基于谱相关的通信和雷达信号识别方法研究
|
摘要
随着现代科学技术的不断发展,现代战争由过去单一式对抗向着系统和体系对抗转变。这一转变使得各国都在努力进行雷达/通信电子战的一体化研究。国内在通信侦察和雷达侦察的单系统方面的研究均已取得了很大的成果,而对于两者一体化的研究则是刚刚起步。本课题就是在此基础上,对雷达/通信侦察一体化技术中的信号调制方式识别方法进行理论探讨。
信号的调制方式识别在单平台雷达/通信侦察一体化技术的研究中具有极其重要的意义。只有正确识别信号的调制方式,才能在后续的处理中更有针对性的对截获的信号进行所需参数的提取。由于无线通信环境的复杂性和不可预知性,尤其在现代电子战环境下,各方都广泛采用了低截获概率(LPI)技术以达到反截获、抗干扰同时能够提高自身作战能力的目的。由于传统的调制方式识别方法大都对噪声很敏感,因此难以满足低信噪比环境下的信号识别及其参数估计。而基于循环谱理论的信号处理方法因其对平稳性干扰和噪声的不敏感性,已被广泛地应用于信号分析及参数提取等方面的理论研究中。因此,循环谱理论很适合用于雷达/通信一体化电子战中关于信号调制方式识别以及所需参数的提取。这种低信噪比情况下调制方式识别的研究不仅具有重要的理论意义,更有重大的应用价值。
本文首先介绍了基于循环平稳模型的循环谱理论；其次,对雷达/通信一体化系统接收到的位于同频段的雷达和通信信号中的单一信号,通过统计占空比的方法将所接收信号粗略的分为两类,即雷达类和常规通信信号类,以便后续选择不同的分析方法分别对雷达信号、通信信号进行调制方式的识别(本课题对两类信号都采用谱相关法进行处理)；接着,利用二阶循环谱的优良特性,采用有限时间平均循环自相关法(CA算法)分析了常规通信信号(AM、FM(mf=0.3和mf=20,其中mf为信号的调频指数)、ASK、FSK、BPSK、QPSK)和脉冲压缩雷达脉内调制信号(LFM信号和二相编码信号)的谱相关特性,以此为参考提取相应的特征参数,进而达到对信号调制方式的识别。
仿真结果表明,利用包络检波,统计信号的占空比因子,确定合适的门限可以很好的将脉冲压缩雷达信号和常规通信信号这两类信号分离；而采用有限时间平均循环自相关法(CA算法),确定合适的门限值,能够在低信噪比的情况下将所要分析的常规通信信号和脉冲压缩雷达信号的不同调制方式识别出来,并对脉压雷达两种调制信号的特征参数进行了估计。最后验证了算法的正确性和可行性。
With the continuous development of modern science and technology, modern warfare is transforming from the previous single-style confrontation to the system confrontation. This change makes countries all over the world strive to carry out the research on the integrative warfare technology for radar and communication. The domestic studies of single-system communication surveillance and radar reconnaissance have got great achievements, but the studies of the integration of the two are just getting started. As it is the background, this paper studies the automatic recognition of modulation types in the integrative reconnaissance technology for radar and communication signals.
The automatic recognition of modulation types is a very important link in the study of the integrative reconnaissance technology for radar and communication signals. Only by correctly identifying the signal modulation, the extraction of parameters for the interception of signals is more targeted in the following process. As the complexity and unpredictability of wireless communications environment are increasing, especially in the modern electronic warfare environment, every country is widely using the low probability of intercept (LPI) techniques to achieve high ability to resist interception and interference. As the traditional recognition of modulation types is very sensitive to noise, it's hard to recognize modulation types and extract the parameters of signals in the case of low SNR. The signal processing methods basing on the theory of cyclic spectrum are widely used in the theoretical studies of signal analysis and parameter extraction for the benefits of insensitivity to the stationary interference and noise. For these reasons, the theory of cyclic spectrum is very suitable for the recognition of modulation types and extraction parameters of signals in the integrative radar and communication warfare. This study of modulation recognition in the case of low SNR is not only of important theoretical significance, but has very important application value.
This article first introduces cyclic spectrum theory based on cyclostationary model, and then through calculating duty cycle of the received signal which is one of the given routine communication signals and pulse compression radar signals will be roughly seperated as one categorie in two (communication signals and the radar signals). After that, different methods are used to recognize the modulation types of communication signals and pulse compression radar signals. With the superior characteristics of second-order cyclic spectrum this paper uses time-variant finite-average cyclic autocorrelation algorithm (CA algorithm) to analize the spectral correlation characteristics of both communication signals (AM, FM (mf=0.3 and mf=20), ASK, FSK, BPSK and QPSK signal) and pulse compression radar signals(linear frequency modulation and two-phase Barker signal). At last by extracting the characteristic parameters the automatic recognition of modulation type is achieved.
Simulation results show that using envelope detection to get duty cycle is effective to separate routine communication signals and the pulse compression radar signals with a appropriate duty cycle threshold. And in the case of low SNR, it is effective to identify the different modulation types of given routine communication signals and the pulse compression radar signals using CA algorithm. The appropriate thresholds are also needed. Finally the correctness and feasibility of the algorithm is verified by simulation.
信号的调制方式识别在单平台雷达/通信侦察一体化技术的研究中具有极其重要的意义。只有正确识别信号的调制方式,才能在后续的处理中更有针对性的对截获的信号进行所需参数的提取。由于无线通信环境的复杂性和不可预知性,尤其在现代电子战环境下,各方都广泛采用了低截获概率(LPI)技术以达到反截获、抗干扰同时能够提高自身作战能力的目的。由于传统的调制方式识别方法大都对噪声很敏感,因此难以满足低信噪比环境下的信号识别及其参数估计。而基于循环谱理论的信号处理方法因其对平稳性干扰和噪声的不敏感性,已被广泛地应用于信号分析及参数提取等方面的理论研究中。因此,循环谱理论很适合用于雷达/通信一体化电子战中关于信号调制方式识别以及所需参数的提取。这种低信噪比情况下调制方式识别的研究不仅具有重要的理论意义,更有重大的应用价值。
本文首先介绍了基于循环平稳模型的循环谱理论；其次,对雷达/通信一体化系统接收到的位于同频段的雷达和通信信号中的单一信号,通过统计占空比的方法将所接收信号粗略的分为两类,即雷达类和常规通信信号类,以便后续选择不同的分析方法分别对雷达信号、通信信号进行调制方式的识别(本课题对两类信号都采用谱相关法进行处理)；接着,利用二阶循环谱的优良特性,采用有限时间平均循环自相关法(CA算法)分析了常规通信信号(AM、FM(mf=0.3和mf=20,其中mf为信号的调频指数)、ASK、FSK、BPSK、QPSK)和脉冲压缩雷达脉内调制信号(LFM信号和二相编码信号)的谱相关特性,以此为参考提取相应的特征参数,进而达到对信号调制方式的识别。
仿真结果表明,利用包络检波,统计信号的占空比因子,确定合适的门限可以很好的将脉冲压缩雷达信号和常规通信信号这两类信号分离；而采用有限时间平均循环自相关法(CA算法),确定合适的门限值,能够在低信噪比的情况下将所要分析的常规通信信号和脉冲压缩雷达信号的不同调制方式识别出来,并对脉压雷达两种调制信号的特征参数进行了估计。最后验证了算法的正确性和可行性。
With the continuous development of modern science and technology, modern warfare is transforming from the previous single-style confrontation to the system confrontation. This change makes countries all over the world strive to carry out the research on the integrative warfare technology for radar and communication. The domestic studies of single-system communication surveillance and radar reconnaissance have got great achievements, but the studies of the integration of the two are just getting started. As it is the background, this paper studies the automatic recognition of modulation types in the integrative reconnaissance technology for radar and communication signals.
The automatic recognition of modulation types is a very important link in the study of the integrative reconnaissance technology for radar and communication signals. Only by correctly identifying the signal modulation, the extraction of parameters for the interception of signals is more targeted in the following process. As the complexity and unpredictability of wireless communications environment are increasing, especially in the modern electronic warfare environment, every country is widely using the low probability of intercept (LPI) techniques to achieve high ability to resist interception and interference. As the traditional recognition of modulation types is very sensitive to noise, it's hard to recognize modulation types and extract the parameters of signals in the case of low SNR. The signal processing methods basing on the theory of cyclic spectrum are widely used in the theoretical studies of signal analysis and parameter extraction for the benefits of insensitivity to the stationary interference and noise. For these reasons, the theory of cyclic spectrum is very suitable for the recognition of modulation types and extraction parameters of signals in the integrative radar and communication warfare. This study of modulation recognition in the case of low SNR is not only of important theoretical significance, but has very important application value.
This article first introduces cyclic spectrum theory based on cyclostationary model, and then through calculating duty cycle of the received signal which is one of the given routine communication signals and pulse compression radar signals will be roughly seperated as one categorie in two (communication signals and the radar signals). After that, different methods are used to recognize the modulation types of communication signals and pulse compression radar signals. With the superior characteristics of second-order cyclic spectrum this paper uses time-variant finite-average cyclic autocorrelation algorithm (CA algorithm) to analize the spectral correlation characteristics of both communication signals (AM, FM (mf=0.3 and mf=20), ASK, FSK, BPSK and QPSK signal) and pulse compression radar signals(linear frequency modulation and two-phase Barker signal). At last by extracting the characteristic parameters the automatic recognition of modulation type is achieved.
Simulation results show that using envelope detection to get duty cycle is effective to separate routine communication signals and the pulse compression radar signals with a appropriate duty cycle threshold. And in the case of low SNR, it is effective to identify the different modulation types of given routine communication signals and the pulse compression radar signals using CA algorithm. The appropriate thresholds are also needed. Finally the correctness and feasibility of the algorithm is verified by simulation.
引文
[1]李朝伟,周希元.雷达/通信信号侦察一体化技术[J].舰船电子对抗.2008,31(2)：5-11页.
[2]李廷军,任建存,赵元立,张金华.雷达—通信一体化研究[J].现代雷达.2001,23(2)：1-2页.
[3]熊少华,吴志文.通信、雷达与电子战部队三位一体化势在必行[J].军队指挥自动化.1999,(1)：46-48页.
[4]殷兆伟,戴国宪.不断发展的电子战接收机[J].航天电子对抗.2002,(6)：42-45页.
[5]余士平.通信信号侦收的算法验证及软件实现[D].杭州：浙江大学.2005,1-2,21-34页.
[6]潘继飞,姜秋喜,毕大平.基于数字接收机脉压雷达信号截获技术研究[J].现代防御技术.2005,33(5)：56-59页.
[7]赵国庆.雷达对抗原理[M].西安电子科技大学出版社.2001,1-33页.
[8]潘继飞.基于数字接收机的脉压雷达信号的截获、识别技术研究[D].现代防御技术.2005,33(5)：55-59页.
[9]张贤达,保铮编著.非平稳信号分析与处理[M].国防工业出版社.1999.324-336页.
[10]黄知涛,周一宇,姜文利.循环平稳信号处理与应用[M].科学出版社.2006.1-35页.
[11]Gardner W A. The spectral correlation theory of cyclostationary time-serise [J]. Signal processing.1986,11(7):13-36P.
[12]Gardner W A. Spectral correlation of modulated signals:PART Ⅰ—Analog Modulation [J]. IEEE Trans. on Comm.,1987, COM-35(6):584-594P.
[13]Gardner W A. Spectral correlation of modulated signals:PART Ⅱ—Digital Modulation [J]. IEEE Trans. on Comm.,1987, COMM-35(6):595-601P.
[14]林志远,刘刚.雷达一电子战一通信的一体化[J].上海航天.2004,(6)：55-58页.
[15]包锡锐,吴瑛,周欣.基于高阶累积量的数字调制信号识别算法[J].郑州：信息工程大学学报.2007,8(4)：464-467页.
[16]李耐根.基于谱分析数字信号调制方式的自动识别[D].南昌：南昌大学.2006,37-60页.
[17]陈洄.基于循环谱相关的调制方式识别的研究[D].北京：北京交通大学.2006,1-6页.,15-38页.
[18]高玉龙.基于循环谱的调制方式识别与高动态同步技术研究[D].哈尔滨：哈尔滨工业大学.2007,3-27页.
[19]张晓林,郭黎利.基于循环谱相关技术的直扩信号的检测与参数估计[D].哈尔滨：哈尔滨工程大学.2003,1-2页,16-27页.
[20]胡延平,李广森,李纲,皇甫堪。利用参数统计方法自动识别数字调制信号[J].通信学报.2002,23(2)：58-65页.
[21]朱晓霞,陆明泉.通信信号细微特征分析[J].电子对抗.1995,(3)：2-11页.
[22]巫胜洪.雷达脉内特征提取方法的研究[J].舰船电子对抗.2002,25(1)：25-28页.
[23]谭营.脉压信号识别的时域方法[J].电子对抗.1991,(2)：27-32页.
[24]王勇,姜秋喜,毕大平.雷达信号指纹分析技术[J].中国雷达.2004,(3)：33-35页.
[25]孟晓琳.电子侦察系统中的雷达信号脉内细微特征分析[D].哈尔滨：哈尔滨工程大学.2007,1-12页.
[26]沙祥.脉内分析综述.现代电子[J].1998,(4)：1-8页.
[27]穆世强.雷达信号脉内细微特征分析[J].电子对抗技术.1991,(2)：28-37页.
[28]罗利春.谱相关的原理、功能与α截面谱表示[J].物理学报.2002,51(10)：2177-2181页.
[29]黄春琳,姜文利,周一宇.低截获概率雷达信号的谱线关函数检测方法 分析[J].长沙：国防科技大学学报.2001,23(4)：102-106页.
[30]Gardner W A. Measurement of spectral correlation [J]. IEEE Trans ASSP.1986,34(5):1111-1123P.
[31]W. A. Gardner. Stationarizable Random Processes [J]. IEEE Trans on Inform Theory,1978,24:8-22P.
[32]W. A. Gardner, C. Spooner. Signal Interception:Performance Advantages of Cyclic feature Detectors [J]. IEEE Trans on Communications,1992,40: 149-159P.
[33]C. Spooner, W. A. Gardner. Robust Feature Detection for Signal Interception [J]. IEEE Trans on Communications,1994,42:2165-2173P.
[34]Y. Li, Z. Ding. Blind Channel Identification Based on Second Order Cyclostationary Statistics [J]. IEEE International Conference on Acoustics, Speech, and Signal Processing,1993,4:81-84P.
[35]张晓林,郭黎利,赵毅.直扩信号识别的有限时间平均循环自相关法[J].哈尔滨：哈尔滨工程大学学报.2006,29(6)：900-903页.
[36]王瑛,程汉文,吴乐南.基于谱相关特征的信号调制方式识别[J].信息技术.2006,28(12)：25-28页.
[37]楼才义.一种多信道、多功能数字化接收机.中国电子学会电子对抗分会第十二届学术年会论文集.485-488页.
[38]石远东.宽带数字接收机信道化及其相关技术的研究与实现[D].南京：南京航空航天大学.2007,14-23页.
[39]张明友,汪学刚.雷达系统[M].电子工业出版社.2006,247-271页.
[40]丁鹭飞,耿富录.雷达原理[M].西安电子科技大学出版社.2005,第三版.1-51页.
[41]姚亚峰,黄载禄.通信信号调制识别技术[J].通信技术.2003,6：41-43页.
[42]W. Brown, H. Loomis. Digital Implementations of Spectral Correlation Analyzers [J]. IEEE Trans on Signal Processing,1993,41:703-720P.
[43]韩国栋,邬江兴.调制参数提取的谱相关方法[J].航天电子对抗.2001,3：37-41页.
[44]吕杰,张胜付,邵伟华,谷明.数字通信信号自动调制识别的谱相关方法[J].南京：南京理工大学学报.1999,23(4)：297-299页.
[45]郭黎利.软件无线电中基于谱相关理论的调制模式识别技术[J].哈尔滨：哈尔滨工程大学学报.2004,(6)：25-29页.
[46]张贤达.时间序列分析—高阶统计量方法[M].清华大学出版社.1996,1-24页.
[47]樊昌信.通信原理[M].国防工业出版社.2001,第五版.13-170页.
[48]郭黎利,张晓林.通信原理[M].哈尔滨工程大学出版社.2003,31-136页.
[49]王生兵.无线电信号的调制识别研究[D].南京：东南大学.2006,1-14页.
[50]史建峰.基于周期谱包络的线性调频信号特征参数估计[J].电子对抗技术.2004,19(3)：7-10页.
[51]史建峰.有限数据下循环谱的频域平滑对称式周期图法估计性能分析[J].电子对抗技术.2004,19(2)：155-159页.
[52]Dandawate A V, Giannakis G B. Nonpararnetric Polyspectral Estimators for kth-Order (Almost) Cyclostationary Processes [J]. IEEE Tram on Information Theory.1994,40(1):67～84P.
[53]刘树棠译.现代通信系统(MATLAB版)[M].电子工业出版社.2005,第二版.1-262页.
[54]朱德君.谱相关理论在电子侦察中的应用[J].电子对抗.1995(2)：43-58页.
[55]Yeung G K, GardnerW A. Search-efficient methods of detection of cyclostationary signals [J]. IEEE Trans on Signal Processing,1996,44(5): 1214-1223P.
[56]Roberts R S, Brown W A, Loom is H H. Computationally efficient algorithms for cyclic spectral analysis [J]. IEEE SP MA G,1991,8(2):38～ 49P.
[2]李廷军,任建存,赵元立,张金华.雷达—通信一体化研究[J].现代雷达.2001,23(2)：1-2页.
[3]熊少华,吴志文.通信、雷达与电子战部队三位一体化势在必行[J].军队指挥自动化.1999,(1)：46-48页.
[4]殷兆伟,戴国宪.不断发展的电子战接收机[J].航天电子对抗.2002,(6)：42-45页.
[5]余士平.通信信号侦收的算法验证及软件实现[D].杭州：浙江大学.2005,1-2,21-34页.
[6]潘继飞,姜秋喜,毕大平.基于数字接收机脉压雷达信号截获技术研究[J].现代防御技术.2005,33(5)：56-59页.
[7]赵国庆.雷达对抗原理[M].西安电子科技大学出版社.2001,1-33页.
[8]潘继飞.基于数字接收机的脉压雷达信号的截获、识别技术研究[D].现代防御技术.2005,33(5)：55-59页.
[9]张贤达,保铮编著.非平稳信号分析与处理[M].国防工业出版社.1999.324-336页.
[10]黄知涛,周一宇,姜文利.循环平稳信号处理与应用[M].科学出版社.2006.1-35页.
[11]Gardner W A. The spectral correlation theory of cyclostationary time-serise [J]. Signal processing.1986,11(7):13-36P.
[12]Gardner W A. Spectral correlation of modulated signals:PART Ⅰ—Analog Modulation [J]. IEEE Trans. on Comm.,1987, COM-35(6):584-594P.
[13]Gardner W A. Spectral correlation of modulated signals:PART Ⅱ—Digital Modulation [J]. IEEE Trans. on Comm.,1987, COMM-35(6):595-601P.
[14]林志远,刘刚.雷达一电子战一通信的一体化[J].上海航天.2004,(6)：55-58页.
[15]包锡锐,吴瑛,周欣.基于高阶累积量的数字调制信号识别算法[J].郑州：信息工程大学学报.2007,8(4)：464-467页.
[16]李耐根.基于谱分析数字信号调制方式的自动识别[D].南昌：南昌大学.2006,37-60页.
[17]陈洄.基于循环谱相关的调制方式识别的研究[D].北京：北京交通大学.2006,1-6页.,15-38页.
[18]高玉龙.基于循环谱的调制方式识别与高动态同步技术研究[D].哈尔滨：哈尔滨工业大学.2007,3-27页.
[19]张晓林,郭黎利.基于循环谱相关技术的直扩信号的检测与参数估计[D].哈尔滨：哈尔滨工程大学.2003,1-2页,16-27页.
[20]胡延平,李广森,李纲,皇甫堪。利用参数统计方法自动识别数字调制信号[J].通信学报.2002,23(2)：58-65页.
[21]朱晓霞,陆明泉.通信信号细微特征分析[J].电子对抗.1995,(3)：2-11页.
[22]巫胜洪.雷达脉内特征提取方法的研究[J].舰船电子对抗.2002,25(1)：25-28页.
[23]谭营.脉压信号识别的时域方法[J].电子对抗.1991,(2)：27-32页.
[24]王勇,姜秋喜,毕大平.雷达信号指纹分析技术[J].中国雷达.2004,(3)：33-35页.
[25]孟晓琳.电子侦察系统中的雷达信号脉内细微特征分析[D].哈尔滨：哈尔滨工程大学.2007,1-12页.
[26]沙祥.脉内分析综述.现代电子[J].1998,(4)：1-8页.
[27]穆世强.雷达信号脉内细微特征分析[J].电子对抗技术.1991,(2)：28-37页.
[28]罗利春.谱相关的原理、功能与α截面谱表示[J].物理学报.2002,51(10)：2177-2181页.
[29]黄春琳,姜文利,周一宇.低截获概率雷达信号的谱线关函数检测方法 分析[J].长沙：国防科技大学学报.2001,23(4)：102-106页.
[30]Gardner W A. Measurement of spectral correlation [J]. IEEE Trans ASSP.1986,34(5):1111-1123P.
[31]W. A. Gardner. Stationarizable Random Processes [J]. IEEE Trans on Inform Theory,1978,24:8-22P.
[32]W. A. Gardner, C. Spooner. Signal Interception:Performance Advantages of Cyclic feature Detectors [J]. IEEE Trans on Communications,1992,40: 149-159P.
[33]C. Spooner, W. A. Gardner. Robust Feature Detection for Signal Interception [J]. IEEE Trans on Communications,1994,42:2165-2173P.
[34]Y. Li, Z. Ding. Blind Channel Identification Based on Second Order Cyclostationary Statistics [J]. IEEE International Conference on Acoustics, Speech, and Signal Processing,1993,4:81-84P.
[35]张晓林,郭黎利,赵毅.直扩信号识别的有限时间平均循环自相关法[J].哈尔滨：哈尔滨工程大学学报.2006,29(6)：900-903页.
[36]王瑛,程汉文,吴乐南.基于谱相关特征的信号调制方式识别[J].信息技术.2006,28(12)：25-28页.
[37]楼才义.一种多信道、多功能数字化接收机.中国电子学会电子对抗分会第十二届学术年会论文集.485-488页.
[38]石远东.宽带数字接收机信道化及其相关技术的研究与实现[D].南京：南京航空航天大学.2007,14-23页.
[39]张明友,汪学刚.雷达系统[M].电子工业出版社.2006,247-271页.
[40]丁鹭飞,耿富录.雷达原理[M].西安电子科技大学出版社.2005,第三版.1-51页.
[41]姚亚峰,黄载禄.通信信号调制识别技术[J].通信技术.2003,6：41-43页.
[42]W. Brown, H. Loomis. Digital Implementations of Spectral Correlation Analyzers [J]. IEEE Trans on Signal Processing,1993,41:703-720P.
[43]韩国栋,邬江兴.调制参数提取的谱相关方法[J].航天电子对抗.2001,3：37-41页.
[44]吕杰,张胜付,邵伟华,谷明.数字通信信号自动调制识别的谱相关方法[J].南京：南京理工大学学报.1999,23(4)：297-299页.
[45]郭黎利.软件无线电中基于谱相关理论的调制模式识别技术[J].哈尔滨：哈尔滨工程大学学报.2004,(6)：25-29页.
[46]张贤达.时间序列分析—高阶统计量方法[M].清华大学出版社.1996,1-24页.
[47]樊昌信.通信原理[M].国防工业出版社.2001,第五版.13-170页.
[48]郭黎利,张晓林.通信原理[M].哈尔滨工程大学出版社.2003,31-136页.
[49]王生兵.无线电信号的调制识别研究[D].南京：东南大学.2006,1-14页.
[50]史建峰.基于周期谱包络的线性调频信号特征参数估计[J].电子对抗技术.2004,19(3)：7-10页.
[51]史建峰.有限数据下循环谱的频域平滑对称式周期图法估计性能分析[J].电子对抗技术.2004,19(2)：155-159页.
[52]Dandawate A V, Giannakis G B. Nonpararnetric Polyspectral Estimators for kth-Order (Almost) Cyclostationary Processes [J]. IEEE Tram on Information Theory.1994,40(1):67～84P.
[53]刘树棠译.现代通信系统(MATLAB版)[M].电子工业出版社.2005,第二版.1-262页.
[54]朱德君.谱相关理论在电子侦察中的应用[J].电子对抗.1995(2)：43-58页.
[55]Yeung G K, GardnerW A. Search-efficient methods of detection of cyclostationary signals [J]. IEEE Trans on Signal Processing,1996,44(5): 1214-1223P.
[56]Roberts R S, Brown W A, Loom is H H. Computationally efficient algorithms for cyclic spectral analysis [J]. IEEE SP MA G,1991,8(2):38～ 49P.