A Study On Multi-scale Feature-level Identification Of Active Deception Jamming

With the development of electronic jamming technology, under more powerful jamming, the radar is facing more serious challenges. In the sophisticated electronic jamming environment, radar systems should intelligently select the most effective anti-jamming methods according to the type of interference signal, which has become an important direction of the development of radar anti-jamming technology.On the basis of the existing theoretical knowledge and related dissertations, this dissertation studies the generation and implementation mechanism of the conventional active deception jammings. The multi-scale decomposition theory is applied to the classification and identification of three types of pull-off jammings and their hybrid interferences in the feature level. Research contents in this dissertation are listed as follows:1. To solve the problem that a single pulse cannot reflect the essential characteristics ofthe pull-off jamming producing and implementation process, effective parts of a pulsesequence in a coherent processing interval are splicied together into a dimension vector.In this dissertation, the effect of the pulse echo number in a coherent processing intervalon feature distinguish degree and the final recognition results is discussed. Simulationexperiments show that the identification result of multi-pulses on the feature level isbetter than a single echo, and the more pulses it has, the better classification results canbe obtained.2. Wavelet decomposition and empirical mode decomposion are two kinds of multi-scale decomposition methods which are used to decompose the received signal. Features such as high-frequency detail component energy ratio and low-frequency approximation component normalized energy are extracted from wavelet coefficients decomposed from the received three kinds of the pull-off jamming signal by the wavelet decomposition. Skewness, kurtosis and noise factor in frequency domain are extracted from the intrinsic mode functions decomposed the received signal by the empirical mode decomposion. After selecting from the extracted features, better features in different jammer-to-noise ratio can constitute a small feature database. Classification and identification of the pull-off jammings can be accomplished under the established feature database. Simulation results show that, through a multi-scale decomposition, the extracted characteristic parameters can effectively distinguish three types of towing interference, and the final recognition result is satisfactory.3. Taking the actual situation that a variety of radar active deception jammings exist simultaneously into account, for seven cases of the mixed signal, multi-scale decomposition algorithms are applied to extract wavelet coefficients features and the features in frequency domain, and their statistical mean in different jammer-to-noise ratio are calculated to create a small mixed interference feature database which can assist in type recognition of the interferences. Simulation results indicate that the algorithm that combines the multi-scale decomposition and the feature database can be relatively effective to distinguish among the seven cases. This method is effective and meaningful.