Study On Fingerprint Recognition Of Frequency-Hopping Signal

Hopping communication is a common military radio communication, also the primary communication means in the field of modern military communication counter measure. It has excellent performance of anti-jamming performance and multi-site networking, which can effectively reduce the probability of communication signals have been detected and the impact of enemy electromagnetic interference. It has great military significance for the army to ensure the smooth flow of communication and command in harsh electromagnetic environment. However, the advantages of the frequency-hopping communication features also increase the difficulty of its research; the characteristics extraction and recognition of the frequency-hopping signal currently are still in the theoretical stage.In this paper, based on the support vector machine, we used the first generation wavelet kernel and second generation wavelet kernel to complish the classification and recognition for frequency-hopping signals, and also pointed out this method’s advantages and disadvantages. Then, extracting the fractal characteristics of frequency-hopping signals as Hopping-signal’s characteristics for the classification and identification of frequency-hopping signals. After that, we used a new method which combined wavelet transform and theory. The specific process is:After wavelet transform, the measured frequency-hopping signal could obtain discrete details and discrete approximation. And so, we extracted the fractal dimensions, which including box dimension, information dimension and LZC. By comparing the performance of those three dimensions, we had selected the box dimension and LZC which have the best clustering effect as the support vector machine input features to finish the classification and recognition experiment of the frequency-hopping signals in different radios.The main areas in this paper are as follows:(1) Compared the two methods which based on the first generation wavelet kernel and the second generation wavelet kernel for the frequency-hopping signals’classification and recognition, and analyzed the experimental results which indicated the advantages and disadvantages of the two identification methods.(2) Introducing the concepts of fractal theory and complexity, by extracting the fractal dimension of the frequency-hopping signal, which including the box dimension, the information dimension and LZC. By comparing the performance of the three dimensions, we have selected the box dimension and LZC which have the best clustering effect for the classification and recognition experiment of the frequency-hopping signal, finally, summarized the result.(3) Using a new method which combined wavelet transform and fractal theory to study the classification and recognition experiment of the frequency-hopping signal. The measured signal would be transformed multi-scale by Daubechies3wavelet, then, we can get the box dimension and LZC which will as the input features to complete the classification and recognition of the frequency-hopping signal. The experiment result shows that the recognition rate has been improved obviously, but there is still more room for improvement.