Research On Radar Emitter Signal Feature Evaluation

As an important process of Electronic warfare support (EWS), electronic attack (EA) and electronic protection (EP), radar emitter signal recognition is the key of mastering the electronic warfare battlefield information control, winning the right to win the initiative on the battlefield and the war. With the new system radar gradually becoming the main force of the electronic countermeasures, electronic warfare environment being more and more complex increasingly, combining with adding the anti-radiation weapons and directed energy weapons, various kinds of interference intensified, which have led to the traditional sorting recognition technology facing the dual challenges both modern electronic war complex environmental and electronic countermeasures to radar signal sorting process mechanism. And modern electronic warfare needs high demands for real-time, accuracy and stability increasingly, the sorting and recognition based on the conventional five parameters (carrier frequency (CF), time of arrival (TOA), pulse amplitude (PA), pulse width (PW), direction of arrival (DOA)) cannot adapt to modern electronic warfare electromagnetic environment. How to build a reasonable and effective evaluation index system, as well as comprehensive evaluation model, to select so-called "the Sixth Parameter Group" with excellent sorting and recognition ability to make up for the traditional five parameters has a very important theoretical and practical significance.This paper designs the radar emitter signal feature evaluation indexes from complexity, separability and stability, establishes a comprehensive evaluation model based on fuzzy matter-element analysis to study the evaluation of the characteristics of complex system radar emitter signal. The main work and research results are as follows:1、The investigation about the present situation at home and abroad of feature evaluation of the radar emitter signal is summarized and analyzed. The problems needed to be solved are also pointed out. Besides the structure model of radar emitter signal feature evaluation is presented.2、Wavelet packet feature extraction algorithm is analyzed and improved, instantaneous frequency derived characteristics and intentional modulation characteristics based on time-frequency extraction algorithms are stated and analyzed. For wavelet packet feature has high dimension,"empty packages" due to noise, redundancy between features, reject "empty packages", decrease redundancy and select the features with better sorting and recognition abilities based upon ReliefF algorithm and correlation detection in order to reduce the dimension of feature and improve the correct recognition rate. Experiment simulations on conventional pulse signal, linear frequency modulation signal, non-linear frequency modulation signal, binary phase shift keying signal, quadruple phase shift keying signal, binary frequency shift keying signal, show that the improve on wavelet packet feature algorithm is effective.3、Radar emitter signal characteristics evaluation indexes are designed based on complexity, separability and stability, theoretical framework of the characteristics evaluation indexes, models and algorithms are given. They include the time complexity of feature extraction algorithm (progressive time complexity and the actual time-consuming resource area under the curve), the space complexity of feature(Fisher’s discriminant ratio, feature efficiency, minimum spanning tree boundary complex,1NN’s error rate based on the Leave One Out (LOO), nonlinearity of INN classifier, linear separability of linear classifier, nonlinearity of linear classifier, characteristic dimension), separability, stability (correct recognition’s stability under the noise, correct recognition’s stability under the classifiers, correct recognition’s stability under both the nosie and classifiers, feature’s stability with the change of noise, feature’s stability with the change of carrier frequency, feature’s stability with the change of sampling frequency). Experiment results show that the indexes are effective and operable.4、Comprehensive evaluation model of radar emitter signal characteristics based on fuzzy matter-element analysis is established. Get objective weights of indexes by projection pursuit classification model based on accelerating genetic algorithm with real coding, and use fuzzy matter element to analyze Euclid approach degree of feature and standard feature, and evaluate each feature’s performances comprehensively. Experiment simulations indicate that the model is effective, easy to operate, has good scalability, and it can provide a more comprehensive and accurate evaluation for the performances of signal characteristics, so as to compensate for the deficiencies of the traditional five parameters to add new features to provide a theoretical reference.This work is supported by the National Natural Science Foundation (60971103).