摘要
提出一种准完全重构(NPR)余弦调制滤波器组(CMFBs)的最优化设计方法。通过使用最优化方法设计原型滤波器,使其过渡带具有平方根余弦滚降(SRCR)特性。这种原型滤波器能够更好地满足设计NPRCMFBs所需的平坦性条件。CMFBs设计问题的核心被归结为一个凸极小极大最优化问题,可以用二阶锥规划(SOCP)算法进行求解。该设计方法不需要用户提供原型滤波器的初始值,并且在滤波器阶数和过渡带宽一定的情况下能获得更小的重构误差和更大的阻带衰减。
An optimization method for designing nearly perfect reconstruction(NPR) cosine modulated filter banks(CMFBs) is proposed. The prototype filters used by the CMFBs have a Square -Root cosine -rolloff (SRCR) transition band due to the optimization design method. It is shown that the flatness condition re- quired for NPR CMFBs can be better fulfilled by using a prototype filter with a SRCR transition band. The design problem of the CMFBs is formulated as a convex mini - max optimization problem, and it can be solved by second- order cone programming(SOCP). The advantage of the proposed method is that it can obtain a lower reconstruction error and a higher stopband attenuation under a certain assigned filter length and transition band without requiring a user - supplied initial value of the prototype filter.
出处
《电讯技术》
北大核心
2010年第2期33-37,共5页
Telecommunication Engineering
基金
军队重点资助项目(7130320)
河南基础与前沿项目(082300413205)~~
关键词
准完全重构
余弦调制滤波器组
二阶锥规划
平方根余弦滚降
nearly sperfect reconstruction (NPR)
cosine modulated filter banks (CMFBs)
second - order cone programming(SOCP)
square - root cosine rolloff(SRCR)