数字信道化器中高阶精确重构滤波器组设计方法与量化分析

数字信道化技术广泛应用于新一代卫星通信系统的星上处理平台中,而精确重构滤波器组的设计是其中的关键技术。针对现有滤波器设计方法通道数较低,难以满足高分辨率数字信道化器需求的问题,提出了一种基于频率取样法和最小均方误差逼近的高阶精确重构原型滤波器的设计方法。该方法首先利用频率取样法求解一个近似精确重构原型滤波器;然后在最小均方误差准则下,用一个由两通道无损网格建模的滤波器对该近似精确重构原型滤波器进行最优拟合。数值实验表明,与Parks-Mc Clellan(PM)算法相比,该方法能设计得到更高通道数的滤波器组;并且原理简单,设计复杂度很低。滤波器性能的量化分析结果表明,综合考虑多相分量长度与量化位宽的设计思想,能够平衡实现复杂度与性能指标,对于实际系统具有重要的参考价值。

Design Method and Quantization Analysis of High Order Perfect Reconstruction Filterbank on Digital Channelizer

Digital channelization technique is widely used in the on-board processing platform for the next generation of satellite communication systems, in which the design of perfect reconstruction prototype filter is the key technique. Filters designed by existing schemes usually have relatively small number of subchannels, which is difficult to meet the demand for high resolution digital channelizers. Aiming at this problem, a new method based on frequency sampling as well as least mean square error is proposed for designing high order perfect reconstruction prototype filters. This method first use the frequency sampling technique to obtain a near perfect reconstruction prototype filter, then in the criterion of least mean square error, a filter modeled by two-channel lossless lattice is optimized to approximate the near perfect reconstruction prototype filter. Numerical results show that compared with the Parks-McClellan (PM) algorithm, this method can design filterbanks with larger number of subchannels in a simple way and with low design complexity. Moreover, quantization analysis on filter performance indicate that considering the length of polyphase component and the quantization width together can make a balance between implementation costs and quantitative criteria, which is of great reference value for practical systems.