resistant to false alarm caused by signal (FAS) than the differential correlation (DC) algorithm.

中文翻译：

---

多通道空基AIS中的动态CSFICA和基于特征的碰撞检测方法

为了在多通道基于空间的自动识别系统（AIS）中检测消息冲突，提出了一种动态复杂对称快速独立成分分析（CSFICA）和基于特征的冲突检测方法。快速稳定的盲源分离算法动态CSFICA用于动态分离信号，并改善每个通道中的信号干扰比（SIR）。频率和相位偏移不敏感特征检测算法用于计算每个通道中的测试统计量。通过将算术应用于几何均值（AGM）方法来抑制误报，并提取具有足够信号质量的信道的测试统计信息以检测前导。仿真结果表明，该算法在碰撞条件下优于参考特征检测算法，对SIR不敏感。所提出的算法比差分相关（DC）算法更能抵抗信号（FAS）引起的虚假警报。