| Modulation recognition and blind equalization are two key technologies of non-cooperative communication, which possess very important actual meanings for blind receiver of the intercepted signal and rear-end information restoration. Based on statistical pattern recognition theory, in this paper, from the angle of third party in the non-cooperative communication, a number of key technologies which is involving in the process of modulation recognition, such as symbol rate precise and blind estimation, symbol synchronization, characteristic parameters extraction, classifier design, has been discussed in detail. Furthermore, blind equalization algorithm has been deeply studied to remove the effect of intersymbol interference existing in the received signal under the multi-path channel environment. The main works of this paper are as follows:1. The cyclostationary and spectrum line generation feature of the received signal has been discussed. The discrete characteristic spectral line of the squaring MPSK/OQPSK/MQAM signal has been analyzed. The method of detection discrete characteristic spectral line and the method of enhancing spectral resolution have been deeply studied. A symbol rate precise estimation algorithm which is combined with FFT-MAC-CZT is proposed. the normalized bias of the symbol rate estimator for MPSK/OQPSK/MQAM signal can be controlled at 10-6 order of magnitude by this algorithm.2. When signal modulation type is unknown, and carrier frequency offset and phase offset is existing, from the angle of third party in the non-cooperative communication, the Gardner Timing Synchronization algorithm has been analyzed and is applied to achieve the best samplings of received signal.3. The discrete characteristic spectral line of the non-linear transforming MPSK/OQPSK/ MQAM signal has been analyzed and the modulation recognition algorithm for OQPSK and MPSK has been presented. The feature parameters of higher-order cumulants which are insensitive to additive white gaussian noise,carrier frequency offset and phase offset are extracted. Symbol amplitude characteristic has been studied after symbol synchronization. A new algorithm constructing symbol amplitude density distribution has been proposed and a set of feature parameters has been extracted, whose ability of anti-noise is relatively strong.4. The method of classifier design based on decision tree has been studied, and the corresponding classifier has been designed according to the feature parameters of MPSK/OQPSK and MQAM signal. thereinto, according to the feature parameters of square MQAM signal, the method of threshold of decision interval partition has been adopted to improve correct recognition rate of the square MQAM signal. The method of classifier design based on RBFNN has been studied, and a new training algorithm for RBFNN has been proposed which are combined with the subtraction clustering and the least mean square(LMS) algorithm.5. The Bussgang blind equalization algorithm has been deeply studied, and the merits and demerits for the algorithms of the CMA,MCMA,DD-LMS and dual-mode blind equalization are analyzed; A carrier frequency offset estimation algorithm based on oversampling signal model is proposed; A jointed algorithm combined with carrier synchronization and blind equalization has been designed. The algorithm is able to realize carrier synchronization and remove the effect of intersymbol interference at a bigger frequency deviation region at the mean time. The channel blind identification and blind equalization algorithm based on subspace decomposition is investigated, and the factors that may affect the algorithm such as the number of statistic samples,SNR,carrier frequency offset and phase offset have been analyzed.6. System realization of the related algorithms has been completed. The testings for the actual satellite received signal has been done to validate usefulness of the algorithms invovling in this paper in practical situations. |
PDF Full Text Request