均衡算法的定点DSP实现研究  

Methods for Equalization Implementing on Fixed-Point DSP

作　　者：刘晨[1] 李庆[1] 王大鸣[1] 

机构地区：[1]信息工程大学信息工程学院,河南郑州450002

出　　处：《信息工程大学学报》2008年第4期447-450,共4页Journal of Information Engineering University

摘　　要：简要介绍了LMS和RLS两类基本的算法,深入分析了RLS算法的发散问题,讨论了数值精度与算法稳定性的关系。在均衡算法的定点DSP实现中,研究了Q格式数的基本原理和运算方法,给出了LMS与RLS实现时的具体精度需求,为实际工程实现提供了有价值的参考。仿真表明,当精度足够高时,均衡算法的性能将趋于恒定。RLS and LMS are two basic algorithms for equalization. This paper introduces these two algorithms and specifically discusses the volatilization phenomenon of RLS from a new angle, as well as the relation between the value precision and the algorithm stability. While implementing equalization on fixed-point DSP, the mathematic theory and operation methods of Q format are studied, and the precision requirements of LMS and RLS implementing are presented in detail, which is quite valuable for practical application. Simulation proves that the algorithm performance would get close to a constant when the iterating precision is high enough.

关 键 词：均衡 RLS DSP Q格式数 

分 类 号：TN911[电子电信—通信与信息系统]