Research On Recognition And Deception Jamming Method Based On Micro-Doppler Signatures For Ground Slow-Moving Target

Since the introduction of micro-Doppler (m-D) effect into radar field, it has found wide applications in radar target detection, feature extraction, classification and recognition, parameter inversion and so on. In addition to the bulk motion of the human torso's center of mass, the micro-movements from shoulders, limbs, hands, feet and other parts are also of great importance and should not be neglected. For vehicle target, except the bulk motion of the vehicle center, there are micro-motion from wheels and other parts. These two types of ground moving targets have typical micro movements. While humans and vehicles are the main monitoring targets for ground surveillance radar. However, most of current ground surveillance radars are narrow-band radar. It is rather difficult for the traditional methods based on Fourier transform (FT) and Doppler analysis to achieve well classification and identification results of humans and slow-moving vehicles. However, the m D signatures of humans and vehicles are their inherent and unique characteristics, which not only contain target instinct attribute information including target size, material, etc., but also the moving information. Therefore, the study on m-D signature analysis will be able to enhance the capabilities of identification and information acquisition of the ground surveillance radar. Meanwhile, the non-stationary characteristic of the m-D signal poses a great challenge to ground slow-moving target recognition based on m-D signatures. On the other hand, since the m-D signature of targets is critical for recognition, it is also a necessity for successful deception jamming. The m-D signal of ground moving targets contains a fine motion information of the ground moving targets, how to make full use of the target m-D information and achieve a further refinement to deception jamming is a hot and difficult problem in the field of radar signal processing and electronic warfare. This dissertation is funded by the National Natural Science Foundation of China, National'863'project and the crosswise project. With the practical application requirements of target detection and recognition for ground slow-moving target, the research of recognition and deception jamming based on m-D signatures for ground slow-moving targets is carried out. This dissertation focuses on the issues of m-D feature extraction for ground slow-moving targets, refined recognition for micro-moving parts and refined deception jamming method based on m-D signatures. The main contents are summarized as follows:1. On the basis of target micro-motion signature analysis, the universal models of typical micro-motion including vibration, rotation and oscillation are established and the analytical expressions of m-D corresponding to different micro-motion models are derived. For the problems of strong non-stationary characteristic and low resolution of conventional FT method, the short time FT (STFT), Wigner-Ville distribution (WVD) and smoothed pseudo WVD are selected to analyze the m-D signatures of different micro-motions in detail, which provides a theoretical basis for the following research on m-D feature extraction, recognition and deception jamming of ground slow-moving targets.2. It is difficult to detect and classify the humans and vehicles based on the measurements of narrow-band radar system. By making use of their respective m-D signature difference, a method of feature extraction and classification for ground slow-moving target based on the textural information of time-frequency (TF) spectrograms is proposed. A superior result of the measured data for humans and vehicles is obtained. Firstly, continuous wave radar data contains a large amount of background clutter, a ground clutter statistical model is established to study the effective clutter suppression methods. Secondly, on the basis of micro-motion model and characteristic of humans and vehicles, STFT is utilized to obtain the spectrogram samples. Pre-processing for the samples further improve the signal-clutter-noise ratio of the spectrograms. Then, three textural features including spectrogram entropy, the third order moment of histogram and directionality are extracted to form an effective feature library. Further, the texture feature library is divided into a training set and a testing set, which are fed into support vector machine classifier sequentially in order to achieve effective classification of humans and vehicles. Finally, simulation and experimental data processing results verify the effectiveness and robustness of the proposed method in a low signal to noise ratio and less training samples situation.3. Human micro-motion parts has complex moving mechanism and it is difficult to recognize and separate out different parts. Combined with the statistical properties of the echo from each part of human body, a refined recognition method of human main micro-motion parts based on principle component analysis (PCA) is proposed, which is better than the existing method based on complex empirical mode decomposition. Firstly, on the basis of the micro-motion characteristic of humans, a refined model of body parts is established, and the statistical properties of each part are analyzed in detail. Secondly, a sliding window technique is utilized to smooth the time series of human echo. The number of PCs are determined by minimizing the Akaike information criterion, which provides the basis of PCA decomposition. Next, decompose the human echo by PCA method to obtain the main feature vectors. Reconstruct the signals by the inverse sliding widow technique to obtain the PCs which denotes the principle components of human. Then, the clustering algorithm based on Ixegram is adopted to recombine the similar PCs in order to achieve the refined recognition result of human main micro-motion parts. Finally, the simulation result demonstrates the effectiveness of the proposed method.4. Traditional deception jamming is easy to be identified without considering the micro-motion of vehicle. To tackle this problem, based on the analysis of the m-D characteristic of the vehicle micro-motion parts, a refined deception jamming method based on m-D signature for vehicle target is proposed, which improves the fidelity of the deceptive target. Firstly, the basic principles and drawback of existing traditional deception jamming method is analyzed. Secondly, the jammer intercepts the narrow band radar signal, locate the radar through electronic surveillance radar system. Thirdly, the translational information is modulated to generate the false vehicle target. Next, according to the m-D signature and attitude information such as the false target azimuth, pitch angle, position, construct the micro-motion modulation function. Then re-modulate the deceptive signal after the translational modulation. Retransmit the modulated signal to accomplish the deception jamming of vehicle target based on m-D signature. The refined jamming effect, the effect on deception jamming of carried frequency error and the computation burden is analyzed. Finally, the simulation results verified that the proposed method is capable of producing deceptive vehicles with high fidelity.