基于极致学习机的通信信号辐射源个体识别技术研究
|
摘要
通信信号辐射源个体识别针对同型号、同批次、相同工作方式的不同辐射源,从接收信号中提取出最能反映辐射源个体差异的细微特征矢量,结合高效实时分类识别策略实现在线的个体识别,在未来电子侦察、无线网络安全及通信资源管理等应用领域将发挥巨大作用。
不同于传统的通信领域研究信息传输的高效可靠性,通信信号辐射源个体识别研究重点转向了调制信息背后隐藏的、由于辐射源器件性能参数、非线性差异在接收信号中留下的、可全面刻画信号辐射源的细微特征集合。本文首次系统地基于通信信号辐射源稳态工作时采样到的信号,深入研究了辐射源个体的细微特征产生机理及提取方法,以极致学习机为基础,提出了自适应在线序贯式极致学习机SLFN网络训练算法用于辐射源个体的在线分类识别,取得了很好的识别效果。
论文主要工作包括:
1、从理论和工程应用的角度出发,提出了以离散HT变换为基础的通信信号测量特征提取理论与方法,包括:基于解绕相位信息的载频线性拟合估计方法和基于绝对包络信息的数字通信信号的码元速率估计方法等,大量仿真和实测结果显示了本文所提方法精确有效且易于实现。
2、为提取包括FM信号调频指数和数字通信信号调制参数在内的调制特征信息,提出了基于FFT迭代插值和CZT的两种通用且易于DSP实现的高性能单音信号参数估计方法,前者通过动态对称频率偏差估计器的不断迭代过程,自适应地调整估计精度,不仅测量精度较传统FFT插值方法高,而且信噪比门限要求低、速度快、全频率范围内性能表现均很接近CRLB下限;后者借助一定的先验知识,提高了频率分辨率,对短时平稳和存在近邻分量干扰时的单音信号参数估计,结果准确且稳定一致,十分接近CRLB下限,具有更强的抵抗近邻分量干扰和噪声的能力。
3、基于通信信号非线性寄生幅度调制杂散输出特征和欠采样技术提出的特征双谱,有效地将信号固有自发频率分量与因非线性QPC所致的频率分量区分开来,保留了高阶谱分析的所有优点同时又大大减少了分析时间,基于积分围线的特征双谱选择结合了积分双谱和选择双谱的优点,抗噪性强、无交叉项干扰、计算简单高效,有效完成了2-D特征双谱向1-D矢量特征的转化,表现出较强的区分能力。
4、针对辐射源个体的时频特性,提出了基于EMD时频谱熵的杂散特征提取方法。EMD时频谱熵从自适应时频分布的角度分析了杂散输出的非线性、非平稳特征,基于相关程度的分解结束标准加快了EMD分解的速度,熵的概念的引入有效刻画了EMD时频分布谱图的能量分布均匀程度,以标量形式简单刻画了辐射源个体识别的时频能量分布,具备一定的区分能力。
5、在极致学习机算法的基础之上,利用回归拟合、正交化等相关技术,提出了SLFN网络中随选隐点对网络学习精度的贡献度指标,实现了网络模型的自适应逐一或成块递增式确定方式,结合序贯式学习的需要,提出了自适应在线序贯式极致学习机算法,实现了SLFN网络的自适应在线序贯式学习,全过程无需人工参与、对训练样本的到达方式和次序无任何要求,提升了算法的通用性,减少了对硬件设备计算存储能力的要求,大量仿真和实测结果说明了其简单有效性,对于FM、FSK实测通信信号辐射源个体的识别率均可达到85%以上。
此外,结合论文研究的需要,本文基于COM组件技术和数据库技术建立了通信信号辐射源个体识别的软件系统,该系统可根据评价辐射源细微特征的衡量标准,从载频、调制参数和杂散输出差异等对辐射源个体进行识别。
Communication signal transmitter recognition includes not only characteristicfine feature extraction from the sampled sequence generated by di?erent transmitterswith the same expected performance parameters and modulation mode, which actsas a comprehensive representation of the transmitters , but a highly e?cient andreal-time classification strategy to implement the online transmitter recognition. Itis of great significance and role in future electronic warfare, wireless network securityand communication resource management.
Unlike usual e?ciency- and reliance- oriented communication theory, communi-cation signal transmitter recognition focuses on the hidden fine feature set associatedwith the variances in actually working performance parameters and nonlinear prop-erty of the device and cares no details about modulated information. This papersystematically and thoroughly discusses the generating mechanism and extractionmethods of fine features from the sampled signal observations of di?erent steady-working transmitters for the first time, proposes an adaptive online sequential learn-ing extreme learning machine to train the Single hidden-Layer Feedforward Neuralnetwork (SLFN) to fulfill the classification and identification of transmitter basedon the extreme learning machine, which attains good recognition rate. The maincontributions are as follows:
1. From the theoretical and engineering point of view, the discrete HT-based theoryand method for communication signal measured feature (carrier and modulationparameters) extraction are presented, which includes unwrapped instantaneousphase based linear fitting carrier estimation and absolute envelope based symbolrate estimation for digital communication signal, etc. Lots of simulations andexperiments show the proposed methods are e?ective and simple to implement.
2. To extract such modulation-relative features as FM modulation index and dig-ital modulation parameters, two generalized, DSP implementable FFT-basediteratively interpolated and CZT-based high performance single tone parameterestimator are proposed. The FFT-based one adaptively adjusts the estima- tion accuracy by the constant iteration of dynamic symmetric frequency o?setestimation to attain not only more accurate estimation but lower SNR thresh-old, higher speed and much more CRLB-approximating performance over thefull frequency range compared to traditional Rife-Jane’s and Quinn’s one. TheCZT-based one enhances the frequency resolution by means of certain priori toobtain an accurate, stable, consistent, closely CRLB-approximating single-toneparameter estimation as well as more robustness to noise in the case of short-timestationarity and adjacent single-tone interference
3. Based on nonlinear parasitic amplitude modulation miscellaneous output and sub-sampling, featured bispectra e?ectively distinguish the spontaneous intrinsic fre-quency from that caused by nonlinear QPC, retain all property of HOS butgreatly reduce the computation e?ort meanwhile. Surrounding line Integration-based featured bispectra selection incorporates advantages of both integrated bis-pectra and selected bispectra: strong noise robustness, no cross interference andsimple, e?cient computation, accomplishes the translation from 2-D featuredbispectra into 1-D vectored feature which shows great individual transmitterdiscriminant ability.
4. Based on the time-frequency property of individual transmitters, EMD time-frequency spectrum entropy- based miscellaneous feature extraction is proposed,which characterizes the nonlinear, nonstationary miscellaneous output by adap-tive time-frequency distribution. Correlation degree- based decomposition fin-ished criterion speeds up the EMD process and the concept of entropy e?ectivelydepicts the degree of uniform distribution of the energy in EMD time-frequencyspectrum as a simple scalar of some individual transimitter discriminant ability.
5. Based on the extreme learning machine, contribution of randomly generated hid-den node in SLFN to network learning accuracy is proposed by regressive fittingand orthogonalization to realize the adaptability of network model one-by-oneor chunk-by-chunk and adaptive online sequential learning of SLFN which notonly runs automatically without manual parameter setting and special require-ment on the arriving mode and order of training samples but greatly reduces therequirements on storage and computation ability of hardware. Lots of simula- tions and experiments show that the proposed algorithm is e?ective and simpleto attain a recognition rate of above 85% for the collected both FM and FSKindividual transimitter observations.
Moreover, based on COM and Database techniques, a communication signaltransmitter recognition software system is established to implement our proposedalgorithms, which identifies individual transmitters by carrier, modulation parametersand miscellaneous feature variance according to conditions and criteria of these finefeatures.
不同于传统的通信领域研究信息传输的高效可靠性,通信信号辐射源个体识别研究重点转向了调制信息背后隐藏的、由于辐射源器件性能参数、非线性差异在接收信号中留下的、可全面刻画信号辐射源的细微特征集合。本文首次系统地基于通信信号辐射源稳态工作时采样到的信号,深入研究了辐射源个体的细微特征产生机理及提取方法,以极致学习机为基础,提出了自适应在线序贯式极致学习机SLFN网络训练算法用于辐射源个体的在线分类识别,取得了很好的识别效果。
论文主要工作包括:
1、从理论和工程应用的角度出发,提出了以离散HT变换为基础的通信信号测量特征提取理论与方法,包括:基于解绕相位信息的载频线性拟合估计方法和基于绝对包络信息的数字通信信号的码元速率估计方法等,大量仿真和实测结果显示了本文所提方法精确有效且易于实现。
2、为提取包括FM信号调频指数和数字通信信号调制参数在内的调制特征信息,提出了基于FFT迭代插值和CZT的两种通用且易于DSP实现的高性能单音信号参数估计方法,前者通过动态对称频率偏差估计器的不断迭代过程,自适应地调整估计精度,不仅测量精度较传统FFT插值方法高,而且信噪比门限要求低、速度快、全频率范围内性能表现均很接近CRLB下限;后者借助一定的先验知识,提高了频率分辨率,对短时平稳和存在近邻分量干扰时的单音信号参数估计,结果准确且稳定一致,十分接近CRLB下限,具有更强的抵抗近邻分量干扰和噪声的能力。
3、基于通信信号非线性寄生幅度调制杂散输出特征和欠采样技术提出的特征双谱,有效地将信号固有自发频率分量与因非线性QPC所致的频率分量区分开来,保留了高阶谱分析的所有优点同时又大大减少了分析时间,基于积分围线的特征双谱选择结合了积分双谱和选择双谱的优点,抗噪性强、无交叉项干扰、计算简单高效,有效完成了2-D特征双谱向1-D矢量特征的转化,表现出较强的区分能力。
4、针对辐射源个体的时频特性,提出了基于EMD时频谱熵的杂散特征提取方法。EMD时频谱熵从自适应时频分布的角度分析了杂散输出的非线性、非平稳特征,基于相关程度的分解结束标准加快了EMD分解的速度,熵的概念的引入有效刻画了EMD时频分布谱图的能量分布均匀程度,以标量形式简单刻画了辐射源个体识别的时频能量分布,具备一定的区分能力。
5、在极致学习机算法的基础之上,利用回归拟合、正交化等相关技术,提出了SLFN网络中随选隐点对网络学习精度的贡献度指标,实现了网络模型的自适应逐一或成块递增式确定方式,结合序贯式学习的需要,提出了自适应在线序贯式极致学习机算法,实现了SLFN网络的自适应在线序贯式学习,全过程无需人工参与、对训练样本的到达方式和次序无任何要求,提升了算法的通用性,减少了对硬件设备计算存储能力的要求,大量仿真和实测结果说明了其简单有效性,对于FM、FSK实测通信信号辐射源个体的识别率均可达到85%以上。
此外,结合论文研究的需要,本文基于COM组件技术和数据库技术建立了通信信号辐射源个体识别的软件系统,该系统可根据评价辐射源细微特征的衡量标准,从载频、调制参数和杂散输出差异等对辐射源个体进行识别。
Communication signal transmitter recognition includes not only characteristicfine feature extraction from the sampled sequence generated by di?erent transmitterswith the same expected performance parameters and modulation mode, which actsas a comprehensive representation of the transmitters , but a highly e?cient andreal-time classification strategy to implement the online transmitter recognition. Itis of great significance and role in future electronic warfare, wireless network securityand communication resource management.
Unlike usual e?ciency- and reliance- oriented communication theory, communi-cation signal transmitter recognition focuses on the hidden fine feature set associatedwith the variances in actually working performance parameters and nonlinear prop-erty of the device and cares no details about modulated information. This papersystematically and thoroughly discusses the generating mechanism and extractionmethods of fine features from the sampled signal observations of di?erent steady-working transmitters for the first time, proposes an adaptive online sequential learn-ing extreme learning machine to train the Single hidden-Layer Feedforward Neuralnetwork (SLFN) to fulfill the classification and identification of transmitter basedon the extreme learning machine, which attains good recognition rate. The maincontributions are as follows:
1. From the theoretical and engineering point of view, the discrete HT-based theoryand method for communication signal measured feature (carrier and modulationparameters) extraction are presented, which includes unwrapped instantaneousphase based linear fitting carrier estimation and absolute envelope based symbolrate estimation for digital communication signal, etc. Lots of simulations andexperiments show the proposed methods are e?ective and simple to implement.
2. To extract such modulation-relative features as FM modulation index and dig-ital modulation parameters, two generalized, DSP implementable FFT-basediteratively interpolated and CZT-based high performance single tone parameterestimator are proposed. The FFT-based one adaptively adjusts the estima- tion accuracy by the constant iteration of dynamic symmetric frequency o?setestimation to attain not only more accurate estimation but lower SNR thresh-old, higher speed and much more CRLB-approximating performance over thefull frequency range compared to traditional Rife-Jane’s and Quinn’s one. TheCZT-based one enhances the frequency resolution by means of certain priori toobtain an accurate, stable, consistent, closely CRLB-approximating single-toneparameter estimation as well as more robustness to noise in the case of short-timestationarity and adjacent single-tone interference
3. Based on nonlinear parasitic amplitude modulation miscellaneous output and sub-sampling, featured bispectra e?ectively distinguish the spontaneous intrinsic fre-quency from that caused by nonlinear QPC, retain all property of HOS butgreatly reduce the computation e?ort meanwhile. Surrounding line Integration-based featured bispectra selection incorporates advantages of both integrated bis-pectra and selected bispectra: strong noise robustness, no cross interference andsimple, e?cient computation, accomplishes the translation from 2-D featuredbispectra into 1-D vectored feature which shows great individual transmitterdiscriminant ability.
4. Based on the time-frequency property of individual transmitters, EMD time-frequency spectrum entropy- based miscellaneous feature extraction is proposed,which characterizes the nonlinear, nonstationary miscellaneous output by adap-tive time-frequency distribution. Correlation degree- based decomposition fin-ished criterion speeds up the EMD process and the concept of entropy e?ectivelydepicts the degree of uniform distribution of the energy in EMD time-frequencyspectrum as a simple scalar of some individual transimitter discriminant ability.
5. Based on the extreme learning machine, contribution of randomly generated hid-den node in SLFN to network learning accuracy is proposed by regressive fittingand orthogonalization to realize the adaptability of network model one-by-oneor chunk-by-chunk and adaptive online sequential learning of SLFN which notonly runs automatically without manual parameter setting and special require-ment on the arriving mode and order of training samples but greatly reduces therequirements on storage and computation ability of hardware. Lots of simula- tions and experiments show that the proposed algorithm is e?ective and simpleto attain a recognition rate of above 85% for the collected both FM and FSKindividual transimitter observations.
Moreover, based on COM and Database techniques, a communication signaltransmitter recognition software system is established to implement our proposedalgorithms, which identifies individual transmitters by carrier, modulation parametersand miscellaneous feature variance according to conditions and criteria of these finefeatures.
引文
[1]张国柱,周一宇,姜文利.基于贝叶斯理论的辐射源分类识别方法研究.信号处理, 2004, 20(4):350–352.
[2]刘刚.雷达指纹分析的基本理论探讨.电子对抗, 2002, 6:1–6.
[3]牛海,马颖.小波一神经网络在辐射源识别中的应用研究.系统工程与电子技术, 2002, 124(15):55–57.
[4] V.Desimir and O.Milorad. Spectral correlation of psk signals. Telecommunica-tions in Modern Satellite, Cable and Broadcasting Services, 1999, 1:273–276.
[5] W. Gardner, W.Brown, and Chih-Kang Chen. Spectral correlation of modulatedsignals: Part ii–digital modulation. IEEE Transactions on Communications,1987, 35(6):595–601.
[6]陈卫东,杨绍全.基于循环累量不变量的mpsk信号调制识别算法.电子与信息学报, 2003, 25(3):320–325.
[7] A. K.Jain and P.Dubin. Statistical pattern recognition: a review. IEEE Transon Pattern Analysis and Machine Intelligence, 2000, 22(1):4–37.
[8] A. Swami and B. Sadler. Hierarchical digital modulation classification usingcumulants. IEEE Transactions on Communications, 2000, 48(3):416–429.
[9] A. K.Nandi and E.E. Azzouz. Algorithms for automatic modulation recognitionof communication signals. IEEE Transactions on Communications, 1998, , Vol:46(4):431–436.
[10] Wu Rong-Ching and Tsao Ta-Peng. The optimization of spectrum analysis fordigital signals. IEEE Transactions on Power Delivery, 2003, 18(2):398–405.
[11] Yang Yawpo and Liu Ching-Hwa. An asymptotic optimal algorithm for modu-lation classification. IEEE Communications Letters, 1998, 2(5):117–119.
[12] E. Jones, P. Runkle, and N. Dasgupta. Genetic algorithm wavelet design forsignal classification. IEEE Transactions on Pattern Analysis and Machine In-telligence, 2001, 23(8):890–895.
[13] D. Z.Vucic and M. M.Obradovic. Cyclic spectral analysis of phase-incoherentfsk signal by matrix-based stochastic method. Telecommunications in ModernSatellite, Cable and Broadcasting Service, 2001, 1:267–270.
[14] T. A.Drumright and Ding Zhi. A new algorithm for qam signal classification inawgn channels. IEEE International Symposium on Circuits and Systems, 2002,1:849–852.
[15]吕铁军,肖先锡.基于快速组合神经网络分类器的数字调制识别新方法.信号处理, 2001, 17(1):17–20.
[16] O. A.Dobre, Y.Bar-Ness, and Wei Su. Higher-order cyclic cumulants for highorder modulation classification. Military Communications Conference, 2003,1:112–117.
[17] J.A Sills. Maximum-likelihood modulation classification for psk/qam. MilitaryCommunications Conference Proceedings, 1999, 1:217–220.
[18] Huo Xiaoming and D. Donoho. A simple and robust modulation classificationmethod via counting. IEEE International Conference on Acoustics, Speech andSignal Processing, 1998, 6:3289–3292.
[19] K. Umebayash, S. Ishii, and R. Kohno. Blind adaptive estimation of modu-lation scheme for software defined radio. Personal, Indoor and Mobile RadioCommunications, 2000, 1:43–47.
[20] C. E. Poole Choe H. and A. M. Yu. Novel identification of intercepted signalsfrom unknown radio transmitters. SPIE, 1995, 2491:504–516.
[21] M. Barbeau Hall J. and E. Kranakis. Detection of transient in radio frequencyfingerprinting using phase characteristics of signals. In proceedings of the 3rdIASTED International Conference on Wireless and Optical Communications(WOC 2003), 2003 Pages = 13-18.
[22] J. Hall M. Barbeau and E. Kranakis. Enhancing intrusion detection in wirelessnetworks using radio frequency fingerprinting. Communications, Internet andInformation Technology (CIIT), 2004 Pages = 32 -38.
[23] Ellis K. and N. Serinken. Characteristics of radio transmitter fingerprints. RadioScience, 2001, 36:585–597.
[24] Fox D. and et al. Bayesian filtering for location estimation. Pervasive Comput-ing, 2003, 24:24–33.
[25] Shaw D. and W. Kinsner. Multifractal modelling of radio transmitter transientsfor classification. IEEE Communications Power and Computing, 1997 Pages =306–312.
[26] Ureten O. and N. Serinken. Detection of radio transmitter turn-on transients.Electronic Letters, 1999, 35:1996–1997.
[27] Serinken N. and O. Ureten. Bayesian detection of wi-fi transmitter rf finger-prints. Electronic Letters, 2005, 41(6):373–374.
[28] K. C. Ho W. Prokopiw et a. Modulation identification of digital signal by thewavelet transfom. IEE Proc. Radar, Sonar Navig, 2000, 147(4):169–176.
[29] W.Gillespie Bradford and L. Atlas. optimizing time-frequency kernes for clas-sification. IEEE Trans. on Siganl Processing, 2001, 49(3):485–495.
[30] J.Chen and W.Kinsner. Multifractal analysis of transients in power systems.2000 Canadian Conference on Electrical and Computer Engineering, 2000,1:307–311.
[31] Sun L. and W. Kinsner. Characterization and feature extraction of transientsignals using multifractal measures. IEEE Canadian Conference on Electricaland Computer Engineering, 1999, 2:781–785.
[32] O. Ureten Tekbas O. and N. Serinken. Improvement of transmitter identificationsystem for low snr transients. Electronic Letters, 2004, 40(3):182–183.
[33]王伦文,钟子发. 2fsk信号“指纹”特征的研究.电讯技术, 2003, 3:45–48.
[34]蔡忠伟.通信信号指纹识别技术研究.无线电工程, 2004, 1:54-57.
[35]王伦文,王铭三,杨俊安.电台“指纹”特征初探.计算机与网络, 2001,10:31-33.
[36] S.M.Kay and S. L. Marple. Spectrum analysis—a modern perspective. Proc.of IEEE, 1981, 69(3):1380–1419.
[37] S. Kay. A fast and accurate single frequency estimator. IEEE Tran. on ASSP,1989, 37(12):1987–1990.
[38] M. P. Fitz. Further results in the fast estimation of a single frequency. IEEETrans. on Commun., 1994, 42(3):862–864.
[39] M. Luise and Regiannini R. Carrier frequency recovery in all-digital modems forburst-mode transimissions. IEEE Trans. on Commun., 1995, 43(2):1169–1178.
[40] U. Mangali and Morelli M. Data-aided frequency estimation for burst digitaltransmission. IEEE.Trans. on Commun, 1997, 45(1):23–25.
[41] M. L. Fowler and Johnson J. A. Extending the threshold and frequency rangefor phase-based frequency estimation. IEEE Trans. Signal Process., 1999,47(10):2857–2863.
[42] Volcker, B., and Handel P. Frequency estimation from proper sets of correla-tions. IEEE Trans. Signal Process., 2002, 50(4):791–802.
[43] Frankie Kit Wing Chan H. C. So. A generalized weighted linear predictorfrequency estimation approach for a complex sinusoid. IEEE Trans. SignalProcess., 2006, 54(4):1304–1315.
[44] Brown Tyler and Wang Michael Mao. An iterative algorithm for single-frequency estimation. IEEE Trans. Signal Process., 2002, 50(11):2671–2682.
[45] M. Macleod. Fast nearly ml estimation of the parameters of real or complexsingle tones or resolved multiple tones. IEEE Trans. Signal Processing, 1998,46(1):141–148.
[46] P. Voglewede. Parabola approximation for peak determination. Global DSPMag., 2004, 3(5):13–17.
[47] Dusan AgreZ. Weighted multipoint interpolated dft to improve amplitude es-timation of multifrequency signal. IEEE Trans. On Instrumentation and Mea-surement, 2002, 51(2):287–292.
[48] Eric Jacobsen and Kootsookos Peter. Fast, accurate frequency estimators. IEEESignal Processing Magazine, 2007 Pages = 123-125.
[49] Romeo Ortega Liu Hsu and Damm Gilney. A globally convergent frequencyestimator. IEEE Trans. on Automatic Control, 1999, 41(4):688–715.
[50]丁康陈健林苏向荣.平稳和非平稳振动信号的若干处理方法及发展.振动工程学报, 2003, 16(1):1–10.
[51]时宇,张贤达.基于局部双谱的高分辨距离像雷达目标识别.清华大学学报(自然科学版), 2002, 42(3):407–410.
[52]姚天任.现代数字信号处理.华中科技大学出版社, 2001.
[53] K Tugnait J. Dection of no-gaussian signals using integrated polyspectrum.IEEE Trans SP, 1994, 42(12):3137–3149.
[54]张贤达.时间序列分析一一高阶统计量方法.清华大学出版社, 1996.
[55]马君国肖怀铁李保国朱江.基于局部围线积分双谱的空间目标识别算法.系统工程与电子技术, 2005, 27(8):1490–1493.
[56] V Chandran and Elgar S. L. Pattern recognition using invariants definedfrom higher order spectra――one-dimensional inputs. IEEE Trans SP, 1993,41(1):205– 212.
[57] K Tsatsanis M. and Giannakis G. B. Object and texture classification usinghigher order statistics. IEEE Trans Pattern Anal and Machine Intel, 1992,14(7):733– 750.
[58] Liao X Bao Z. Circularly integrated bispectra: Novel shift invariant featuresfor high-resolution radar target recognition. Elect Letters, 1998, 34(19):1879–1880.
[59] Zhang Xian-Da. A new feature vector using selected bispectrum for signalclassification with application in radar target recognition. IEEE Trans. SignalProcessing, 2001, 49(3):1875–1885.
[60]张贤达,保铮.非平稳信号分析与处理.国防工业出版社, 1998.
[61] Richaro G Baraniuk Douglas L. Jones. Signal-dependent time-frequency repre-sentation. Signal Processing, 1993, 32:263–284.
[62] G Mallat S. and Zhang Z. Matching pursuits with time-frequency dictionaries.IEEE Transactions on signal processing, 1993, 41(12):3397–3415.
[63] Shen Z E Huang N. and Long S. R. The empirical mode decomposition and thehilbert spectrum for nonlinear and non-stationary time series analysis. Proceed-ings of the Royal Society of London Series, 1998, 454:903–995.
[64] Shen Z E Huang N. and Long S. R. A new view of nonlinear water waves: thehilbert spectrum. Annual Review of Fluid Mechanics, 1999, 31:417–457.
[65] Norden E. huang Hwang Paul A. and David W. Wang. A note on analyz-ing nonlinear and nonstationary ocean data. Applied Ocean Research, 2003,25(4):187–193.
[66] Veltcheva A. D. and C. Guedes Soares. Identification of the components of wavespectra by the hilbert huang transform method. Applied Ocean Research, 2004,26(1):1一12.
[67] Datig Marcus and Torsten Schlurmann. Performance and limitations of thehilbert-huang transformation (hht) with an application to irregular water waves.Ocean Engineering, 2004,31(14):, 2004, 31(14):1783–1834.
[68]边肇祺,张学工.模式识别(第二版).清华大学出版社, 2000.
[69] Petre E. Hard Duda Richard O. and David G. Stock. Pattern classification.北京:机械工业出版社, 2003.
[70] V. N. Vapnik张学工译.统计学习理论的本质.清华大学出版社, 2000.
[71] Keerthi S. S. and E. G. Gilbert. Convergence of a generalized smo algorithmfor svm classifier design. Machine Learning, 2002, 46(1):351–360.
[72] A. Smola Scholkopf B. and P. L. Bartlett. New support vector algorithms.Neural Computation, 2000, 12(5):1207–1245.
[73] Chang C. C. and C. J. Lin. Training v-supportvector classifiers: theory andalgorithms. Neural Computation, 2001, 13(9):2119–2147.
[74] Lee Y. J. and O. L. Mangasarian. Rsvm: Reduced support vector machines.Proceedings of the First SIAM International Conference on Data Mining, 2001,3:321–328.
[75] Tax D. and R. Duin. Support vector domain description. Pattern RecognitionLetters, 1999, 20(11):1191–1199.
[76] R. Bogener Chew H. and C. Lm. Dual v-support vector machine with error rateand training size beasing. Proceedings of IEEE Int. Conf. on Acoustics, Speechand Signal Processing, 2001, 3:1269–1272.
[77] Suyken S. and J. Vandewale. Least squares support vector machine classifiers.Neural Processing Letters, 1999, 9(3):293–300.
[78] Lin C. and S. Wan. Fuzzy support vector machines. IEEE Trans. on NeuralNetworks, 2002, 13(2):464–471.
[79] Hsu C. W. and C. J. Lin. A comparison of methods for multi-class supportvector machines. IEEE Transactions on Neural Networks, 2002, 13(2):415–425.
[80] Danjou A M Grana and Albizuri F. X. Experiments of fast learning with highorder boltzmann machines. Applied Intelligence, 1997, 7(4):287–303.
[81] Bilski J Rutkowski L. A fast training algorithm for neural networks. IEEETransactions on Circuits and Systems II:Analog and Digital Signal Processing,1998, 45(6):123–140.
[82] Cogswell R J Barhen and Protopopescu V. Single-iteration training algorithmfor multi-layer feed-forward neural networks. Neural Processing Letters, 2000,11(2):113–129.
[83] Siu yeung Cho and Chow Tommy W. S. Training multilayer neural networksusing fast global learning algorithm least-squares and penalized optimizationmethods. Neurocomputing, 1999, 25:115–131.
[84]张玉兴.射频摸拟电路.电子工业出版社, 2001.
[85]黄载禄,殷卫华,黄本雄.通信原理.科学出版社, 2007.
[86] Razavi B. A study of phase noise in cmos oscillators. IEEE Publisher of Solid-State Circuits, 1996, 31(3):331–343.
[87]张厥盛,郑继禹,万心平.锁相技术.西安电子科技大学出版社, 2002.
[88] W.Couch Leon.数字与模拟通信系统.电子工业出版社, 2003.
[89]汪紫荆.调频接收机静噪电路的原理和设计.移动通信, 1999, 4:21–24.
[90] Tsui著,杨小牛,陆安南,金鹰译James.宽带数字接收机.电子工业出版社,2002.
[91] Hornik K. Approximation capabilities of multilayer feedforward networks. Neu-ral Netw., 1991, 4:251–257.
[92] Leshno L and Lin V Y. Multilayer feedforward networks with a nonpolyno-mial activation function can approximate any function. Neural Netw., 1993,6:861–867.
[93] Park J and Sandberg I W. Universal approximation using radial-basis-functionnetworks. Neural Comput., 1991, 3:246–257.
[94] Huang G B and Babri H A. Upper bounds on the number of hidden neuronsin feedforward networks with arbitrary bounded nonlinear activation functions.IEEE Trans. Neural Netw., 1998, 9(1):224–229.
[95] Kwok T Y and Yeung D Y. Objective functions for training new hiddenunits in constructive neural networks. IEEE Trans. Neural Netw., 1997,8(5):1131–1148.
[96] Romero E. . Function approximation with saocif: a general sequen-tial method and a partivular algorithm with feedforward neural networks.http://www.lsi.upc.es/dept/techreps, 2001.
[97] Baum E. . On the capabilities of multilayer perceptrons. J. Complex., 2004,40(4):193–215.
[98] S. Ferrari and R.F. Stengel. Smooth function approximation using neural net-works. IEEE Trans. Neural Networks, 2005, 16(1):24–38.
[99] Sartori M A. . A simple method to derive bounds on the size and to trainmultilayer neural networks. IEEE Trans. Neural Netw., 1991, 50(2):467–471.
[100] Tammura S and Tateishi M. . Capabilities of a four-layered feedforwardneural network:four layers versus three. IEEE Trans. Neural Netw., 1997,8(2):251–255.
[101] Huang G B. Learning capability and storage capacity of two-hidden layer feed-forward networks. IEEE Trans. Neural Netw., 2003, 14(2):274–281.
[102] Huang G B and Siew C K. . Extreme learning machine with randomly assignedrbf kernels. Int. J. Inf. Technol., 2005, 11(1):16–24.
[103] Huang G B and Zhu Q Y. Extreme learning machine: theory and applications.Neurocomp., 2006, 70:489–501.
[104] G.B.Huang, L.Chen, and Chee-Kheong. Universal approximation using incre-mental constructive feedforward networks with random hidden nodes. IEEETrans. Neural Netw., Jul. 2006, 17(4):879–892.
[105] G.B.Huang and L.Chen. Convex incremental extreme learning machine. Neu-rocomp., to appear in Mar. 2007.
[106] N.Y.Liang, G.B.Huang, and P. Saratchandran. A fast accurate online sequentiallearning algorithm for feedforward networks. IEEE Trans. Neural Netw., Nov.2006, 17(6):1411–1422.
[107] P.L.Bartlett. The sample complexity of pattern classification with neural net-works: the size of the weights is more important than the size of the network.IEEE Trans. Inf. Theory, Nov. 1998, 44(2):525–536.
[108]王松桂,杨振海.广义逆矩阵及其应用.北京工业大学出版社, 1996.
[109] D.Serre. Matrices:theory and applications. Springer,NewYork, 2002.
[110] Li M.B., Huang G.B., and P. Saratchandran. Fully complex extreme learningmachine. Neurocomputing, 2005, 68:306–314.
[111] Hsue S. Z. and S.S.Soliman. Automatic modulation classification using zero-crossing. IEEE Proceedings, 1990, 6:459–464.
[112] Gardner F. M. Interpolation in digital modems- i: Fundamentals. IEEE Trans-actions on Communications, 1993, 3:501–507.
[113] L.Erup, F. M. Gardner, and Robert A. Hands. Interpolation in digital modems.ii. implementation and performance. IEEE Transactions on Communications,1993, 41(6):998–1005.
[114] O Shea P. A fast algorithm for estimating the parameters of a quadratic fmsignal. IEEE Trans. on Signal Processing, 2004, 52(2):385–393.
[115] K Lee and Kim Y. A new symbol timing recovery for all-digital high-speedsymbol synchronization. IEEE Trans. Common, 1997, 80(9):1290–1298.
[116]张晓冬,王桥,吴乐南.利用脊的特征进行信号盲分离. 2004, 32(7):1156–11590.
[117] Jinghuai G Xiaolong D. and Wen-Bing W. Instantaneous parameters extractionvia wavelet transform. IEEE Trans. on Geoscience and Remote Sensing, 1999,37(2):867–870.
[118] K.C.Ho, W.Prokipiw, and Y.T.Chan. Modulation identification of digital sig-nals by the wavelet transform. IEE Proc-Radar.Sonar Navig, 2000, 4:169–176.
[119] S.Chandrasekhar and T. V. Sreenivas. Instantaneous frequency estimation usinglevel-crossing information. 2003, 4:141一144.
[120] B Boashash. Estimating and interpreting the instantaneous frequency of asignal. i—fundamentals. Proc. of IEEE, 1992, 80(4):520–538.
[121] Nuttall A. On the quadrature approximation to the hilbert transform of mod-ulated signals. Proc. of IEEE, 1966, 54:1458–1462.
[122] D. Slepian H. Landan and H. Pollak. Prolate spheroidal wave functions—fourieranalysis and uncertainty, i ,ii and iii. Bell Syst. Tech. J., 1961, 40(1):43–84.
[123] K.Wmartin. Complex signal processing is not complex. IEEE Trans. On Cir-cuits and System, 2004, 9:1823–1836.
[124] K Tong L., Guanghan X., and Thomas. Blind identification and equalizationbased on second-order statistics: A time domain approach. 1995, 40(2):340–349.
[125] G Tong L. and Xu. Blind channel identification based on second-order statistics:a frequency-domain approach. 1995, 41(1):329–334.
[126] A Reed D. E. and Wckert M. Minimization of detection of symbol-rate spectrallines by delay and multiply receivers. 1988, 36:118–120.
[127] hertner A Solve C. Symbol-rate timing recovery comprising the optimum signal-to-noise ratio in a digital subscriber loop. IEEE Trans. Commun, 45(8):925–936.
[128] A Scheper R. and Teolis J. A. cramer-rao bounds for wavelet transform basedinstantaneous frequency estimates. 2003, 51(6):1593一1603.
[129] Y.T Chan. Symbol rate estimation by the wayelet transform. 1997, 1:177–180.
[130] Koh B. S. and H. S. Lee. Detection of symbol rate of unknown digital commu-nication signals. Electronic Letters, 1993, 29(3):278–279.
[131] Fang T. T. I and q decomposition of self-noise in square-law clock regenerators.IEEE Trans. On Commun., 1988, 36:1044–1052.
[132]齐国清,贾欣乐.基于dft相位的正弦波频率和初相的高精度估计方法.电子学报, 2001, 29(9):1164– 1167.
[133] Rife D C and Boorstyn R. R. Single tone parameter estimation from discretetime observation. IEEE Trans. Info. Theory, 1974, 20(5):591–598.
[134] V. M. Baronkin Y. V. Zakharov and Tozer T. C. Dft-based frequency estima-tors with narrow acquisition range. Proc. Inst. Elect. Eng.―Commun., 2001,148:1–7.
[135] Y. V. Zakharov and Tozer T. C. Frequency estimator with dichotomous searchof periodogram peak. Electron. Lett., 1999, 35:1608–1609.
[136] Elias Aboutanios. A modified dichotomous search frequency estimator. IEEESignal Processing Letters, 2004, 11(2):186–188.
[137] Rife D C and Vincent G. A. Use of the discrete fourier transform in the measure-ment of frequencies and levels of tones. Bell. Sys. Tech. J., 1970, 49(2):197–228.
[138] Collins W. L. Jr Davis D. C. Jane V K. High-accuracy analog measurementsvia interpolated ?t. IEEE Trans. IM, 1979, 28(2):113–122.
[139] Quinn B. G. Estimation of frequency, amplitude and phase from the dft of atime series. IEEE Trans. SP, 1997, 45(3):814–817.
[140]贾欣乐齐国清.插值?t估计正弦信号频率的精度分析.电子学报, 2004,32(4):625-629.
[141] Z. G. Chen An H. Z. and E. J. Hannan. The maximum of the periodogram. J.Multivariate Anal., 1983, 13:383–400.
[142] G Quinn B. Estimating frequency by interpolation using fourier coe?cients.IEEE Trans. On Signal Processing, 1994, 42(5):1264–1268.
[143] N Kwak and Choi C. H. Input feature selection by mutual information basedon parzen window. IEEE Trans. on Pattern Analysis and Machine Intelligence,2002, 24(12):1667一1671.
[144] R Meo. Maximum independence and mutual information. IEEE Trans. onInformation Theory, 2002, 48(1):318一324.
[145] R Battiti. Using mutual information for selecting features in supervised neuralnet learning. IEEE Trans. on Neural Networks, 1994, 5(4):537–550.
[146] K. R. Muller S. Mika G. Ratsch K. Tsuda and B. Scholkopf. An introduction tokernel-based learning algorithm. IEEE Transactions on Neural Networks, 2001,12(2):181–201.
[147] Chapelle O. Choosing kemel parameters for support vector machines. MachineLearning, 2002, 46(1):131–159.
[148] Platt J. A resource-allocating network for function interpolation. Neural Com-putat., 1991, 3:213–225.
[149] Kadirkamanathan V. and M. Niranjan. A function estimation approach tosequential learning with neural networks. Neural Computat., 1993, 5:954–975.
[150] N. Sundararajan Yingwei L. and P. Saratchandran. A sequental learning schemefor function approximation using minimal radial basis function (rbf) neuralnetworks. Neural Computat., 1997, 9:461–478.
[151] N. Sundararajan Yingwei L. and P. Saratchandran. Performance evaluation of asequental minimal radial basis function (rbf) neural network learning algorithm.IEEE Trans. Neural Netw., 1998, 9(2):308–318.
[152] C. G. Puntonet J. Ortega Salmerón M. and A. Prieto. Improved ran sequentialprediction using orthogonal techniques. Neurocomput., 2001, 41:153–172.
[153] I. Rojas H. Pomares and A. Prieto. Time series analysis using normalized pg-rbfnetwork with regression weights. Neurocomput., 2002, 42:267–285.
[154] P. Saratchandran Huang G. B. and N. Sundararajan. An e?cient sequentiallearning algorithm for growing and pruning rbf (gap-rbf) networks. IEEE Trans.Syst., Man, Cybern., B, Cybern., 2004, 34(6):2284–2292.
[155] P. Saratchandran Huang G. B. and N. Sundararajan. A generalized growingand pruning rbf (ggap-rbf) neural network for function approximation. IEEETrans. Neural Netw., 2005, 16(1):57–67.
[156] Lee K Kim Y. A new symbol timing recovery for all-digital high-speed symbolsynchronization. IEEE Trans. Common., 1997, 80(9):1290–1298.
[2]刘刚.雷达指纹分析的基本理论探讨.电子对抗, 2002, 6:1–6.
[3]牛海,马颖.小波一神经网络在辐射源识别中的应用研究.系统工程与电子技术, 2002, 124(15):55–57.
[4] V.Desimir and O.Milorad. Spectral correlation of psk signals. Telecommunica-tions in Modern Satellite, Cable and Broadcasting Services, 1999, 1:273–276.
[5] W. Gardner, W.Brown, and Chih-Kang Chen. Spectral correlation of modulatedsignals: Part ii–digital modulation. IEEE Transactions on Communications,1987, 35(6):595–601.
[6]陈卫东,杨绍全.基于循环累量不变量的mpsk信号调制识别算法.电子与信息学报, 2003, 25(3):320–325.
[7] A. K.Jain and P.Dubin. Statistical pattern recognition: a review. IEEE Transon Pattern Analysis and Machine Intelligence, 2000, 22(1):4–37.
[8] A. Swami and B. Sadler. Hierarchical digital modulation classification usingcumulants. IEEE Transactions on Communications, 2000, 48(3):416–429.
[9] A. K.Nandi and E.E. Azzouz. Algorithms for automatic modulation recognitionof communication signals. IEEE Transactions on Communications, 1998, , Vol:46(4):431–436.
[10] Wu Rong-Ching and Tsao Ta-Peng. The optimization of spectrum analysis fordigital signals. IEEE Transactions on Power Delivery, 2003, 18(2):398–405.
[11] Yang Yawpo and Liu Ching-Hwa. An asymptotic optimal algorithm for modu-lation classification. IEEE Communications Letters, 1998, 2(5):117–119.
[12] E. Jones, P. Runkle, and N. Dasgupta. Genetic algorithm wavelet design forsignal classification. IEEE Transactions on Pattern Analysis and Machine In-telligence, 2001, 23(8):890–895.
[13] D. Z.Vucic and M. M.Obradovic. Cyclic spectral analysis of phase-incoherentfsk signal by matrix-based stochastic method. Telecommunications in ModernSatellite, Cable and Broadcasting Service, 2001, 1:267–270.
[14] T. A.Drumright and Ding Zhi. A new algorithm for qam signal classification inawgn channels. IEEE International Symposium on Circuits and Systems, 2002,1:849–852.
[15]吕铁军,肖先锡.基于快速组合神经网络分类器的数字调制识别新方法.信号处理, 2001, 17(1):17–20.
[16] O. A.Dobre, Y.Bar-Ness, and Wei Su. Higher-order cyclic cumulants for highorder modulation classification. Military Communications Conference, 2003,1:112–117.
[17] J.A Sills. Maximum-likelihood modulation classification for psk/qam. MilitaryCommunications Conference Proceedings, 1999, 1:217–220.
[18] Huo Xiaoming and D. Donoho. A simple and robust modulation classificationmethod via counting. IEEE International Conference on Acoustics, Speech andSignal Processing, 1998, 6:3289–3292.
[19] K. Umebayash, S. Ishii, and R. Kohno. Blind adaptive estimation of modu-lation scheme for software defined radio. Personal, Indoor and Mobile RadioCommunications, 2000, 1:43–47.
[20] C. E. Poole Choe H. and A. M. Yu. Novel identification of intercepted signalsfrom unknown radio transmitters. SPIE, 1995, 2491:504–516.
[21] M. Barbeau Hall J. and E. Kranakis. Detection of transient in radio frequencyfingerprinting using phase characteristics of signals. In proceedings of the 3rdIASTED International Conference on Wireless and Optical Communications(WOC 2003), 2003 Pages = 13-18.
[22] J. Hall M. Barbeau and E. Kranakis. Enhancing intrusion detection in wirelessnetworks using radio frequency fingerprinting. Communications, Internet andInformation Technology (CIIT), 2004 Pages = 32 -38.
[23] Ellis K. and N. Serinken. Characteristics of radio transmitter fingerprints. RadioScience, 2001, 36:585–597.
[24] Fox D. and et al. Bayesian filtering for location estimation. Pervasive Comput-ing, 2003, 24:24–33.
[25] Shaw D. and W. Kinsner. Multifractal modelling of radio transmitter transientsfor classification. IEEE Communications Power and Computing, 1997 Pages =306–312.
[26] Ureten O. and N. Serinken. Detection of radio transmitter turn-on transients.Electronic Letters, 1999, 35:1996–1997.
[27] Serinken N. and O. Ureten. Bayesian detection of wi-fi transmitter rf finger-prints. Electronic Letters, 2005, 41(6):373–374.
[28] K. C. Ho W. Prokopiw et a. Modulation identification of digital signal by thewavelet transfom. IEE Proc. Radar, Sonar Navig, 2000, 147(4):169–176.
[29] W.Gillespie Bradford and L. Atlas. optimizing time-frequency kernes for clas-sification. IEEE Trans. on Siganl Processing, 2001, 49(3):485–495.
[30] J.Chen and W.Kinsner. Multifractal analysis of transients in power systems.2000 Canadian Conference on Electrical and Computer Engineering, 2000,1:307–311.
[31] Sun L. and W. Kinsner. Characterization and feature extraction of transientsignals using multifractal measures. IEEE Canadian Conference on Electricaland Computer Engineering, 1999, 2:781–785.
[32] O. Ureten Tekbas O. and N. Serinken. Improvement of transmitter identificationsystem for low snr transients. Electronic Letters, 2004, 40(3):182–183.
[33]王伦文,钟子发. 2fsk信号“指纹”特征的研究.电讯技术, 2003, 3:45–48.
[34]蔡忠伟.通信信号指纹识别技术研究.无线电工程, 2004, 1:54-57.
[35]王伦文,王铭三,杨俊安.电台“指纹”特征初探.计算机与网络, 2001,10:31-33.
[36] S.M.Kay and S. L. Marple. Spectrum analysis—a modern perspective. Proc.of IEEE, 1981, 69(3):1380–1419.
[37] S. Kay. A fast and accurate single frequency estimator. IEEE Tran. on ASSP,1989, 37(12):1987–1990.
[38] M. P. Fitz. Further results in the fast estimation of a single frequency. IEEETrans. on Commun., 1994, 42(3):862–864.
[39] M. Luise and Regiannini R. Carrier frequency recovery in all-digital modems forburst-mode transimissions. IEEE Trans. on Commun., 1995, 43(2):1169–1178.
[40] U. Mangali and Morelli M. Data-aided frequency estimation for burst digitaltransmission. IEEE.Trans. on Commun, 1997, 45(1):23–25.
[41] M. L. Fowler and Johnson J. A. Extending the threshold and frequency rangefor phase-based frequency estimation. IEEE Trans. Signal Process., 1999,47(10):2857–2863.
[42] Volcker, B., and Handel P. Frequency estimation from proper sets of correla-tions. IEEE Trans. Signal Process., 2002, 50(4):791–802.
[43] Frankie Kit Wing Chan H. C. So. A generalized weighted linear predictorfrequency estimation approach for a complex sinusoid. IEEE Trans. SignalProcess., 2006, 54(4):1304–1315.
[44] Brown Tyler and Wang Michael Mao. An iterative algorithm for single-frequency estimation. IEEE Trans. Signal Process., 2002, 50(11):2671–2682.
[45] M. Macleod. Fast nearly ml estimation of the parameters of real or complexsingle tones or resolved multiple tones. IEEE Trans. Signal Processing, 1998,46(1):141–148.
[46] P. Voglewede. Parabola approximation for peak determination. Global DSPMag., 2004, 3(5):13–17.
[47] Dusan AgreZ. Weighted multipoint interpolated dft to improve amplitude es-timation of multifrequency signal. IEEE Trans. On Instrumentation and Mea-surement, 2002, 51(2):287–292.
[48] Eric Jacobsen and Kootsookos Peter. Fast, accurate frequency estimators. IEEESignal Processing Magazine, 2007 Pages = 123-125.
[49] Romeo Ortega Liu Hsu and Damm Gilney. A globally convergent frequencyestimator. IEEE Trans. on Automatic Control, 1999, 41(4):688–715.
[50]丁康陈健林苏向荣.平稳和非平稳振动信号的若干处理方法及发展.振动工程学报, 2003, 16(1):1–10.
[51]时宇,张贤达.基于局部双谱的高分辨距离像雷达目标识别.清华大学学报(自然科学版), 2002, 42(3):407–410.
[52]姚天任.现代数字信号处理.华中科技大学出版社, 2001.
[53] K Tugnait J. Dection of no-gaussian signals using integrated polyspectrum.IEEE Trans SP, 1994, 42(12):3137–3149.
[54]张贤达.时间序列分析一一高阶统计量方法.清华大学出版社, 1996.
[55]马君国肖怀铁李保国朱江.基于局部围线积分双谱的空间目标识别算法.系统工程与电子技术, 2005, 27(8):1490–1493.
[56] V Chandran and Elgar S. L. Pattern recognition using invariants definedfrom higher order spectra――one-dimensional inputs. IEEE Trans SP, 1993,41(1):205– 212.
[57] K Tsatsanis M. and Giannakis G. B. Object and texture classification usinghigher order statistics. IEEE Trans Pattern Anal and Machine Intel, 1992,14(7):733– 750.
[58] Liao X Bao Z. Circularly integrated bispectra: Novel shift invariant featuresfor high-resolution radar target recognition. Elect Letters, 1998, 34(19):1879–1880.
[59] Zhang Xian-Da. A new feature vector using selected bispectrum for signalclassification with application in radar target recognition. IEEE Trans. SignalProcessing, 2001, 49(3):1875–1885.
[60]张贤达,保铮.非平稳信号分析与处理.国防工业出版社, 1998.
[61] Richaro G Baraniuk Douglas L. Jones. Signal-dependent time-frequency repre-sentation. Signal Processing, 1993, 32:263–284.
[62] G Mallat S. and Zhang Z. Matching pursuits with time-frequency dictionaries.IEEE Transactions on signal processing, 1993, 41(12):3397–3415.
[63] Shen Z E Huang N. and Long S. R. The empirical mode decomposition and thehilbert spectrum for nonlinear and non-stationary time series analysis. Proceed-ings of the Royal Society of London Series, 1998, 454:903–995.
[64] Shen Z E Huang N. and Long S. R. A new view of nonlinear water waves: thehilbert spectrum. Annual Review of Fluid Mechanics, 1999, 31:417–457.
[65] Norden E. huang Hwang Paul A. and David W. Wang. A note on analyz-ing nonlinear and nonstationary ocean data. Applied Ocean Research, 2003,25(4):187–193.
[66] Veltcheva A. D. and C. Guedes Soares. Identification of the components of wavespectra by the hilbert huang transform method. Applied Ocean Research, 2004,26(1):1一12.
[67] Datig Marcus and Torsten Schlurmann. Performance and limitations of thehilbert-huang transformation (hht) with an application to irregular water waves.Ocean Engineering, 2004,31(14):, 2004, 31(14):1783–1834.
[68]边肇祺,张学工.模式识别(第二版).清华大学出版社, 2000.
[69] Petre E. Hard Duda Richard O. and David G. Stock. Pattern classification.北京:机械工业出版社, 2003.
[70] V. N. Vapnik张学工译.统计学习理论的本质.清华大学出版社, 2000.
[71] Keerthi S. S. and E. G. Gilbert. Convergence of a generalized smo algorithmfor svm classifier design. Machine Learning, 2002, 46(1):351–360.
[72] A. Smola Scholkopf B. and P. L. Bartlett. New support vector algorithms.Neural Computation, 2000, 12(5):1207–1245.
[73] Chang C. C. and C. J. Lin. Training v-supportvector classifiers: theory andalgorithms. Neural Computation, 2001, 13(9):2119–2147.
[74] Lee Y. J. and O. L. Mangasarian. Rsvm: Reduced support vector machines.Proceedings of the First SIAM International Conference on Data Mining, 2001,3:321–328.
[75] Tax D. and R. Duin. Support vector domain description. Pattern RecognitionLetters, 1999, 20(11):1191–1199.
[76] R. Bogener Chew H. and C. Lm. Dual v-support vector machine with error rateand training size beasing. Proceedings of IEEE Int. Conf. on Acoustics, Speechand Signal Processing, 2001, 3:1269–1272.
[77] Suyken S. and J. Vandewale. Least squares support vector machine classifiers.Neural Processing Letters, 1999, 9(3):293–300.
[78] Lin C. and S. Wan. Fuzzy support vector machines. IEEE Trans. on NeuralNetworks, 2002, 13(2):464–471.
[79] Hsu C. W. and C. J. Lin. A comparison of methods for multi-class supportvector machines. IEEE Transactions on Neural Networks, 2002, 13(2):415–425.
[80] Danjou A M Grana and Albizuri F. X. Experiments of fast learning with highorder boltzmann machines. Applied Intelligence, 1997, 7(4):287–303.
[81] Bilski J Rutkowski L. A fast training algorithm for neural networks. IEEETransactions on Circuits and Systems II:Analog and Digital Signal Processing,1998, 45(6):123–140.
[82] Cogswell R J Barhen and Protopopescu V. Single-iteration training algorithmfor multi-layer feed-forward neural networks. Neural Processing Letters, 2000,11(2):113–129.
[83] Siu yeung Cho and Chow Tommy W. S. Training multilayer neural networksusing fast global learning algorithm least-squares and penalized optimizationmethods. Neurocomputing, 1999, 25:115–131.
[84]张玉兴.射频摸拟电路.电子工业出版社, 2001.
[85]黄载禄,殷卫华,黄本雄.通信原理.科学出版社, 2007.
[86] Razavi B. A study of phase noise in cmos oscillators. IEEE Publisher of Solid-State Circuits, 1996, 31(3):331–343.
[87]张厥盛,郑继禹,万心平.锁相技术.西安电子科技大学出版社, 2002.
[88] W.Couch Leon.数字与模拟通信系统.电子工业出版社, 2003.
[89]汪紫荆.调频接收机静噪电路的原理和设计.移动通信, 1999, 4:21–24.
[90] Tsui著,杨小牛,陆安南,金鹰译James.宽带数字接收机.电子工业出版社,2002.
[91] Hornik K. Approximation capabilities of multilayer feedforward networks. Neu-ral Netw., 1991, 4:251–257.
[92] Leshno L and Lin V Y. Multilayer feedforward networks with a nonpolyno-mial activation function can approximate any function. Neural Netw., 1993,6:861–867.
[93] Park J and Sandberg I W. Universal approximation using radial-basis-functionnetworks. Neural Comput., 1991, 3:246–257.
[94] Huang G B and Babri H A. Upper bounds on the number of hidden neuronsin feedforward networks with arbitrary bounded nonlinear activation functions.IEEE Trans. Neural Netw., 1998, 9(1):224–229.
[95] Kwok T Y and Yeung D Y. Objective functions for training new hiddenunits in constructive neural networks. IEEE Trans. Neural Netw., 1997,8(5):1131–1148.
[96] Romero E. . Function approximation with saocif: a general sequen-tial method and a partivular algorithm with feedforward neural networks.http://www.lsi.upc.es/dept/techreps, 2001.
[97] Baum E. . On the capabilities of multilayer perceptrons. J. Complex., 2004,40(4):193–215.
[98] S. Ferrari and R.F. Stengel. Smooth function approximation using neural net-works. IEEE Trans. Neural Networks, 2005, 16(1):24–38.
[99] Sartori M A. . A simple method to derive bounds on the size and to trainmultilayer neural networks. IEEE Trans. Neural Netw., 1991, 50(2):467–471.
[100] Tammura S and Tateishi M. . Capabilities of a four-layered feedforwardneural network:four layers versus three. IEEE Trans. Neural Netw., 1997,8(2):251–255.
[101] Huang G B. Learning capability and storage capacity of two-hidden layer feed-forward networks. IEEE Trans. Neural Netw., 2003, 14(2):274–281.
[102] Huang G B and Siew C K. . Extreme learning machine with randomly assignedrbf kernels. Int. J. Inf. Technol., 2005, 11(1):16–24.
[103] Huang G B and Zhu Q Y. Extreme learning machine: theory and applications.Neurocomp., 2006, 70:489–501.
[104] G.B.Huang, L.Chen, and Chee-Kheong. Universal approximation using incre-mental constructive feedforward networks with random hidden nodes. IEEETrans. Neural Netw., Jul. 2006, 17(4):879–892.
[105] G.B.Huang and L.Chen. Convex incremental extreme learning machine. Neu-rocomp., to appear in Mar. 2007.
[106] N.Y.Liang, G.B.Huang, and P. Saratchandran. A fast accurate online sequentiallearning algorithm for feedforward networks. IEEE Trans. Neural Netw., Nov.2006, 17(6):1411–1422.
[107] P.L.Bartlett. The sample complexity of pattern classification with neural net-works: the size of the weights is more important than the size of the network.IEEE Trans. Inf. Theory, Nov. 1998, 44(2):525–536.
[108]王松桂,杨振海.广义逆矩阵及其应用.北京工业大学出版社, 1996.
[109] D.Serre. Matrices:theory and applications. Springer,NewYork, 2002.
[110] Li M.B., Huang G.B., and P. Saratchandran. Fully complex extreme learningmachine. Neurocomputing, 2005, 68:306–314.
[111] Hsue S. Z. and S.S.Soliman. Automatic modulation classification using zero-crossing. IEEE Proceedings, 1990, 6:459–464.
[112] Gardner F. M. Interpolation in digital modems- i: Fundamentals. IEEE Trans-actions on Communications, 1993, 3:501–507.
[113] L.Erup, F. M. Gardner, and Robert A. Hands. Interpolation in digital modems.ii. implementation and performance. IEEE Transactions on Communications,1993, 41(6):998–1005.
[114] O Shea P. A fast algorithm for estimating the parameters of a quadratic fmsignal. IEEE Trans. on Signal Processing, 2004, 52(2):385–393.
[115] K Lee and Kim Y. A new symbol timing recovery for all-digital high-speedsymbol synchronization. IEEE Trans. Common, 1997, 80(9):1290–1298.
[116]张晓冬,王桥,吴乐南.利用脊的特征进行信号盲分离. 2004, 32(7):1156–11590.
[117] Jinghuai G Xiaolong D. and Wen-Bing W. Instantaneous parameters extractionvia wavelet transform. IEEE Trans. on Geoscience and Remote Sensing, 1999,37(2):867–870.
[118] K.C.Ho, W.Prokipiw, and Y.T.Chan. Modulation identification of digital sig-nals by the wavelet transform. IEE Proc-Radar.Sonar Navig, 2000, 4:169–176.
[119] S.Chandrasekhar and T. V. Sreenivas. Instantaneous frequency estimation usinglevel-crossing information. 2003, 4:141一144.
[120] B Boashash. Estimating and interpreting the instantaneous frequency of asignal. i—fundamentals. Proc. of IEEE, 1992, 80(4):520–538.
[121] Nuttall A. On the quadrature approximation to the hilbert transform of mod-ulated signals. Proc. of IEEE, 1966, 54:1458–1462.
[122] D. Slepian H. Landan and H. Pollak. Prolate spheroidal wave functions—fourieranalysis and uncertainty, i ,ii and iii. Bell Syst. Tech. J., 1961, 40(1):43–84.
[123] K.Wmartin. Complex signal processing is not complex. IEEE Trans. On Cir-cuits and System, 2004, 9:1823–1836.
[124] K Tong L., Guanghan X., and Thomas. Blind identification and equalizationbased on second-order statistics: A time domain approach. 1995, 40(2):340–349.
[125] G Tong L. and Xu. Blind channel identification based on second-order statistics:a frequency-domain approach. 1995, 41(1):329–334.
[126] A Reed D. E. and Wckert M. Minimization of detection of symbol-rate spectrallines by delay and multiply receivers. 1988, 36:118–120.
[127] hertner A Solve C. Symbol-rate timing recovery comprising the optimum signal-to-noise ratio in a digital subscriber loop. IEEE Trans. Commun, 45(8):925–936.
[128] A Scheper R. and Teolis J. A. cramer-rao bounds for wavelet transform basedinstantaneous frequency estimates. 2003, 51(6):1593一1603.
[129] Y.T Chan. Symbol rate estimation by the wayelet transform. 1997, 1:177–180.
[130] Koh B. S. and H. S. Lee. Detection of symbol rate of unknown digital commu-nication signals. Electronic Letters, 1993, 29(3):278–279.
[131] Fang T. T. I and q decomposition of self-noise in square-law clock regenerators.IEEE Trans. On Commun., 1988, 36:1044–1052.
[132]齐国清,贾欣乐.基于dft相位的正弦波频率和初相的高精度估计方法.电子学报, 2001, 29(9):1164– 1167.
[133] Rife D C and Boorstyn R. R. Single tone parameter estimation from discretetime observation. IEEE Trans. Info. Theory, 1974, 20(5):591–598.
[134] V. M. Baronkin Y. V. Zakharov and Tozer T. C. Dft-based frequency estima-tors with narrow acquisition range. Proc. Inst. Elect. Eng.―Commun., 2001,148:1–7.
[135] Y. V. Zakharov and Tozer T. C. Frequency estimator with dichotomous searchof periodogram peak. Electron. Lett., 1999, 35:1608–1609.
[136] Elias Aboutanios. A modified dichotomous search frequency estimator. IEEESignal Processing Letters, 2004, 11(2):186–188.
[137] Rife D C and Vincent G. A. Use of the discrete fourier transform in the measure-ment of frequencies and levels of tones. Bell. Sys. Tech. J., 1970, 49(2):197–228.
[138] Collins W. L. Jr Davis D. C. Jane V K. High-accuracy analog measurementsvia interpolated ?t. IEEE Trans. IM, 1979, 28(2):113–122.
[139] Quinn B. G. Estimation of frequency, amplitude and phase from the dft of atime series. IEEE Trans. SP, 1997, 45(3):814–817.
[140]贾欣乐齐国清.插值?t估计正弦信号频率的精度分析.电子学报, 2004,32(4):625-629.
[141] Z. G. Chen An H. Z. and E. J. Hannan. The maximum of the periodogram. J.Multivariate Anal., 1983, 13:383–400.
[142] G Quinn B. Estimating frequency by interpolation using fourier coe?cients.IEEE Trans. On Signal Processing, 1994, 42(5):1264–1268.
[143] N Kwak and Choi C. H. Input feature selection by mutual information basedon parzen window. IEEE Trans. on Pattern Analysis and Machine Intelligence,2002, 24(12):1667一1671.
[144] R Meo. Maximum independence and mutual information. IEEE Trans. onInformation Theory, 2002, 48(1):318一324.
[145] R Battiti. Using mutual information for selecting features in supervised neuralnet learning. IEEE Trans. on Neural Networks, 1994, 5(4):537–550.
[146] K. R. Muller S. Mika G. Ratsch K. Tsuda and B. Scholkopf. An introduction tokernel-based learning algorithm. IEEE Transactions on Neural Networks, 2001,12(2):181–201.
[147] Chapelle O. Choosing kemel parameters for support vector machines. MachineLearning, 2002, 46(1):131–159.
[148] Platt J. A resource-allocating network for function interpolation. Neural Com-putat., 1991, 3:213–225.
[149] Kadirkamanathan V. and M. Niranjan. A function estimation approach tosequential learning with neural networks. Neural Computat., 1993, 5:954–975.
[150] N. Sundararajan Yingwei L. and P. Saratchandran. A sequental learning schemefor function approximation using minimal radial basis function (rbf) neuralnetworks. Neural Computat., 1997, 9:461–478.
[151] N. Sundararajan Yingwei L. and P. Saratchandran. Performance evaluation of asequental minimal radial basis function (rbf) neural network learning algorithm.IEEE Trans. Neural Netw., 1998, 9(2):308–318.
[152] C. G. Puntonet J. Ortega Salmerón M. and A. Prieto. Improved ran sequentialprediction using orthogonal techniques. Neurocomput., 2001, 41:153–172.
[153] I. Rojas H. Pomares and A. Prieto. Time series analysis using normalized pg-rbfnetwork with regression weights. Neurocomput., 2002, 42:267–285.
[154] P. Saratchandran Huang G. B. and N. Sundararajan. An e?cient sequentiallearning algorithm for growing and pruning rbf (gap-rbf) networks. IEEE Trans.Syst., Man, Cybern., B, Cybern., 2004, 34(6):2284–2292.
[155] P. Saratchandran Huang G. B. and N. Sundararajan. A generalized growingand pruning rbf (ggap-rbf) neural network for function approximation. IEEETrans. Neural Netw., 2005, 16(1):57–67.
[156] Lee K Kim Y. A new symbol timing recovery for all-digital high-speed symbolsynchronization. IEEE Trans. Common., 1997, 80(9):1290–1298.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |