近似完全重构的DFT调制滤波器组迭代设计方法

An iterative design algorithm of near-perfect-reconstruction DFT modulated filter banks

针对传统离散傅里叶变换(DFT)调制滤波器组设计算法存在优化规模大、求解较为复杂的缺点,提出一种近似完全重构的DFT调制滤波器组迭代设计方法。该方法通过凸优化松弛将高度非线性非凸的优化问题转化成凸的优化问题,迭代化地优化迭代增量,使得原型滤波器系数的重构误差达到较低的水平。仿真结果表明,本方法设计的DFT调制滤波器组重构误差更低,整体性能更好。

A new algorithm is proposed to design near-perfect-reconstruction DFT modulated filter banks due to the large optimized scale and the poor performance of the traditional algorithms.In each iteration,the nonlinear nonconvex constraint function of the original optimization problem is transformed into a convex optimization problem by the convex optimization slack.This problem is easier to be solved.Then the iterative increment is iteratively optimized such that the prototype filter coefficient with lower reconstruction error can be obtained.The simulation results show that the DFT modulation filter group obtained by the new algorithm has lower reconstruction error and better overall performance.