Research On Key Technology Of Radar Signal Recongnition

Radar signal recognition as a key link in electronic warfare plays a pivotal role in the electronic surveillance. And detailed information on radar emitter signals reflects some parameter information of the radar directly or indirectly, and thus the identification of radar emitter signals has become an important part of the electronic warfare. Electromagnetic environment in the electronic warfare is very complex, radiation signal density increases as signal structure complexity increases. There are overlapping parts among the radiation source signals and a huge difference in the intensity of radiation at different times of the same signal, so parameters of conventional radar emitter signals such as PA, PW, DOA, TOA and RF can not meet the growing demands of the new system of complex radar. In this context, finding more effective radar emitter signal parameter vector becomes all the more significant. Among these parameter vectors, the pulse repetition interval(PRI) has an irreplaceable role in the sorting phase of the radar signal, LFM signal is signal is the common form of pulsed radar. LFM instantaneous frequency information obtained is valuable to the subsequent identification of radar emitter signals extract.1. Radar signal sorting is to separate radiation source signals which overlaps together. It emphasizes the integrity of the signal radiation source of information in the environment, especially of sorting pulse sequence for the PRI, the correlation function method and the histogram method is the most basic sorting algorithm, the correlation function method does not work well in pulse sequence sorting about varies and slide change, and harmonic interference seriously. Histogram method is based on statistics: the pulse jitter is 1%, the sorting function performs well. When the pulse jitter exceeds 5%, harmonics interference too interruptive to sort which results in an influence on the sorting process; PRI transform method improves on the base of the correlation function method and the histogram method, in the absence of PRI dither, traditional PRI conversion and improved PRI conversion can conduct PRI estimation effectively when the jitter rate is 5%, the conventional transformation method PRI has lost the ability to estimate the value of the PRI, but the improved PRI transform is able to estimate PRI value at the accuracy rate of more than 90%.when the jitter rate reaches 10%, the improved PRI transform method can not effectively estimate the PRI.2. Time-frequency analysis technique has a great advantage in dealing with non-stationary signals. Given model characteristics of the LFM signal, the instantaneous frequency LFM feature can be extracted through frequency analysis techniques. WVD transform is a typical nonlinear frequency conversion time, but it is difficult to achieve the instantaneous frequency LFM feature from time-frequency analysis. And it will be interruped by cross-term interference and noise. WVD-Hough Transform by image detection algorithm to some extent inhibited the cross-term interference WVD distribution and when SNR is 5d B, the instantaneous frequency estimation of LFM signal accuracy is higher than 85%. In the low SNR at 0d B poor estimate the performance of the algorithm, the error reaches 50%; improved WVD-Hough transform further increase in noise suppression and suppression of cross-term interference, the SNR is 5d B, the LFM signal for instantaneous frequency estimation accuracy is maintained at more than 60%, and in the low SNR 0d B, the new algorithm is estimated to reach 50% accuracy.