基于并行约束延迟LMS的高速FPGA信息采集
|
摘要
在实际的信号传输过程中,原始信号受噪声的影响,噪声源特性未能在不同的环境中提前预测。传统的滤波器系数在设计的阶段已被固定,不可再根据应用调整。传统数字信号处理器实现的自适应滤波器处理速度低、抗干扰性差,导致FPGA信息采集速度比较慢。为了自动消除系统中的非目标带谐波,自适应滤波器根据随机信号的统计描述和模型来预测信号趋势。同时,为了降低资源消耗,提高运行速度,提出了一个并行约束延迟最小均方算法,采用数据转移的方法实现步长因子的更新,理论分析和实验结果表明,与传统方法相比,该方法对于FPGA的高速信息采集方面具有更大的优越性。
In the process of the actual signal transmission, the original signal is affected by noise, the noise characteristics cannot be predicted in advance in different environment. The traditional filter coefficients have been fixed in the design stage and cannot be adjusted according to the application. The adaptive filter implemented by traditional digital signal processor has low processing speed and poor anti-interference performance, which results in the slow speed of FPGA information acquisition. To automatically eliminate the target zone system harmonic, adaptive filter based on random signals of statistical description and model to predict trends, at the same time, in order to reduce resource consumption and improve the running speed, it puts forward a parallel constrained delay least mean square algorithm, and adopts the method of data transfer implementation step length factor. The theoretical analysis and experimental results show that compared with traditional methods, the method for FPGA has greater advantages in the respect of high-speed data collection.
In the process of the actual signal transmission, the original signal is affected by noise, the noise characteristics cannot be predicted in advance in different environment. The traditional filter coefficients have been fixed in the design stage and cannot be adjusted according to the application. The adaptive filter implemented by traditional digital signal processor has low processing speed and poor anti-interference performance, which results in the slow speed of FPGA information acquisition. To automatically eliminate the target zone system harmonic, adaptive filter based on random signals of statistical description and model to predict trends, at the same time, in order to reduce resource consumption and improve the running speed, it puts forward a parallel constrained delay least mean square algorithm, and adopts the method of data transfer implementation step length factor. The theoretical analysis and experimental results show that compared with traditional methods, the method for FPGA has greater advantages in the respect of high-speed data collection.
引文
[ 1 ]雷宇,靳宝全,王云才,等.LMS自适应滤波器模块化设计及其FPGA实现[J].电子器件,2017,40(2):390-395.
[ 2 ]王新,闫文源,王丹璐.复杂输入下自适应滤波器的过渡过程处理方法[J].电机与控制学报,2017,21(3):97-103.
[ 3 ]汤书森,苟煜春,张文强,等.用FPGA和改进的LMS算法实现自适应滤波器[J].单片机与嵌入式系统应用,2018,18(2):18-20.
[ 4 ]NARIMANI G,MARTIN P A,TAYLOR D P.Spectral analysis of fractionally-spaced MMSE equalizers and stability of the LMS algorithm[J].IEEE Transactions on Communications,2018,66(4):1675-1688.
[ 5 ]CHAUDHARY N I,MANZAR M A,RAJA M A Z.Fractional Volterra LMS algorithm with application to Hammerstein control autoregressive model identification[J].Neural Computing & Applications,2018(3):1-14.
[ 6 ]GOLDSTEIN J S,REED I S,SCHARF L L.A multistage representation of the Wiener filter based on orthogonal projections[J].IEEE Trans Inf Theory,2007,44(7):2943-2959.
[ 7 ]GARDNER W A.Cyclic Wiener filtering:Theory and methods[J].IEEE Transactions on Communications,1993,40(1):151-163.
[ 8 ]沈希忠,叶秋泽.基于维纳滤波的超声增强实现方法[J].数据采集与处理,2018,33(3):71-76.
[ 9 ]温金玉,宣士斌,肖石林,等.基于小波变换与改进维纳滤波的SAR图像分割[J].广西民族大学学报(自然科学版),2018,24(2):75-84.
[10]柴万东,张立萍,张迪.延迟LMS(DLMS)自适应滤波器的FPGA设计与实现[J].赤峰学院学报(自然科学版),2016(2):1-4.
[11]高金定.一种改进的延迟型LMS自适应滤波器及其FPGA实现[J].核电子学与探测技术,2012,32(12):1450-1453.
[12]马上.实时LMS信道模拟器设计与实现[D].成都:电子科技大学,2006.
[13]冯丁.基于延迟矢量跟踪环的变步长LMS空时滤波算法[J].自动化与仪表,2017(12):5-9.
[14]郭莹,孟彩云.改进的LMS自适应时延估计方法[J].沈阳工业大学学报,2013,35(2):200-203.
[15]郇战,戴永惠,王振海,等.基于FxLMS算法和预测滤波器的数字耳机降噪研究[J].常州大学学报(自然科学版),2015,27(2):64-67.
[16]董祥雷,袁上策,王玲,等.一种改进LMS算法的高速FPGA实现[J].电子信息对抗技术,2015(3):72-75.
[ 2 ]王新,闫文源,王丹璐.复杂输入下自适应滤波器的过渡过程处理方法[J].电机与控制学报,2017,21(3):97-103.
[ 3 ]汤书森,苟煜春,张文强,等.用FPGA和改进的LMS算法实现自适应滤波器[J].单片机与嵌入式系统应用,2018,18(2):18-20.
[ 4 ]NARIMANI G,MARTIN P A,TAYLOR D P.Spectral analysis of fractionally-spaced MMSE equalizers and stability of the LMS algorithm[J].IEEE Transactions on Communications,2018,66(4):1675-1688.
[ 5 ]CHAUDHARY N I,MANZAR M A,RAJA M A Z.Fractional Volterra LMS algorithm with application to Hammerstein control autoregressive model identification[J].Neural Computing & Applications,2018(3):1-14.
[ 6 ]GOLDSTEIN J S,REED I S,SCHARF L L.A multistage representation of the Wiener filter based on orthogonal projections[J].IEEE Trans Inf Theory,2007,44(7):2943-2959.
[ 7 ]GARDNER W A.Cyclic Wiener filtering:Theory and methods[J].IEEE Transactions on Communications,1993,40(1):151-163.
[ 8 ]沈希忠,叶秋泽.基于维纳滤波的超声增强实现方法[J].数据采集与处理,2018,33(3):71-76.
[ 9 ]温金玉,宣士斌,肖石林,等.基于小波变换与改进维纳滤波的SAR图像分割[J].广西民族大学学报(自然科学版),2018,24(2):75-84.
[10]柴万东,张立萍,张迪.延迟LMS(DLMS)自适应滤波器的FPGA设计与实现[J].赤峰学院学报(自然科学版),2016(2):1-4.
[11]高金定.一种改进的延迟型LMS自适应滤波器及其FPGA实现[J].核电子学与探测技术,2012,32(12):1450-1453.
[12]马上.实时LMS信道模拟器设计与实现[D].成都:电子科技大学,2006.
[13]冯丁.基于延迟矢量跟踪环的变步长LMS空时滤波算法[J].自动化与仪表,2017(12):5-9.
[14]郭莹,孟彩云.改进的LMS自适应时延估计方法[J].沈阳工业大学学报,2013,35(2):200-203.
[15]郇战,戴永惠,王振海,等.基于FxLMS算法和预测滤波器的数字耳机降噪研究[J].常州大学学报(自然科学版),2015,27(2):64-67.
[16]董祥雷,袁上策,王玲,等.一种改进LMS算法的高速FPGA实现[J].电子信息对抗技术,2015(3):72-75.