变步长的LMS盲多用户检测

LMS Algorithm with Variable Step Size for Blind Multiuser Detection

提出了一种变步长 LMS自适应盲多用户检测算法 ,克服了定步长自适应算法的收敛步长与输入信号自相关矩阵特征值大小有关的缺点 ,使算法的步长能够随输入信号矢量的变化而变化 ,提高了算法的收敛性能 ,使其更加符合实际工作的需要。并通过计算机模拟仿真 ,比较了变步长算法和固定步长算法的性能。

A multiuser detection method based on time-varying step size LMS algorithm is proposed. This algorithm can overcome the defect of constant step size LMS algorithm whose step size is related to the eigenvalue distribution of the received data autocorrelation matrix and can make the step size change with the input signal vector. The evaluation of the performance of convergence meets the practical need. By simulation, the performance of time-varying step size algorithm is compared with that of constant step size algorithm.