【摘要】 基于信号稀疏表示理论,提出一种离散系数FIR滤波器设计方法。在加权最小二乘准则下,首先将原始设计问题转换为半定规划问题。为了有效控制硬件实现开销,借鉴信号稀疏表示理论,在目标函数中引入加权项,并在此基础之上,利用随机采样技术获得最终设计结果。仿真结果表明该方法所设计的离散系数滤波器,在控制加法器个数的同时,能有效减少因系数量化所导致的性能损失。与传统方法相比,在阶数较高的情况下,该方法依然能取得较好效果。更多还原

【Abstract】 Base on the sparse signal representation theory,a method is proposed to design FIR filters with discrete coefficients.The original design problem is transformed to a semidefinite programming(SDP)problem under weighted least-squares criterion.Inspired by the sparse signal representation theory,the proposed method introduces an auxiliary term in the objective function to control the cost of hardware implementation.Final solutions are obtained by random sampling technique.Simulation results show that FIR filters obtained by the proposed algorithm can efficiently control the number of adders and reduce the performance loss incurred by the quantification of filter coefficients.Compared with traditional design methods,the proposed method can achieve better performance when the filter order is high.更多还原