变步长神经网络盲均衡算法的研究
|
摘要
均衡技术是数字通信系统中克服码间干扰的有效方法，其中盲均衡是均衡技术的最新发展，它不需要借助训练序列就能自适应调整均衡器的抽头系数，因此得到广泛应用。本文对盲均衡技术主要分支之一——神经网络盲均衡算法进行了深入研究，分析了其在收敛性能上存在的缺点，将变步长思想引入神经网络盲均衡算法中，提出了两种改进算法，并通过计算机仿真对收敛性能进行了验证。
本文所做的主要工作有：
(1)归纳总结了神经网络和盲均衡的基础理论，推导了BP算法，并分析了其缺陷。
(2)分析研究了传统神经网络盲均衡算法由于采用固定步长，使得收敛速度和收敛精度之间相互制约。针对这一情况，提出将变步长理论引入神经网络盲均衡算法中。
(3)分析了神经网络盲均衡算法中剩余误差的变化规律，指出将其作为步长控制因子的可行性。并通过对剩余误差进行相应变换，提出了两种变步长神经网络盲均衡算法，分析了改进算法的参数选取原则，推导了算法形式，通过计算机仿真验证了改进算法在收敛性能上有一定的提高。
Equalization technology is an effective method which can remove intersymbol interference (ISI) in digital communication. Blind equalization is the latest development. It has abroad use because it can adaptively equalize without training sequence. This paper deduces BP algorithm in detail, analyzes blind equalization based on neural network, propose two improved algorithms by putting variable step-size theories into neural network blind equalization algorithm. The convergence performance of the improved algorithms is illustrated by computer simulations. The main works of this paper can be summarized as follows:
(1) In this paper, the fundamental theory of neural network and blind equalization is summarized. BP algorithm is deduced in detail, and its drawback is analyzed.
(2 ) This paper analyzes the convergence performance of the traditional blind equalization algorithm. Convergence speed and residual error become a contradiction due to the fixed step-size. In order to solve the contradiction, the paper proposes two improved algorithms by putting variable step-size theory into neural network blind equalization algorithm.
(3) This paper analyzes residual error of neural network blind equalization algorithm, bring forward the feasibility that residual error
本文所做的主要工作有：
(1)归纳总结了神经网络和盲均衡的基础理论，推导了BP算法，并分析了其缺陷。
(2)分析研究了传统神经网络盲均衡算法由于采用固定步长，使得收敛速度和收敛精度之间相互制约。针对这一情况，提出将变步长理论引入神经网络盲均衡算法中。
(3)分析了神经网络盲均衡算法中剩余误差的变化规律，指出将其作为步长控制因子的可行性。并通过对剩余误差进行相应变换，提出了两种变步长神经网络盲均衡算法，分析了改进算法的参数选取原则，推导了算法形式，通过计算机仿真验证了改进算法在收敛性能上有一定的提高。
Equalization technology is an effective method which can remove intersymbol interference (ISI) in digital communication. Blind equalization is the latest development. It has abroad use because it can adaptively equalize without training sequence. This paper deduces BP algorithm in detail, analyzes blind equalization based on neural network, propose two improved algorithms by putting variable step-size theories into neural network blind equalization algorithm. The convergence performance of the improved algorithms is illustrated by computer simulations. The main works of this paper can be summarized as follows:
(1) In this paper, the fundamental theory of neural network and blind equalization is summarized. BP algorithm is deduced in detail, and its drawback is analyzed.
(2 ) This paper analyzes the convergence performance of the traditional blind equalization algorithm. Convergence speed and residual error become a contradiction due to the fixed step-size. In order to solve the contradiction, the paper proposes two improved algorithms by putting variable step-size theory into neural network blind equalization algorithm.
(3) This paper analyzes residual error of neural network blind equalization algorithm, bring forward the feasibility that residual error
引文
[01] Proakis J G. Digital communications (3rd Edition). McGraw Hill, 1995.
[02] 张贤达，保铮．通信信号处理．北京：国防工业出版社，2000．
[03] 邹谋炎．反卷积和信号复原．北京：国防工业出版社，2001．
[04] 沈风麟，陈和晏．生物医学随机信号处理．合肥：中国科学技术大学出版社，1999．
[05] 周耀华，王凯仁．数字信号处理．上海：复旦大学出版社，1993．
[06] 何振亚．自适应信号处理．北京：科学出版社，2002．
[07] 张贤达．现代信号处理(第二版)．北京：清华大学出版社，2003．
[08] L. tong, J. K. tugnait. Signle-user channel estimation and equalization. IEEE Signal processing Magazine, 2000: 17～28.
[09] Y. Sato. A method of self-recovering equalization for multiple amplitude modulation schemes. IEEE Trans. on communication, 1975, 23: 679～682.
[10] A. Benveniste, M. Goursat, G. Ruget. Robust Identification of a Nonminimum Phase System:Blind Adjustment of a Linear equalizer in Data Communication. IEEE Trans. on automatic control, 1980, 25: 385～399.
[11] D. N. Godard. Self-recovering equalization and carrier tracking in two dimensional data communication systems. IEEE Trans. on communication, 1980, 28: 1867～1875.
[12] G. Picchi, G. Prati. Blind equalization and carrier recovering using a "Stop-and-Go" Decision-Directed algorithm. IEEE Trans. on communication, 1987, 35: 877～887.
[13] L. Tong, G. Xu, T. Kailath. Blind channel identification and equalization using second-order statistics: a time domain approach. IEEE Trans. on Information Theory, 1994, 40(2): 340～349.[14] Moulines E, Duhamel P. Subspace methods for the blind identification of multipath channel FIR filters. IEEE Trans. on Signal Processing, 1995, 43: 516～525.
[15] Strom Erik G, Malmsten F. A maximum likelihood approach for extimating DS-CDMA multipath fading channels. IEEE Trans. On Communication, 2000, 18: 132～140.
[16] Shen J, Ding Z. Direct Blind MMSE channel equalization based on second-order statistics. IEEE Trans. On Signal Processing, 2000, 48: 1015～1022.
[17] Abed-Mer K, Duhamel P, Gesbert D. Prediction error methods for time-domain blind identification of multichannel FIR filters. Proc. of IEEE ICASSP, 1995, 1968～1971.
[18] Tugnait J T. Linear prediction error method for blind identification of time-varying channels: theoretical results. Proc. of 2001 IEEE ICASSP, 2001.
[19] Zhang Liyi, Jia Hairong, Wang Huakui etc. Analysis of the Blind Equalization Algorithm Based on Signal Detection Theory. Proceedings of the Sixth International Conference on Electronic Measurement and Instruments, 2001, 240～243.
[20] 张立毅，鲁瑞，王华奎等．基于神经网络盲均衡算法的分析．电子测量与仪器学报，2002，27(6)：1867～1875．
[21] S. Mo, B. Shafai. Blind equalization using higher order cumulates and neural network IEEE Trans. on signal processing, 1994, 42: 3209.
[22] 梁启联，周正．基于多层神经网络的盲均衡算法．北京邮电大学学报，1996，19(3)：27～32．
[23] Rosario R A. rapid multi-layer perceptron training algorithm. Proc of IJCNN-92. Baltimore, Mary-land: 1992, 824～829.
[24] Solis F J, Wets J B. Minimization by random search techniques. Mathematics of Operation Research, 1981, 6: 19～30.[25] G. Kechriotis. Using recurrent neural network for adaptive communica-tion channel equalization. IEEE Trans. on Neural Network, 1994, 5(2): 267-278.
[26] 梁启联，周正，刘泽民．基于递归神经网络的盲均衡算法的改进．北京邮电大学学报，1997，20(4)：6-11．
[27] 张立毅．数字通信系统中盲均衡技术的研究．北京：北京理工大学博士学位论文，2003．
[28] 鲁瑞．基于神经网络理论盲均衡算法的研究．太原：太原理工大学硕士学位论文，2003．
[29] Yong Fang, W. S. Tommy. Linear neural network based blind equalization. IEEE Trans. on Signal Processing, 1999, 76(1): 37～42.
[30] 刘琚，何振亚．一种基于ICA和过采样技术的盲反卷积方法．现代雷达，1998，20(4)：41～45．
[31] R. Ben Abdennour, F. Bouani, M. Ksouri etc. Neural network structures applied for both channel modeling and blind equalization. In Proc. IEEE ICS, 1999, 53: 395～400.
[32] 刘建成，张清泰，蔡湛宇．用于多信道盲反卷积的级联神经网络．计算机与网络，1999，14(2)：20．
[33] 赵建业，余道衡．用细胞神经网络实现盲均衡的一种方法．电子科学学刊，2000，22(3)：423～428．
[34] 谷源涛，唐昆，崔慧娟，杜文．新的变步长归一化最小均方算法．清华大学学报，2002，42(1)：15～18．
[35] 高鹰，谢胜利．一种变步长LMS自适应滤波算法及分析．电子学报，2001，29(8)：1094～1097．
[36] 覃景繁，欧阳景正．一种新的变步长LMS自适应滤波算法．数据采集与处理，1997，12(9)：171～174．
[37] 蒋明峰，郑小林，彭承琳．一种新的变步长LMS自适应算法及其在自适应噪声对消中的应用．信号处理，2001，17(6)：282～286．[38] 李盈颖，万建伟，周良柱．一种改进的变步长归一化LMS算法．国防科技大学学报，1999，21(1)：94～96．
[39] 盛三元，王建华．一种新的变步长LMS自适应滤波算法．华东船舶工业学院学报，2002，16(6)：50．
[40] 邓江波，侯新国．一种新的变步长LMS自适应算法及其性能分析．电声技术，2004，12：4～6．
[41] Dimitrios I. Pazaitis, A.G.Constantinides. A Novel kurtosis Driven Variable Step-Size adaptive Algorithm. IEEE Yrans.on signal processing, 1999, 47(3): 864.
[42] Andreas Mader, Henning Puder. Step-size Control for Acoustic Echo Cancellation Filters-an Overview. Signal Processing, 2000, 80(1): 1697～1719.
[43] Wee-peng Ang, B.Farhang-Boroujeny. A New Class of Gradient Adaptive Step-Size LMS Algorithms. IEEE Trans.on signal processing, 2001, 49(4): 805～810.
[44] V. John Mathews, Zhenhua Xie. A Stochastic Gradient Adaptive Filter with Gradient Adaptive Step Size. IEEE Trans. on signal processing, 1993, 41(6): 2075～2087.
[45] R. H. Kwrong, E. W. Johnson. A Variable step size LMS algorithm. IEEE Trans. on signal processing, 1992, 40(7): 1633～1642.
[46] K. Mayyas, T. Aboulnasr. A Robust Variable Step-Size LMS-Type Algorithm: Analysis and Simulations. IEEE Trans. on signal processing, 1997, 45(3): 631～639.
[47] 愈洋，杨俊松，田亚菲．一种新的变步长LMS算法及其仿真．甘肃科学学报，2005，17(2)：34～37．
[48] 张秦；冯存前．变步长LMS算法及其在自适应消噪中的应用．现代电子技术，2003，14：88～90．
[49] 谷源涛，唐昆，崔慧娟，杜文．独立假设下最优变步长LMS模型和算法．中??国科学，2003，33(8)：760～768．
[50] 冯存前，张永顺，赵宗宝．一种新的变步长LMS自适应算法及其仿真．电子对抗技术，2003，34(4)：31～33．
[51] 成磊，葛临东．变步长LMS性能比较与仿真．信息工程大学学报，2003，4(4)：70～73．
[52] 彭劲东，段正华，王梓展．一种变步长解相关LMS算法．计算机工程与应用，2004，30：136～138．
[53] 曾少华，刘贵忠．基于变步长LMS的自适应匹配算法．电子与信息学报，2004，26(2)：241～247．
[54] 徐凯，纪红，乐光新．一种改进的变步长自适应滤波器LMS算法．电路与系统学报，2004，9(4)：115～117．
[55] 王敏强，郑宝玉．一种新的可变步长LMS自适应滤波算法．信号处理，2004，20(6)：613～617．
[56] 陈凯，张平．一种新的变步长LMS自适应滤波算法．电子技术，2004，23：56～57．
[57] 赵俊渭，郭业才，李金明．基于峭度激励的变步长自适应谱线增强算法．哈尔滨工程大学学报，2004，25(4)：412～422．
[58] 张雄．基于Bussgang技术盲均衡算法的研究．太原：太原理工大学硕士学位论文，2003．
[59] 赵宝峰，赵菊敏，张立毅．基于MSE变换的变步长恒模盲均衡算法．太原理工大学学报，2005，36(4)：395～397．
[60] 白煜．基于模糊神经网络理论盲均衡算法的研究．太原：太原理工大学硕士学位论文，2005．
[61] 田俊霞，匡镜明，王华．基于修正常系数模板的变步长双模式盲自适应均衡算法．北京理工大学学报，2004，24(8)：716～718．
[62] 叶世伟，史忠植．神经网络原理．北京：机械工业出版社，2004．
[63] 焦李成．神经网络系统理论．西安：西安电子科技大学出版社，1992．[64] J. J. Hopfield. Neural networks and physical system with emergent collective computational abilities. Proc. Natl. Acad. Sci., U. S. A., 1982, 79: 2554～2558.
[65] J. J. Hopfield. Neurons with Graded Respone have collective computational properties like those of two state neurons. Proc. Natl. Acad. Sci., U. S. A., 1984, 81: 3088～3092.
[66] D. E. Rumelhart, J. L. Meclelland. Parallel distributed processing. MIT Press, Cambridge, MA, 1986, 1: 2.
[67] L. O. Chua, L. Yang. Cellular Neural Networks: Theory. IEEE Trans. on Circuits and Systems, 1988, 35: 1257～1272.
[68] L. O. Chua, L. Yang. Cellular Neural Networks: Application. IEEE Trans. on Circuits and Systems, 1988, 35: 273～1290.
[69] G. E. Hinton, T. J. Sejuowski, D. H. Ackley. Boltzmann Machines: Cotraint Satisfaction Networks that Learn. Camegie-Mellon University, Tech, Report CMU-CS-84-119, 1984.
[70] 张立明．人工神经网络的模型及其应用．上海：复旦大学出版社，1993．
[71] 李学桥，马莉．神经网络工程应用．重庆：重庆大学出版社，1996．
[72] 赵振宇．模糊理论和神经网络的基础应用．北京：清华大学出版社，1996．
[73] 袁曾任．人工神经元网络及其应用．北京：清华大学出版社，1999．
[74] 李士勇．模糊控制、神经控制和智能控制论．哈尔滨：哈尔滨工业大学出版社，1998．
[75] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2005．
[76] R. Hecht-Nielsen. Kolmogorov's Mapping Neural Network Existence Theorem. IEEE International Conference on Neural Network, San Diego, SOS Printing, 1987, 2: 11～14.
[77] 陈戍，常胜江．神经网络的自适应删减学习算法及其应用．物理学报，2001，50(4)：674～681．[78] Rumelhart D E. Lesrning representation by BP Errors. Nature (London), 1986, 7: 149～154.
[79] P. Patrick. Minimization Method for Training Feedforward Neural Networks. NeuralNetworks, 1994, 7: 1～11.
[80] F. Palmieri. Sound Localization with a Neural Network Trained with the Multiple Extended Kalmamn Algorithm. Proc IJCNN, 1991, 125～131.
[81] R. Parisi. A Generalized Learning Paradiam Exploiting the Structure of Feed-forwardNN. IEEETransNN, 1996, 7: 1450～1459.
[82] 姚天任，孙洪．现代数字信号处理．武汉：华中理工大学出版社，2000．
[83] O. Shalvi, E. Weinstein. New criteria for blind deconvolution of non-minimum phase systems (channels). IEEE Trans. on information theory, 1990, 36:312～321.
[84] J. Cadzow. A blind deconvolution via cumulant extrema. IEEE Trans. on signal processing, 1996, 13: 24～42.
[85] 严太山．神经网络BP算法研究及其在工业检测中的应用．南宁：广西师范大学硕士学位论文，2001年．
[86] 吴仁志．单神经元和多层前向人工神经网络的研究与应用．南宁：广西大学硕士学位论文，2002年．
[87] Cheolwoo You, Daesik Hong. Nonlinear Blind Equalization Schemes Using Complex-Valued Multilayer Feedforward Neural Networks. IEEE transactions on neural networks, 1998, 8(6): 1442～1455.
[88] P. balay, J. Palicot. Equalization of nonlinear perturbations by a multiplayer perceptron in satellite channel transmission. In Proc. IEEE CROBECOM., 1994, 22～26.
[89] Luo. F. L, Unbehauen. Applied neural network for signal processing. New York Cambirdge university Press, 1997.
[90] Douglas. Neural network for blind deconvolution of signal. IEEE Trans on signal??processing, 1997, 45: 2829～2842.
[91] 欧阳喜，葛临东，一种新的基于误差切换的盲均衡算法，数字通信，2000，2：10-11．
[92] Sridhar V, Convex Cost Function in Blind Equalization, IEEE Trans. On Signal Processing, 1994, 42 (8): 1257.
[93] H. Raymond, A Variable Step Size LMS Algorithm, IEEE Trans. On Signal Processing, 1992, 40: 1633-1642.
[02] 张贤达，保铮．通信信号处理．北京：国防工业出版社，2000．
[03] 邹谋炎．反卷积和信号复原．北京：国防工业出版社，2001．
[04] 沈风麟，陈和晏．生物医学随机信号处理．合肥：中国科学技术大学出版社，1999．
[05] 周耀华，王凯仁．数字信号处理．上海：复旦大学出版社，1993．
[06] 何振亚．自适应信号处理．北京：科学出版社，2002．
[07] 张贤达．现代信号处理(第二版)．北京：清华大学出版社，2003．
[08] L. tong, J. K. tugnait. Signle-user channel estimation and equalization. IEEE Signal processing Magazine, 2000: 17～28.
[09] Y. Sato. A method of self-recovering equalization for multiple amplitude modulation schemes. IEEE Trans. on communication, 1975, 23: 679～682.
[10] A. Benveniste, M. Goursat, G. Ruget. Robust Identification of a Nonminimum Phase System:Blind Adjustment of a Linear equalizer in Data Communication. IEEE Trans. on automatic control, 1980, 25: 385～399.
[11] D. N. Godard. Self-recovering equalization and carrier tracking in two dimensional data communication systems. IEEE Trans. on communication, 1980, 28: 1867～1875.
[12] G. Picchi, G. Prati. Blind equalization and carrier recovering using a "Stop-and-Go" Decision-Directed algorithm. IEEE Trans. on communication, 1987, 35: 877～887.
[13] L. Tong, G. Xu, T. Kailath. Blind channel identification and equalization using second-order statistics: a time domain approach. IEEE Trans. on Information Theory, 1994, 40(2): 340～349.[14] Moulines E, Duhamel P. Subspace methods for the blind identification of multipath channel FIR filters. IEEE Trans. on Signal Processing, 1995, 43: 516～525.
[15] Strom Erik G, Malmsten F. A maximum likelihood approach for extimating DS-CDMA multipath fading channels. IEEE Trans. On Communication, 2000, 18: 132～140.
[16] Shen J, Ding Z. Direct Blind MMSE channel equalization based on second-order statistics. IEEE Trans. On Signal Processing, 2000, 48: 1015～1022.
[17] Abed-Mer K, Duhamel P, Gesbert D. Prediction error methods for time-domain blind identification of multichannel FIR filters. Proc. of IEEE ICASSP, 1995, 1968～1971.
[18] Tugnait J T. Linear prediction error method for blind identification of time-varying channels: theoretical results. Proc. of 2001 IEEE ICASSP, 2001.
[19] Zhang Liyi, Jia Hairong, Wang Huakui etc. Analysis of the Blind Equalization Algorithm Based on Signal Detection Theory. Proceedings of the Sixth International Conference on Electronic Measurement and Instruments, 2001, 240～243.
[20] 张立毅，鲁瑞，王华奎等．基于神经网络盲均衡算法的分析．电子测量与仪器学报，2002，27(6)：1867～1875．
[21] S. Mo, B. Shafai. Blind equalization using higher order cumulates and neural network IEEE Trans. on signal processing, 1994, 42: 3209.
[22] 梁启联，周正．基于多层神经网络的盲均衡算法．北京邮电大学学报，1996，19(3)：27～32．
[23] Rosario R A. rapid multi-layer perceptron training algorithm. Proc of IJCNN-92. Baltimore, Mary-land: 1992, 824～829.
[24] Solis F J, Wets J B. Minimization by random search techniques. Mathematics of Operation Research, 1981, 6: 19～30.[25] G. Kechriotis. Using recurrent neural network for adaptive communica-tion channel equalization. IEEE Trans. on Neural Network, 1994, 5(2): 267-278.
[26] 梁启联，周正，刘泽民．基于递归神经网络的盲均衡算法的改进．北京邮电大学学报，1997，20(4)：6-11．
[27] 张立毅．数字通信系统中盲均衡技术的研究．北京：北京理工大学博士学位论文，2003．
[28] 鲁瑞．基于神经网络理论盲均衡算法的研究．太原：太原理工大学硕士学位论文，2003．
[29] Yong Fang, W. S. Tommy. Linear neural network based blind equalization. IEEE Trans. on Signal Processing, 1999, 76(1): 37～42.
[30] 刘琚，何振亚．一种基于ICA和过采样技术的盲反卷积方法．现代雷达，1998，20(4)：41～45．
[31] R. Ben Abdennour, F. Bouani, M. Ksouri etc. Neural network structures applied for both channel modeling and blind equalization. In Proc. IEEE ICS, 1999, 53: 395～400.
[32] 刘建成，张清泰，蔡湛宇．用于多信道盲反卷积的级联神经网络．计算机与网络，1999，14(2)：20．
[33] 赵建业，余道衡．用细胞神经网络实现盲均衡的一种方法．电子科学学刊，2000，22(3)：423～428．
[34] 谷源涛，唐昆，崔慧娟，杜文．新的变步长归一化最小均方算法．清华大学学报，2002，42(1)：15～18．
[35] 高鹰，谢胜利．一种变步长LMS自适应滤波算法及分析．电子学报，2001，29(8)：1094～1097．
[36] 覃景繁，欧阳景正．一种新的变步长LMS自适应滤波算法．数据采集与处理，1997，12(9)：171～174．
[37] 蒋明峰，郑小林，彭承琳．一种新的变步长LMS自适应算法及其在自适应噪声对消中的应用．信号处理，2001，17(6)：282～286．[38] 李盈颖，万建伟，周良柱．一种改进的变步长归一化LMS算法．国防科技大学学报，1999，21(1)：94～96．
[39] 盛三元，王建华．一种新的变步长LMS自适应滤波算法．华东船舶工业学院学报，2002，16(6)：50．
[40] 邓江波，侯新国．一种新的变步长LMS自适应算法及其性能分析．电声技术，2004，12：4～6．
[41] Dimitrios I. Pazaitis, A.G.Constantinides. A Novel kurtosis Driven Variable Step-Size adaptive Algorithm. IEEE Yrans.on signal processing, 1999, 47(3): 864.
[42] Andreas Mader, Henning Puder. Step-size Control for Acoustic Echo Cancellation Filters-an Overview. Signal Processing, 2000, 80(1): 1697～1719.
[43] Wee-peng Ang, B.Farhang-Boroujeny. A New Class of Gradient Adaptive Step-Size LMS Algorithms. IEEE Trans.on signal processing, 2001, 49(4): 805～810.
[44] V. John Mathews, Zhenhua Xie. A Stochastic Gradient Adaptive Filter with Gradient Adaptive Step Size. IEEE Trans. on signal processing, 1993, 41(6): 2075～2087.
[45] R. H. Kwrong, E. W. Johnson. A Variable step size LMS algorithm. IEEE Trans. on signal processing, 1992, 40(7): 1633～1642.
[46] K. Mayyas, T. Aboulnasr. A Robust Variable Step-Size LMS-Type Algorithm: Analysis and Simulations. IEEE Trans. on signal processing, 1997, 45(3): 631～639.
[47] 愈洋，杨俊松，田亚菲．一种新的变步长LMS算法及其仿真．甘肃科学学报，2005，17(2)：34～37．
[48] 张秦；冯存前．变步长LMS算法及其在自适应消噪中的应用．现代电子技术，2003，14：88～90．
[49] 谷源涛，唐昆，崔慧娟，杜文．独立假设下最优变步长LMS模型和算法．中??国科学，2003，33(8)：760～768．
[50] 冯存前，张永顺，赵宗宝．一种新的变步长LMS自适应算法及其仿真．电子对抗技术，2003，34(4)：31～33．
[51] 成磊，葛临东．变步长LMS性能比较与仿真．信息工程大学学报，2003，4(4)：70～73．
[52] 彭劲东，段正华，王梓展．一种变步长解相关LMS算法．计算机工程与应用，2004，30：136～138．
[53] 曾少华，刘贵忠．基于变步长LMS的自适应匹配算法．电子与信息学报，2004，26(2)：241～247．
[54] 徐凯，纪红，乐光新．一种改进的变步长自适应滤波器LMS算法．电路与系统学报，2004，9(4)：115～117．
[55] 王敏强，郑宝玉．一种新的可变步长LMS自适应滤波算法．信号处理，2004，20(6)：613～617．
[56] 陈凯，张平．一种新的变步长LMS自适应滤波算法．电子技术，2004，23：56～57．
[57] 赵俊渭，郭业才，李金明．基于峭度激励的变步长自适应谱线增强算法．哈尔滨工程大学学报，2004，25(4)：412～422．
[58] 张雄．基于Bussgang技术盲均衡算法的研究．太原：太原理工大学硕士学位论文，2003．
[59] 赵宝峰，赵菊敏，张立毅．基于MSE变换的变步长恒模盲均衡算法．太原理工大学学报，2005，36(4)：395～397．
[60] 白煜．基于模糊神经网络理论盲均衡算法的研究．太原：太原理工大学硕士学位论文，2005．
[61] 田俊霞，匡镜明，王华．基于修正常系数模板的变步长双模式盲自适应均衡算法．北京理工大学学报，2004，24(8)：716～718．
[62] 叶世伟，史忠植．神经网络原理．北京：机械工业出版社，2004．
[63] 焦李成．神经网络系统理论．西安：西安电子科技大学出版社，1992．[64] J. J. Hopfield. Neural networks and physical system with emergent collective computational abilities. Proc. Natl. Acad. Sci., U. S. A., 1982, 79: 2554～2558.
[65] J. J. Hopfield. Neurons with Graded Respone have collective computational properties like those of two state neurons. Proc. Natl. Acad. Sci., U. S. A., 1984, 81: 3088～3092.
[66] D. E. Rumelhart, J. L. Meclelland. Parallel distributed processing. MIT Press, Cambridge, MA, 1986, 1: 2.
[67] L. O. Chua, L. Yang. Cellular Neural Networks: Theory. IEEE Trans. on Circuits and Systems, 1988, 35: 1257～1272.
[68] L. O. Chua, L. Yang. Cellular Neural Networks: Application. IEEE Trans. on Circuits and Systems, 1988, 35: 273～1290.
[69] G. E. Hinton, T. J. Sejuowski, D. H. Ackley. Boltzmann Machines: Cotraint Satisfaction Networks that Learn. Camegie-Mellon University, Tech, Report CMU-CS-84-119, 1984.
[70] 张立明．人工神经网络的模型及其应用．上海：复旦大学出版社，1993．
[71] 李学桥，马莉．神经网络工程应用．重庆：重庆大学出版社，1996．
[72] 赵振宇．模糊理论和神经网络的基础应用．北京：清华大学出版社，1996．
[73] 袁曾任．人工神经元网络及其应用．北京：清华大学出版社，1999．
[74] 李士勇．模糊控制、神经控制和智能控制论．哈尔滨：哈尔滨工业大学出版社，1998．
[75] 阎平凡，张长水．人工神经网络与模拟进化计算．北京：清华大学出版社，2005．
[76] R. Hecht-Nielsen. Kolmogorov's Mapping Neural Network Existence Theorem. IEEE International Conference on Neural Network, San Diego, SOS Printing, 1987, 2: 11～14.
[77] 陈戍，常胜江．神经网络的自适应删减学习算法及其应用．物理学报，2001，50(4)：674～681．[78] Rumelhart D E. Lesrning representation by BP Errors. Nature (London), 1986, 7: 149～154.
[79] P. Patrick. Minimization Method for Training Feedforward Neural Networks. NeuralNetworks, 1994, 7: 1～11.
[80] F. Palmieri. Sound Localization with a Neural Network Trained with the Multiple Extended Kalmamn Algorithm. Proc IJCNN, 1991, 125～131.
[81] R. Parisi. A Generalized Learning Paradiam Exploiting the Structure of Feed-forwardNN. IEEETransNN, 1996, 7: 1450～1459.
[82] 姚天任，孙洪．现代数字信号处理．武汉：华中理工大学出版社，2000．
[83] O. Shalvi, E. Weinstein. New criteria for blind deconvolution of non-minimum phase systems (channels). IEEE Trans. on information theory, 1990, 36:312～321.
[84] J. Cadzow. A blind deconvolution via cumulant extrema. IEEE Trans. on signal processing, 1996, 13: 24～42.
[85] 严太山．神经网络BP算法研究及其在工业检测中的应用．南宁：广西师范大学硕士学位论文，2001年．
[86] 吴仁志．单神经元和多层前向人工神经网络的研究与应用．南宁：广西大学硕士学位论文，2002年．
[87] Cheolwoo You, Daesik Hong. Nonlinear Blind Equalization Schemes Using Complex-Valued Multilayer Feedforward Neural Networks. IEEE transactions on neural networks, 1998, 8(6): 1442～1455.
[88] P. balay, J. Palicot. Equalization of nonlinear perturbations by a multiplayer perceptron in satellite channel transmission. In Proc. IEEE CROBECOM., 1994, 22～26.
[89] Luo. F. L, Unbehauen. Applied neural network for signal processing. New York Cambirdge university Press, 1997.
[90] Douglas. Neural network for blind deconvolution of signal. IEEE Trans on signal??processing, 1997, 45: 2829～2842.
[91] 欧阳喜，葛临东，一种新的基于误差切换的盲均衡算法，数字通信，2000，2：10-11．
[92] Sridhar V, Convex Cost Function in Blind Equalization, IEEE Trans. On Signal Processing, 1994, 42 (8): 1257.
[93] H. Raymond, A Variable Step Size LMS Algorithm, IEEE Trans. On Signal Processing, 1992, 40: 1633-1642.