用于高阶MPPSK信号检测的多分类SVM新算法
|
摘要
为了降低支持向量机(SVM)算法在高阶多元位置相移键控(M-ary Position Phase Shift Keying,MPPSK)系统的信号检测复杂度,在分析常用SVM多分类算法的基础上,提出了一种新的具有更低复杂度的类二分法SVM。为了进一步提高高阶MPPSK信号检测性能,提出一种新的SVM特征向量提取方法,调制矩阵法,并将两种方法结合起来,用于高阶MPPSK系统的信号检测。仿真结果表明:类二分法SVM能显著降低多分类SVM的算法复杂度,调制矩阵选取特征向量法能够显著提高高阶MPPSK系统的检测性能,两种方法结合用于高阶MPPSK系统,可以在有效降低复杂度的前提下保证期望的检测性能。
In order to lower the complexity of M-ary Position Phase Shift Keying( MPPSK) signal detection based on support vector machines( SVM),a new method named similar dichotomy SVM with low complexity was proposed after the discussion of common methods for the multiclass classification of SVM. Then a new approach of choosing SVM characteristic vector,known as modulation array method,was proposed,which meant to improve the detection performance of high-order MPPSK system. At last,these two methods were combined to be used in the detection of higher-order MPPSK system.The simulation results indicate that: the complexity of multiclass classification SVM can be significantly lowered and the detection performance of the high-order MPPSK system is effectively improved by using similar dichotomy and modulation array method respectively. And combining these two methods,the complexity can be effectively lowered while the expected detection performance is guaranteed.
In order to lower the complexity of M-ary Position Phase Shift Keying( MPPSK) signal detection based on support vector machines( SVM),a new method named similar dichotomy SVM with low complexity was proposed after the discussion of common methods for the multiclass classification of SVM. Then a new approach of choosing SVM characteristic vector,known as modulation array method,was proposed,which meant to improve the detection performance of high-order MPPSK system. At last,these two methods were combined to be used in the detection of higher-order MPPSK system.The simulation results indicate that: the complexity of multiclass classification SVM can be significantly lowered and the detection performance of the high-order MPPSK system is effectively improved by using similar dichotomy and modulation array method respectively. And combining these two methods,the complexity can be effectively lowered while the expected detection performance is guaranteed.
引文
[1]靳一,吴乐南,余静,陈艺方.MPPSK调制解调器研究[J].信号处理,2012,28(7):917-925.Jin Y,Wu L N,Yu J,Chen Y F.The research of MPPSK modem[J].Signal Processing,2012,28(7):917-925.(in Chinese)
[2]吴乐南,戚晨皓.多元位置相移键控调制和解调方法[P].中国:CN100586116C,2010.1.27.Wu L N,Qi C H.The modulation and demodulation method of M-ary Position Phase Shift Keying[P].China:CN100586116C,2010.1.27.(in Chinese)
[3]邓乃杨,田英杰.支持向量机:理论、算法与拓展[M].北京:科学出版社,2009:10-249.Deng N Y,Tian Y J.Support Vector Machine:theory、algorithms and expansion[M].Beijing:Science Press,2009:10-249.(in Chinese)
[4]Vapnik V N.The nature of statistical learning theory[M].New York:Springer Verilog,1995:4-80.
[5]Platt J C,Cristianini N,Shawe-Taylor J.Large Margin DAGs for Multiclass classification[A].Advances in Neural Information Processing Systems,2000,12(3):547-553.
[6]夏建涛,何明一.支持向量机与纠错编码相结合的多类分类算法[J].西北工业大学学报,2003,21(4):443-448.Xia J T,He M Y.The multi-class classification algorithm of Support vector machines combined with error correction coding[J].Northwestern Industry University Journal,2003,21(4):443-448.(in Chinese)
[7]段修生,单甘霖,张岐龙.用于多类分类的层次式支持向量机[J].军械工程学院学报,2009,21(1):64-66.Duan X S,Shan G L,Zhang Q L.Hierarchical Support Vector Machine used in multi-class classification[J].Ordnance Engineering College Journal,2009,21(1):64-66.(in Chinese)
[8]窦智宙,孟昭进.基于MATLAB的一对一SVM算法实现[J].干部学院学报,2008,1:145-147.Dou Z Z,Meng S J.The complement of one-versus-one SVM algorithm based on MATLAB[J].Cadres College Journal,2008,1:145-147.(in Chinese)
[9]吴乐南,应鹏魁.扩展的二元相移键控调制解调器及其实现方法[P].中国:CN102014093A,2011.4.13.Wu L N,Ying P K.The accomplishment method of Extended Binary Shift Keying[P].China:CN102014093A,2011.4.13.(in Chinese)
[10]Stefan Knerr,Personnaz Léon,Dreyfus Gérard,et al.Single-layer learning revisited:a stepwise procedure for building and training a neural network[J].Optimization Methods and Software,1990,1:23-34.
[11]陈贤卿.高效调制通信系统中的检测与信道编码的研究[D].南京:东南大学信息科学与工程学院,2012.Chen X Q.Research on Detection and Channel Coding of Efficient Modulation communications[D].Nanjing:School of Information Science and Engineering Southeast University,2012.(in Chinese)
[12]刘勇,全廷伟.基于DAG-SVMS的SVM多类分类方法[J].统计与决策,2007,248(20):145-148.Liu Y,Quan T W.SVM multi-class Classification Method based on DAG-SVMS[J].Statistics and Decision,2007,248(20):145-148.(in Chinese)
[13]吴乐南,冯熳,高鹏.用于增强不对称二元调制信号的冲击滤波方法[P].中国:CN101599754 A,2009.12.9.Wu L N,Feng M,Gao P.A method to Enhance the Impulse Filter of Asymmetric dual elements modulated Signal[P].China:CN101599754 A,2009.12.9.(in Chinese)
[2]吴乐南,戚晨皓.多元位置相移键控调制和解调方法[P].中国:CN100586116C,2010.1.27.Wu L N,Qi C H.The modulation and demodulation method of M-ary Position Phase Shift Keying[P].China:CN100586116C,2010.1.27.(in Chinese)
[3]邓乃杨,田英杰.支持向量机:理论、算法与拓展[M].北京:科学出版社,2009:10-249.Deng N Y,Tian Y J.Support Vector Machine:theory、algorithms and expansion[M].Beijing:Science Press,2009:10-249.(in Chinese)
[4]Vapnik V N.The nature of statistical learning theory[M].New York:Springer Verilog,1995:4-80.
[5]Platt J C,Cristianini N,Shawe-Taylor J.Large Margin DAGs for Multiclass classification[A].Advances in Neural Information Processing Systems,2000,12(3):547-553.
[6]夏建涛,何明一.支持向量机与纠错编码相结合的多类分类算法[J].西北工业大学学报,2003,21(4):443-448.Xia J T,He M Y.The multi-class classification algorithm of Support vector machines combined with error correction coding[J].Northwestern Industry University Journal,2003,21(4):443-448.(in Chinese)
[7]段修生,单甘霖,张岐龙.用于多类分类的层次式支持向量机[J].军械工程学院学报,2009,21(1):64-66.Duan X S,Shan G L,Zhang Q L.Hierarchical Support Vector Machine used in multi-class classification[J].Ordnance Engineering College Journal,2009,21(1):64-66.(in Chinese)
[8]窦智宙,孟昭进.基于MATLAB的一对一SVM算法实现[J].干部学院学报,2008,1:145-147.Dou Z Z,Meng S J.The complement of one-versus-one SVM algorithm based on MATLAB[J].Cadres College Journal,2008,1:145-147.(in Chinese)
[9]吴乐南,应鹏魁.扩展的二元相移键控调制解调器及其实现方法[P].中国:CN102014093A,2011.4.13.Wu L N,Ying P K.The accomplishment method of Extended Binary Shift Keying[P].China:CN102014093A,2011.4.13.(in Chinese)
[10]Stefan Knerr,Personnaz Léon,Dreyfus Gérard,et al.Single-layer learning revisited:a stepwise procedure for building and training a neural network[J].Optimization Methods and Software,1990,1:23-34.
[11]陈贤卿.高效调制通信系统中的检测与信道编码的研究[D].南京:东南大学信息科学与工程学院,2012.Chen X Q.Research on Detection and Channel Coding of Efficient Modulation communications[D].Nanjing:School of Information Science and Engineering Southeast University,2012.(in Chinese)
[12]刘勇,全廷伟.基于DAG-SVMS的SVM多类分类方法[J].统计与决策,2007,248(20):145-148.Liu Y,Quan T W.SVM multi-class Classification Method based on DAG-SVMS[J].Statistics and Decision,2007,248(20):145-148.(in Chinese)
[13]吴乐南,冯熳,高鹏.用于增强不对称二元调制信号的冲击滤波方法[P].中国:CN101599754 A,2009.12.9.Wu L N,Feng M,Gao P.A method to Enhance the Impulse Filter of Asymmetric dual elements modulated Signal[P].China:CN101599754 A,2009.12.9.(in Chinese)