| In modern electronic warfare, the feature of the signal receivesd by the electronic reconnaissance equipment is unknown. So it is very necessary and important to analyze the confidence of the result when the signal modulation type is identified and the feature parameters are estimated. This thesis sdudies the confidence of the radar signal modulation type identification.Estimating the frequency of sinusoid is the basis of the intrapulse analyzing of radar signal, and is also a key technique. Compared with several existing algorithms, a new algorithm of frequency estimation is proposed. And a synthetic algorithm is also presented due to the performance complementarities of this algorithm and Rife algorithm in the whole frequency segment. The computation complexity of this synthetic algorithm is a little more than that of FFT, but less than that of M-Rife algorithm. Simulation results show that the performance of the synthetic algorithm is close to M-Rife algorithm, and approach to CRLB when the result is chosen as the starting point for one Newton's iteration.Several methods of signal modulation type identification and feature parameters estimation are introduced and the performance and localization of these methods are analyzed. Then from the view of statistical analysis, two effective methods of confidence analyzing of modulation type identification are proposed. One is a threshold decision based on correlative summation using multivariate hypothesis test and the principle of correlation receivers. For the other method, first, a local signal is generated using the estimated parameters; then a new signal is got by making a correlative accumulation to the local signal and original signal; finally by making curve fit to the new signal, the confidence can be decided by variance analysis and significant test.The simulation results show that the performance of the two methods is very good for judging modulation type identification of familiar radar signal. |
PDF Full Text Request