中文摘要：

恒模算法（CMA）是一种广泛应用于阵列处理、均衡、多用户检测中的盲算法。最小二乘恒模算法（LSCMA）由于其全局收敛性及稳定性受到关注。本文研究块衰落信道下基于训练序列辅助的LSCM算法的性能。本文的研究基于同步CDMA系统,系统中每个用户经历独立的平坦瑞利衰落信道。在发送端,为了在接收端获得数据的信号子空间,目标用户的第一帧中插入少量的训练序列。通过信号子空间信息的辅助,本文提出的训练序列非周期辅助最小二乘恒模（TSALSCM -NP,Training Sequence Aided Least Squares Constant Modulus non-periodically）算法在提高少许复杂度的基础上能够获得优于LSCM算法的检测性能。

英文摘要：

The constant modulus algorithm（CMA） is a blind algorithm. It is applied in the array processing,equalization and multiuser detection widely. The least square CMA is a popular CMA because of its global convergence and stability. In this paper, a training sequence aided least square constant modulus algorithm in block flat rayleigh fading CDMA channels is discussed. A synchro- nous code division multiple access system where each user undergoes independent flat Rayleigh fading is considered, and where training sequences are inserted into the first data stream of the desired user in order to achieve the signal subspace of the data symbols. By the assistance of the signal subspace information, the proposed TSA-LSCM-NP （Training Sequence Aided Least Squares Constant Modulus non-periodically） algorithm with only a few training sequence can achieve better detection performance than the LSCM algorithm.