摘要
出于一些特殊的目的,我们需要监视通信信号的活动状况,甚至破译它的通信内容,这些离不开调制分析这一关键技术。调制分析是通信双方以外的第三者以不明信号为对象,通过提取载波频率、符号速率等未知参数,最终辨明调制方式的信号分析过程。从目前公开发表的资料来看,调制分析可用于无线电侦察、非通信侦察、电子对抗、无线电频谱管理和软件无线电等领域。
调制分析与信号解调有很强的关联性,前者可以借鉴或沿用后者采用的多种方法和手段,这是因为两者的处理对象一致,所要完成的任务也有很多共同之处。然而与信号解调不同的是,调制分析中的载波频率、符号速率和调制方式都是未知因素,这一点使得调制分析更富有挑战性。
从实现过程来看,调制分析可分为参数提取、调制分类两个环节。由于具备分类特征提取、样本特征库构造和判决算法设计等基本步骤,以辨明信号调制方式为目的的调制分类通常被视为模式识别的一种典型应用,参数提取则被看作调制分类的辅助手段。从统计信号处理的角度看待调制分析,参数提取属于估计问题,调制分类属于二元或多元检测问题,在实现过程中两者往往相辅相成、没有固定的主从关系或先后顺序。
本文研究了包括BPSK、QPSK、8PSK、OQPSK(MSK,π/2BPSK)、π/4QPSK、16QAM、32QAM、64QAM、128QAM、256QAM在内的多种线性调制信号的参数提取、识别或分类问题,并力图采用谐波检测与估计(SWDE;Sinusoidal Wave Detection and Estimation)这一理论上最基本、应用上最广泛的信号处理手段贯穿线性调制分析的整个研究过程。
文中针对色噪声中的谐波信号首次引入局部信噪比的概念,与全局信噪比(谐波功率与噪声方差的比值)相比,新的定义能够真实地反映噪声干扰对信号质量的影响程度。文中在讨论线性调制信号统一模型的基础上,给出了信噪比(每符号能量与噪声功率谱密度的比值)的计算方法。基于计算机仿真的观测结果表明,线性调制分析中的背景噪声表现出有色特性,而且不服从高斯分布,正是背景噪声的非高斯有色特性增加了线性调制分析的难度。
文中首次引入信号MAC(Moving Averaging and Comparing)谱的概念,通过计算信号幅度谱每一抽头与临近若干抽头平均值的比值便能得到信号MAC谱,利用MAC谱检测谐波信号能够有效抑制色噪声带来的不利影响。在提出色噪声中谐波检测的MAC算法之后,文中结合不同的非线性变换进一步提出了分类BPSK/QPSK/8PSK信号的2~k次方律算法,以及分类错位调制与非错位调制的C-DOT(Classifier on Detecting Offset-Timing)算法。
文中基于数字外差、滤波和抽取提出一种用于谐波频率估计的FFT-Kay算法,该算法估计精度高、噪声门限低,各方面的性能已逼近理论上最优的最大似然估计(MLE;Maximum-Likelihood Estimator)。结合信号的MAC谱,FFT-Kay算法可推广至色噪声中的谐波频率估计,进而导出MAC-Kay算法。结合特定的非线性变换,以上两种算法可用于进行高性能的载波频率、符号速率估计。
调制分析的应用背景极为复杂,常常可能出现一些无法预料的情况,本文因此给出了两个涉及异常处理的算法,其一是周期序列检测,其二是信道均衡的失效检测。周期序列的存在和无效的信道均衡都有可能导致调制分类或识别算法给出错误的结果,在现有关于调制分析的论文中,几乎没有关于此类问题的讨论,因此这里的工作具有很强的借鉴意义。
最后给出的应用结果表明,SWDE理论完全能够用于指导和实现线性调制分析。换言之,本文给出的系列算法能够精确地估计出信号的载波频率和符号速率,并能有效地完成信号分类和识别。
An important technology called communication signal modulation analysis(CSMA)will be used if the third party in non-cooperative communication wants to surveil the activity of communication or even decipher its content for some special purpose.Although with little prior knowledge,CSMA can provide the estimation of some key parameters,such as carrier frequency and symbol rate,and determine the modulation types ultimately.As much as we know,CSMA can be found in many applications including Radio Reconnaissance,Non-Communication Intelligence Reconnaissance,Electronic Warfare,Radio Spectrum Management,and Software-Defined Radio.
CSMA is highly related to signal demodulation because of the same object and many common tasks.As a result,some algorithm found in signal demodulation can also be used for reference by CSMA.However,CSMA is much more challenging than demodulation because of a good many disadvantageous factor.
CSMA can be divided into the estimation of unknown paramters and the classification of modulation types.From the point of view of pattern recogonition,modulation classification is also composed of classifying feature extraction,samples feature set construction and classifying algorithm design,and parameter estimation will only be assistant.From the point of view of statistical signal processing,parameter estimation and modulation classification,which can be performed through signal detection,are equally interrelated.
This paper discusses the parameters estimation and modulation classification of linearly modulated signals on the theory of sinusoidal signal detection and estimation,a basic signal processing method can be found in all kinds of applications.The modulated signal set is composed of BPSK,QPSK,8PSK,OQPSK(MSK,π/2 BPSK),π/4 QPSK,16QAM,32QAM, 64QAM,128QAM,256QAM.
This paper gives out a new definition called local SNR to describe the quality of sinusoidal wave in colored noise.Local SNR is the ratio of the signal power to the variance of a nominal white noise,whose power spectrum density equals that of colored noise where the spectral line of sinusoidal wave is located.Based on a general model of linearly modulated signals,this paper also discusses how to calculate E_sN_0,which equals the ratio of energy per symbol to power spectrum density of noise.A series of computer simulations indicate that the background noise of CSMA is colored and non-Gaussian,which leads to the difficulty in CSMA.
A new definition called signal MAC(Moving Averaging and Comparing)spectrum, which can be got by calculating the ratio of every frequency bin and its nearby ones of magnitude spectrum and is immune from the magnitude spectrum fluctuation of colored noise, has been proposed for sinusoidal wave detecting and estimating in colored noise.Combining corresponding nonlinear transform and signal MAC spectrum,two signal classifying algorithm have been given.The one is 2~k-Power algorithm for classification of BPSK/QPSK/8PSK,the other is C-DOT(Classifier on Detecting Offset-Timing)algorithm for classification of offset modulated signal(MSK and OQPSK,for example)and non-offset modulated signal(QPSK and QAM,for example).
Based on digitally heterodyning,filtering and decimating,this paper puts forward the FFT-Kay algorithm,which owns good estimation precision and low noise threshold,for frequency estimation of sinusoidal wave in white noise.It is important that the FFT-Kay algorithm is comparable to the theoretic optimum MLE(Maximum-Likelihood Estimator) algorithm.Combined with signal MAC spectrum,the FFT-Kay algorithm can be easily extended to MAC-Kay algorithm,which can be used for frequency estimation of sinusoidal wave in colored noise.Following different nonlinear transform,above-mentioned two algorithms can be used to estimate the carrier frequency and symbol rate with high precision.
There are many accidents in CSMA because of its highly complicated application background.Two algorithms for countermeasure have been given to illustrate how these accidents can be dealt;the detection of periodic sequence and invalid channel equalization.Such algorithms are strongly recommended if wrong conclusion is unexpected in CSMA.
The application of above-mentioned algorithms indicates that the theory of sinusoidal wave detection and estimation can resolve many technique issues appearing in CSMA.
引文
[1]樊昌信,詹道庸,徐炳祥,等.通信原理[M].第4版.北京;国防工业出版社,1995;4-5.
[2]Nandi A K,Azzouz E E.Algorithms for automatic modulation recognition of communication signals[J].IEEE Transactions on Communications,1998,46(4);431-436.
[3]张贤达.现代信号处理[M].第1版.北京;清华大学出版社,1995;115-120.
[4]So H C,Chan Y T,Ma Q,et al.Comparision of various periodograms for sinusoid detection and frequency estimation[J].IEEE Transactions on Aerospace and Electronic Systems,1999,35(3);945-952.
[5]Kay S M,Gabriel J R.Optimal invariant detection of a sinusoid with unknown parameter[J].IEEE Transactions on Signal Processing,2002,50(1);27-40.
[6]P Baggenstoss M,Kay S M.An adaptive detector for deterministic signals in noise of unkown spectra using the rao test[J].IEEE Transactions on Signal Processing,1992,40(6);1460-1468.
[7]Bose S,Steinhardt A O.Adaptive array detection of uncertain rank one waveforms[J].IEEE Transactions on Signal Processing,1996,44(11);2801-2809.
[8]Kay S M,Sengupta D.Detection in incompletely characterized colored non-gaussian noise via parametric modeling[J].IEEE Transactions on Signal Processing,1993,41(10);3066-3070.
[9]Kay S M.Asymptotically optimal detection in incompletely characterized non-gaussian noise[J].IEEE Transactions on Acoustics,Speech and Signal Processing,1989,37(5);627-633.
[10]Wan C R,Goh J T,Chee H T.Optimal tonal detectors based on the power spectrum[J].IEEE Journal of Oceanic Engineering,2000,25(4);540-552.
[11]Rife D,Boorstyn R.Single-tone parameter estimation from discrete-time observations[J].IEEE Transactions on Information Theory,1974,20(5);591-598.
[12]Holm S.Optimum FFT-based frequency acquisition with application to COSPAS-SARSAT[J].IEEE Transactions on Aerospace and Electronic System,1993,29(2);464-475.
[13]Quinn B G.Estimating frequency by interpolation using Fourier coefficients[J].IEEE Transactions on Signal Processing,1994,42(5);1264-1268.
[14]Zakharov Y V,Baronkin V M,Tozer T C.DFT-based frequency estimators with narrow acquisition Range[J].IEE Proceedings on Communications,2001,148(1);1-7.
[15]Tretter S.Estimating the frequency of a noisy sinusoid by linear regression[J].IEEE Transactions on Information Theory,1985,31(6);832-835.
[16]Kay S M.A fast and accurate single frequency estimator[J].IEEE Transactions on Acoustics,Speech,and Signal Processing,1989,37(12);1987-1990.
[17]Lang S,Musicus B.Frequency estimation from phase differences[C].IEEE ICASSP,1989;2140-2143.
[18] Kim D, Cox D. An improved single frequency estimator [J]. IEEE Signal Processing Letters, 1996, 3(7): 212-214.
[19] Umesh S, Nelson D. Computationally efficient estimation of sinusoidal frequency at low SNR [C]. IEEE ICASSP, 1996: 2797-2800.
[20] Fowler M, Johnson J. Extending the threshold and frequency range for phase-based frequency estimation [J]. IEEE Transactions on Signal Processing, 1999, 47(10): 2857-2863.
[21] Jackson L, Tufts D. Frequency estimation by linear prediction [C]. IEEE ICASSP, 1978: 352-356.
[22] Brown T, Wang M. An iterative algorithm for single-frequency estimation [J]. IEEE Transactions on Signal Processing, 2002, 50(11): 2671-2682.
[23] Swami A, Mendel J M. Cumulant-based approach to harmonic retrieval and Related Problems [J]. IEEE Transactions on Signal Processing, 1991,39(5): 1099-1109.
[24] Anderson J M M, Giannakis G B, Swami A. Harmonic retrieval using higher order statistics: A deterministic formulation [J]. IEEE Transactions on Signal Processing, 43(8): 1880-1889, Aug. 1995.
[25] Gharieb R R, Horita Y, Murai T. Retrieving sinusoids in colored Rayleigh noise by a Cumulant-Based FBLP Approach [C]. IEEE ICASSP, 2000: 741-744.
[26] Gharieb R R. New results on employing cumulants for retrieving sinusoids in colored non-gaussian noise [J]. IEEE Transactions on Signal Processing, 2000, 48(7):2164-2168.
[27] Gardner W A. Exploitation of spectral redundancy in cyclostationary signals [J]. IEEE Signal Processing Magazine, 1991, 8(2): 14-19.
[28] Gardner W A. Signal interception: a unifying theoretical framework for feature detection [J]. IEEE Transactions on Communications, 1988, 36(8): 897-906.
[29] Dandawate A X, Giannakis G B. Statistical tests for presence of cyclostationarity [J]. IEEE Transactions on Signal Processing, 1994,42(9): 2355-2369.
[30] Ciblat P, Loubaton P, Serpedin E, et al. Asymptotic analysis of blind cyclic correlation- based symbol-rate estimators [J]. IEEE Transactions on Information Theory, 2002, 48(7): 1922-1934.
[31] Mazet L, Loubaton R. Cyclic correlation based symbol rate estimation [C]. Record of the Thirty-Third Asilomar Conference on Signals, Systems, and Computers, 1999: 1008-1012.
[32] Gini F, Giannakis G B. Frequency offset and symbol timing recovery in flat-fading channels: a cyclostationary approach [J]. IEEE Transactions on Communications, 1998, 46(3): 400-411.
[33] Gaby J E. Automatic bit pattern analysis of communication signals [C]. ICASSP'90, 1990: 1715-1718.
[34] Wegener A W. Practical techniques for baud rate estimation [C]. ICASSP'92, 1992: 681-684.
[35] Koh B S, Lee H S. Detection of symbol rate of unknown digital communication Signals [J]. Electronic Letters, 1993, 29(3): 278-279.
[36] Chan Y T, Plews J W, Ho K C. Symbol rate estimation by the wavelet transform [C]. IEEE ISCAS, Hong Kong, 1997: 177-180.
[37] Sawai R, Harada H, et al. General-purpose symbol rate and symbol timing estimation method by using multi-resolution analysis based on wavelet transform for multimode software radio [C]. IEEE VTC, 1999: 2128-2132.
[38] Shintaro Taira, Eiichiro Murakami. Automatic classification of analogue modulation signals by statistical parameters [C], Military Communications Conference Proceedings, 1999: 202-207.
[39] Lopez-Salcedo J A, Vazquez G. Stochastic approach to square timing estimation with frequency uncertainty [C]. IEEE ICC, 2003: 3555-3559.
[40] Wickert W A, Turcotte R L. Rate-line detection using higher-order spectra [C]. IEEE MILCOM, San Diego, 1992: 1221-1225.
[41] Truong T, Wickert M A. Performance of spectral line detection using integrated trispectrum [C], IEEE MILCOM, Boston, 1998: 659-663.
[42] Reed D E, Wickert M A. Symbol-rate detection by a power-series-nonlinear envelope detector receiver [C]. Proceedings of the Seventh Annual International Phoenix Conference on Computers and Communications, Scottsdale, 1988:179-183.
[43] Kueckenwaitz M, Quint F, Reichert J. A robust baud rate estimator for noncooperative demodulation [C]. IEEE MILCOM, Los Angeles, 2000: 971-975.
[44] Chen L, Kusaka H, Kominami M. Blind phase recovery in QAM communication systems using higher order statistics [J]. IEEE Signal Processing Letters, 1996, 3(5): 147-149.
[45] Cartwright K V. Blind phase recovery in general QAM communication systems using alternative higher order statistics [J]. IEEE Signal Processing Letters, 1999, 6(12): 327-329.
[46] Serpedin E, Ciblat P, Giannakis G B, et al. Performance analysis of blind carrier phase estimators for general QAM constellations [J]. IEEE Transactions on Signal Processing, 2001,49(8): 1816-1823.
[47] Moeneclaey M, Jonghe G de. ML-oriented NDA carrier synchronization for general rotationally symmetric signal constellations [J]. IEEE Transactions on Communications, 1994, 42(8): 2531-2533.
[48] Cartwright K V. Blind phase recovery in cross QAM communication systems with eighth-order statistics [J]. IEEE Signal Processing Letters, 2001, 8(12): 304-306.
[49] Georghiades C N. Blind carrier phase acquisition for QAM constellations [J]. IEEE Transactions on Communications, 1997,45(11): 1477-1486.
[50] Morelli M, Mengali U, D'Andrea A. N. Feedforward estimation techniques for carrier recovery in 16-QAM modulation [J]. Broadband Wireless Communications, Luise M, Pupolin S. New York: Springer-Verlag, 1988.
[51] Patrizio Campisi, Gianpiero Panci, Stefania Colonnese et al. Blind phase recovery for QAM communication systems [J]. IEEE Transactions on Signal Processing, 2005, 53(4): 1348-1358.
[52]D'Andrea A,Mengali U,Reggiannini R.The modified Cramer-Rao bound and its application to synchronization problems[J].IEEE Transactions on Communications,1994,42(2/3/4);1391-1399.
[53]Gini F,Reggiannini R,Mengali U.The modified Cramer-Rao bound in vector parameter estimation[J].IEEE Transactions on Communications,1998,46(1);52-60.
[54]Mengali U,D'Andrea A.Sychronization Techniques for Digital Receivers.New York;Plenum,1997.
[55]Moeneclaey M.On the true and the modified Cramer-Rao bounds for the estimation of a scalar parameter in the presence of nuisance parameters[J].IEEE Transactions on Communications,1998,46(11);1536-1544.
[56]Cowley W G.Phase and frequency estimation for PSK packets;Bounds and algorithms[J].IEEE Transactions on Communications,1996,44(1);26-28.
[57]Rice F,Cowley B,Moran B,et al.Cramer-Rao Lower Bounds for QAM Phase and Frequency Estimation[J].IEEE Transactions on Communications,2001,49(9);1582-1591.
[58]Yuan W S,Georghiades C N.Rapid carrier aquisition from baud-rate samples[J].IEEE Transactions on Communications,1999,47(4);631-641.
[59]Luise M,Pupolin S.Broadband Wireless Communications[M].London,U.K.;Springer-Verlag,1998.
[60]Rice F,Rice M,Cowley B.A new bound and algorithm for star 16-QAM carrier phase estimation[J].IEEE Transactions on Communications,2003,51(2);161-165.
[61]陈卫东.数字通信信号调制识别算法研究;西安电子科技大学博士论文.西安;西安电子科技大学,2001.
[62]詹亚锋.通信信号自动调制识别和参数估计;清华大学博士论文.北京;清华大学,2004.
[63]Huang C Y,Polydoros A.Likelihood methods for mPSK modulation classification[J].IEEE Transactions on Communications,1995,43(2/3/4);1493-1504.
[64]Kim K,Polydoros A.Digital modulation classification;the BPSK versus QPSK case[C].MILCOM'88,San Diego,CA,1988;431-436.
[65]Han Y Y,Lee W C,Yoon D W.On the error rate evaluation and modulation classification for coherent and noncoherent PSK signals using the transformation of random variable[C].ICC'93,Geneva,Switzerland,1993;1508-1514.
[66]Yang Yawpo,Soliman S S.Optimum classifier for M-ary PSK signals[C].ICC'91,Denver,CO,1991;1693-1697.
[67]Yang Yawpo,Soliman S S.Statistical moments based classifier for mPSK signals[C].GLOBECOM'91,Phoenix,AZ,1991;72-76.
[68]Yang Yawpo,Soliman S S.An improved moment-based algorithm for signal classification[J].Signal Processing,1995,43(3);231-244.
[69]Yang Yawpo,Soliman S S.Suboptimal algorithm for modulation classification[J].IEEE Transactions on Aerospace and Electronic Systems,1997,33(1);38-45.
[70] Soliman S S, Hsue Shue-Zen. Signal classification using statistical moments [J]. IEEE Transactions on Communications, 1992,40(5): 908-916.
[71] Schreyoegg C, Reichert J. Modulation classification of QAM schemes using the DFT of phase histogram combined with modulus imformation [C]. MILCOM'97, Monterey, CA, 1997: 1372-1376.
[72] Sills J A. Maximum-likelihood modulation classification for PSK/QAM [C]. MILCOM'99, Atlantic, City, NJ, 1999: 217-220.
[73] Wei Wen, Mendel J M. Maximum-likelihood classification for digital amplitude-phase modulations [J]. IEEE Transactions on Communications, 2000,48(2): 189-193.
[74] Yawpo Yang, Liu Ching-Hwa, Soong Ta-Wei. A log-likelihood function-based algorithm for QAM signal classification [J]. Signal Processing, 1998, 70(1): 61-71.
[75] Long C S, Chugg K M, Polydoros A. Further results in likelihood classification of QAM signals [C]. MILCOM'94, Long Branch, NJ. 1994: 57-61.
[76] Beitas B F, Weber C L. Higher-order correlation-based approach to modulation classification of digitallly frequency-modulated signals [J]. IEEE Journal on Selected Areas in Communications, 1995, 13(1): 89-101.
[77] Beitas B F, Weber C L. Asynchronous classification of mFSK signals using the higher order correlation domain [J]. IEEE Transactions on Communications, 1998,46(4): 480-493.
[78] Boiteau D, Le Martret C. A general maximum likelihood framework for modulation classification [C]. ICASSP'98, Seattle, WA, 1998: 2165-2168.
[79] Vergara-Dominguez L, Paez-Borrallo J M, Portillo-Garcia J. A general approach to the automatic classification of radio communication signals [J]. Signal Processing, 1991, 22(3): 239-250.
[80] Lamontagne R. Modulation recognition: An overview. NTIS No: N92-20159/9/HDM, Defense Research Establishment, Ottawa, Ontario, March 1991.
[81] Liedtke F F. Computer simulation of an automatic classification procedure for digitally modulated communication signals with unknown parameters [J]. Signal Processing, 1984, 6(4): 311-323.
[82] Portillo-Garcia J I, Sancho-Marco J P, Vergara-Dominguez L. A microcomputer-based general architecture for radiocommunication signal classification and digital demodulation [C]. EUSIPCO'90, Barcelona, Spain, 1990: 1919-1922.
[83] Hong Liang, Ho K C. Identification of digtial modulation types using the wavelet transform [C]. MILCOM'99. Atlantic City, NJ, 1999: 427-431.
[84] Hsue S Z, Soliman, S S. Automatic modulation classification using zeros crossing [J]. IEE Proceedings of Radar and Signal Processing, 1990: 459-464.
[85] Azzouz E E, Nandi A K. Automatic modulation recognition of communication signals [M]. Netherlands: Kluwer Academic Publishes, 1996.
[86] Eimicke G A. A flexible architecture for real-time modulation recognition [C]. ASSPA'89, Australia, 1989: 133-137.
[87] Hipp J E. Modulation classification based on statistical moment [C]. MILCOM'86, Monterey, CA, 1986: 2021-2026.
[88] Whelchel J E, McNeill D L, Hughes R D. Signal understanding: an artificial intelligence approach to modulation classification [J]. IEEE International Workshop on Tools for Artificial Intelligence, Architectures, Languages and Algorithm, 1998:231-236.
[89] Louis C, Sehier P. Automatic modulation recognition with a hierarchical neural network [C]. MILCOM'94, Long Branch, NJ, 1994: 713-717.
[90] Hill P C J, Orzeszko G R. Performance comparison of neural network and statistical discriminant processing techniques for automatic modulation recognition [J]. SPIE, 1991, 1469(1): 329-340.
[91] Swami A, Sadler B M. Modulation classification via hierarchical agglomerative cluster analysis [J]. IEEE Signal Processing Workshop on Signal Processing Advances in Wireless Communications, SPAWC'97, 1997: 141-144.
[92] Swami A, Sadler B M. Hierarchical digital modulation classification using cumulants [J]. IEEE Transactions on Communications, 2000, 48(3): 416-429.
[93] Le Martret C J, Boiteau D M. Modulation classification by means of different orders statistical moments [C]. MILCOM'97, Monterey, CA, 1997: 1387-1391.
[94] HeroIII A O, Hadinejad-Mahram H. Digital modulation classification using power moment matrices [C]. ICASSP'98, Seattle, WA, 1998: 3285-3288.
[95] Akmouche W. Detection of multicarrier modulations using 4th-order cumulants [C]. MILCOM'99, Atlantci City, NJ, 1999: 432-436.
[96] Weaver C S, Cole C A, Krumland R B. The automatic classification of modulation types by pattern recognition [C]. AD691069, 1969, April.
[97] Ghani N. Lamontagne R., Neural Networks Applied to the Classification of Spectral Features for Automatic Modulation Recognition [C], MILCOM'93, Boston, MA, Vol.1, pp.111-115, 1993.
[98] Zhu Q. Kam M. Yeager R., Non-parametric Identification of QAM Constellations in Noise [C], ICASSP'93, Minneapolis, MN, Vol.4, ppl84-187, 1993.
[99] Sapiano P C, Martin J D. Maximum likihood PSK classificaiton using the DFT of phase histogram [C]. GLOBECOM'95, Singapore, 1995: 1029-1033.
[100] Sapiano P C, Martin J D, Holbeche R J. Classification of PSK signals using the DFT of phase histogram [C]. ICASSP'95, Detroit, Ml, 1995: 1868-1871.
[101] Wood S.L., Ready M.J., and Treichler J.R., Constellation Identification Using the Radon Transform [C], ICASSP'88, New York City, Vol.3, pp.1878-1881, 1988.
[102] Wood S L, Ready M J, Treichler J R. Performance of the radon transform method for constellation identification [C]. Proceeding of the Twenty-Second Asilomar Conference on Circuits, System and Computers, 1988: 119-122.
[103] Wood S L, Larimore M G, Treichler J R. Modem constelltion identification: a performance comparison of two methods [C]. ICASSP'90, Albuquerque, MI, 1990: 1651-1654.
[104] Wood S L, Treichler J R. Computational and performance analysis of radon transform based constellation identificaiton [C]. ICASSP'94, Adelaide, Australia, 1994: 241-244.
[105]Gardner W A,Spooner C M.Cyclic spectral analysis for signal detection and modulation recognition[C].MILCOM'88,San Diego,CA,1988;419-424.
[106]Kim Y K,Weber C L.Generalized single cycle classifier with applications to SQPSK vs.2~kPSK[C].MILCOM'89,Boston,MA,1989;754-758.
[107]Reichert J.Automatic classification of communication signals using higher order statistics[C].ICASSP'92,San Francisco,1992;221-224.
[108]Marchand P,Le Martret C,Lacounme J L.Classification of linear modulation by a combination of different orders cyclic cumulants[C].Proceedings of IEEE Signal Processing Workshop on Higher-Order Statistics,1997;47-51.
[109]Gardner W A.The role of spectral correlation in design and performance analysis of synchronizers[J].IEEE Signal Processing Magazine,1986,34(11);1089-1095.
[110]Polydoros A,Kim K.On the detection and classification of quadrature digital modulations in broad-band noise[J].IEEE Transactions on Communications,1990,38(8);1199-1211.
[111]Bernardini A,Fina S D.Optimal decision boundaries for m-QAM signal formats using neural classifiers[J].IEEE Transactions on Neural Networks,1998,9(2);241-246.
[112]Daponte P,Mercurio G,Rapuano S.A wavelet networks-based method for the digital telecommunication system monitoring[J].IEEE Transactions on Instrumentation and Measurement,2001,50(6);1773-1780.
[113]El-Mahdy A E,Namazi N M.Classification of multiple m-ary frequency-shift keying signals over a rayleigh fading channel[J].IEEE Transactions on Communications,2002,50(6);967-974.
[114]Wei W,Mendel J M.A fuzzy logic method for modulation classification in nonideal environments[J].IEEE Transactions on Fuzzy System,1999,7(3);333-344.
[115]Kim N J,Kehtarnavaz N.Dsp-based hierarchical neural network modulation signal classification[J].IEEE Transactions on Neural Networks,2003,14(5);1065-1071.
[116]胡广书.数字信号处理-理论、算法与实现[M].第2版,北京;清华大学出版社,2003;410-413.