Abstract: Digital shaping filters are important part of digital baseband communication systems. It is designed for limiting the spectrum of signal with introducing Inter Symbol Interference （ISI） as little as possible. Traditional way to design filters can be divided into two steps：Estimating the orders and computing the ripple coefficient. However, current methods for estimating the orders are only available for Finite Impulse Response （FIR） filters. They are not available for digital shaping filters, a special kind of FIR. In order to solve this problem, an algorithm which combines the two steps of designing digital filters is proposed. The experiments show that the proposed algorithm can get better results in estimating the orders and computing the ripple coefficient.

Key words: digital shaping filters, genetic algorithm, filters order, ripple coefficient

摘要： 数字成形滤波器是数字基带通信系统的重要组成部分。基带信号的频谱范围较宽，频带拥挤问题日益突出，成形滤波技术可以在消除码间干扰（ISI）的前提下，压缩信号频带，提高频谱的利用率。传统的数字滤波器设计包含阶数估计和波纹系数计算两个步骤。而阶数估计方法仅对普通的有限脉冲响应滤波器（FIR）有效，对成形滤波器阶数估计的结果不能达到最优。针对这一问题，将遗传算法的阶数估计和系数计算两个步骤结合起来设计数字成形滤波器，简化了算法复杂度。仿真对比结果表明，该算法在阶数估计和波纹系数计算方面都可以得到更优的结果。

关键词: 数字成形滤波器, 遗传算法, 滤波器阶数, 波纹系数