本文提出一种基于Fisher线性判别(FLD)的多用户检测器。该检测器将信道的时变特性隐含在学习算法得出的最优投影方向矢量中,因而适合各种信道环境,而不必知道噪声和干扰的类型;它需要训练序列,除了有用信号的扩频序列外,不需要其他用户的扩频序列;算法简单、实现复杂度低。本文对所提出的检测器性能进行了仿真实验,给出了FLD检测器与MMSE检测器和SVM检测器的性能比较。 This paper presents a multiuser detector based on Fisher linear discriminant (FLD). This detector implies the time-varying nature of the channel in the optimal projection direction vector derived from the learning algorithm and is therefore suitable for a wide range of channel environments without having to know the type of noise and interference; it requires a training sequence, in addition to the expansion of the useful signal Frequency sequence, does not require other users of the spread spectrum sequence; algorithm is simple, low complexity. In this paper, we simulate the proposed detector performance and compare the performance of FLD detector with MMSE detector and SVM detector.