Short Burst Signal Reconnaissance Technology Research

As we know, short burst signal is hard to intercept and decrypt, so short burst signal technology are gaining more and more application and attention in the military and civilian fields. What’s more, short burst signal have shown its superiority in hard detection and classification in the complicated electromagnetic environment, so it is particularly urgent to research on reconnaissance technology about short burst signal. This thesis focuses on the study of reconnaissance technology about short burst signal, mainly includes signal detection, SNR estimation and modulation classification, the works and contributions of this thesis are as follows:1. CUSUM algorithm and GLR algorithm these two sequential detection methods are recommended to detect short burst signal in the ideal or quasi ideal case. It’s the first time to quote the non-parameter method which is based on the importance fitting to accomplish the blind signal detection in the case with no ststistic information about the signal.2. The traditional Maximum Likelihood Estimation technology of SNR is recommended. Based on the newest theory of Maximum Likelihood Estimation, an affinely modified version of the maximum likelihood estimator is uesed to estimate SNR under the background of burst signal which can deal with short data well.3. The traditional Modulation Classification algorithm based on maximum likelihood is recommended and a modified vesion is proposed to reduce the computational complexity by half. The maximum-likelihood modulation classification based on the amplitude is recommended to overcome the shortcoming which is not robust to some of the signal parameters. Further, a new maximum-likelihood modulation classification based on the Gaussian Mixture Model is proposed which builds a new constellation model. This new constellation model is more robust to the majority of signal parameters. The final simulation results show the new algorithm can execute the modulation classification perfectly.