Font Size: a A A

Research On Emitter Identification

Posted on:2006-04-15Degree:DoctorType:Dissertation
Country:ChinaCandidate:G Z ZhangFull Text:PDF
GTID:1118360155972173Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
In modern electronic war, emitter identification is a important function of ELINT and ESM. With the rapidly developing of electronic technology, the system and modulation manner of radar signals became more and more complicated and various, and circumstance of signals became increasing denseness. It results in that the routine method and theory of recognition can hardly satisfy practical requirement and can't effectively recognize for radar signals. So, the strict demand has been presented for study on recognition of radar signals.For the last several decade years, the scholars have helpfully explored many new methods of solving the problem of emitter recognition. However, these methods were presented in condition of high signal noise ratio (SNR>10dB), fixed signal noise ratio and fewer samples. In practice, the SNR of receiving signals are always low and changed, the observation time of signals is short. It results in that the recognition probability of conventional classification is reduced.Considering the characteristic of radar signal's low SNR, changed SNR and limited observation time, this dissertation includes four aspects. They are the phase noise (PN) distributing of emitter, feature extraction based on artificial modulation of pulse (AMOP), feature extraction based on unmeant modulation of pulse (UMOP) and classifier design. The dissertation presents new excellent features and designs a group of effective single and combined classifiers.In chapter 2, we study the principle of AMOP and UMOP, and draw a conclusion that PN is the main factor of UMOP. So, we discuss the transmitter model, including the model of transmitter's structure, the model of transmitter's phase noise. Then, we investigate the characteristic of PN in detail.In chapter 3, the feature extraction of AMOP is investigated. In this chapter, we present two methods of feature extraction. One is based on image of signal's time frequency distribution (TFD) . This method regards signal's TFD as a image. The first step is image edge acquirement using maximum entropy region segmentation way, and the second step is singular valued decompose (SVD) feature extraction of radar signal's TFD image. The other is based on wavelet transform. In this method, wavelet transform (WD) coefficient spectrum feature is discussed.In chapter 4, the feature extraction of UMOP is investigated. This chapter extracts three types of feature based on WD and higher order statistics (HOS): time field, frequency field and transform field. These features are meaningful to the identification of specific emitter.The classification of radar signals is crucial element for the recognition of radar system. In chapter 5, a new algorithm of modulation identification of radar signals based on support vector machine (SVM) is developed. And the relation between the performance of SVM and kernel function, training set and so on is studied.In chapter 6, two new algorithms of radar type recognition based on combined classifiers are presented. The combined classifiers can achieveeffective modulation recognition in changed SNR.The results of computer simulation and real radar data have shown their good performance of presented methods in this dissertation. This dissertation does a little helpful research on modulation recognition of radar signal. For the future, many works need to do in this aspects.Considering the characteristic of radar signal's low SNR, changed SNR and limited observation time, this dissertation includes four aspects. They are the phase noise (PN) distributing of emitter, feature extraction based on minded modulation of pulse (MMOP), feature extraction based on unmeant modulation of pulse (UMOP) and classifier design. The dissertation presents new excellent features and designs a group of effective single and combined classifiers.In chapter 2, we study the principle of MMOP and UMOP, and draw a conclusion that PN is the main factor of UMOP. So, we discuss the transmitter model, including the model of transmitter's structure, the model of transmitter's phase noise. Then, we investigate the characteristic of PN in detail.In chapter 3, the feature extraction of MMOP is investigated. In this chapter, we present two methods of feature extraction. One is based on picture of signal's Time Frequency Distribution (TFD). The singular valued composition feature of radar signal's time-frequency distribution is presented. The other is based on wavelet transform. The wavelet coefficient spectrum feature is discussed.In chapter 4, the feature extraction of UMOP is investigated. This dissertation extracts three types of feature: the stability of carrier frequency, the characteristic of pulse envelope and higher order statistics coefficient. The results are meaningful to the identification of specific emitter.The classification of radar signals is crucial element for the recognition of radar system. In chapter 5, a new algorithm of modulation identification of radar signals based on support vector machine (SVM) is developed. And the relation between the performance of SVM and kernel function, training set and so on is studied. In chapter 6, two new algorithms of radar type identification based on combined classifiers are presented. The combined classifiers can achieve effective modulation recognition in changed SNR.The results of computer simulation and real radar data have shown their good performance of presented methods in this dissertation. This dissertation does a little helpful research on modulation recognition of radar signal and structure of radar recognition system. For the future, many works need to do in this aspect.
Keywords/Search Tags:Modulation Recognition, ELINT, Unmeant Modulation in Pulse (UMOP), Phase Noise, Feature Extraction, Higher Order Statistics (HOS), Combined Classifiers, Statistical Learning Theory (SLT), Support Vector Machine (SVM)
PDF Full Text Request
Related items