一种改进的自适应恒模算法收敛特性的研究

The Study on Convergence Property of A Modified CMA

常数模算法是数字通信中经常使用的一种盲均衡技术,但它的收敛速率非常缓慢,为了改善收敛特性,本文就4种典型的误差方程进行了分析、仿真,发现误差方程对CMA的收敛速度以及调整过程中是否能收敛到全局最小点都有影响,并在此基础上提出了一种改进的误差方程。仿真结果表明,改进后的误差方程具有较好的收敛速度,而且总能收敛到全局最小点。

The Constant Modulus Algorithm (CMA), which is the most commonly used blind channel e-qualization algorithm in digital communications, exhibits very slow convergence rate. In this paper, in order to improve the convergence property, four typical error equations are analyzed and simulated under different channel conditions. Analysis and computer simulation results demonstrate that the error equations not only determine the convergence rate of the adaptation process, but also have effect on the convergence to a global or a local minimum. Based on this analysis we present a modified error equation. Simulation shows that this error equation will have better convergence property and will always lead the equalizer to converge to a global minimum.