一种片率增量SVM多用户检测器算法仿真研究

码分多址(CDMA)系统中的多用户检测(MUD)实质上是一盲信源分离问题,通过在上行链路接收端获取采样的混合信号,然后根据扩频码特征及波束形成算法进行初步预分离,进而将片率信号作为支持向量机(SVM)的训练信号,在SVM中将所有用户通过1-1模式构造多目标分类器,通过类增量学习方式进行目标信号分类,最终获取期望用户的检测值。文中最后给出了仿真结果及分析比较,验证了算法的可行性。

Chip-Increment Support Vector Machine-based Algorithm for Multi-user Detection

Multi-user detection (MUD) in code-division multi access (CDMA) system is in essence of a problem of blind source separation (BSS). This paper firstly presents a pre-separation framework based on spreading code for mixed signal at uplink receiver, and then formulates a multi-object classifier through one-against-one mode. A chip-increment learning algorithm is implemented in classifier to obtain the desired sign information. Finally the simulation is given to support the theoretical result.