Research On Modulation Recognition Algorithm For Multiple Signals

Modulation recognition of signals is an important topic in the field of signal processing.The purpose of modulation recognition is to recognize the modulation types of the receivedsignals and estimate the related parameters, and provides the basis for the demodulation. Theexisted modulation recognition methods mainly employed single signal as the model. Butwith the communication environment increasingly complex, the existence of interferencesignal can make two or more than two signals simultaneously exist in the range of the receiverband, therefore the modulation recognition of multiple signals has practical significance. Thisdissertation researches the recognition methods of single signal, and on this basis selects orreconstructs characteristic parameters which fits to distinguish the multiple signals, andachieve the modulation recognition of multiple signals. The main work of this dissertation canbe summarized in three aspects as follows:Firstly, on the basis of analyzing the cyclic spectrum of MPSK signals and thesuperposition of cyclic spectrum, a parameter estimation method of two MPSK signals basedon cyclic spectrum is put forward. This method extracts cyclic frequency domaincharacteristics of the mixed signals’ cyclic spectrum to realize the estimation of symbol rate,and on this basis, estimates the carrier frequency through searching the maximum in thesection that cyclic frequency is equal to symbol rate.Secondly, aim at the modulation recognition of multiple signals based on signalseparation, this dissertation respectively proposes the method based on EMD and fractal boxdimension and the method based on MVDR beamforming to achieve modulation recognitionof two MFSK signals. The method based on EMD firstly uses empirical mode decompositionto separate the mixed signal that to be recognized into multiple components, and then extractsthe fractal box dimension characteristics of components which have been separated to realizecategories judgment. The method based on MVDR firstly uses MVDR beamforming toseparate the mixed signal, and then the instantaneous frequency normalized variance of theseparated signals are extracted as characteristic parameters to realize categories judgment.Finally, aim at the modulation recognition of multiple signals based on direct featureextraction, a recognition method of two MPSK signas based on constellation diagram and a recognition method based on cyclic spectrum are respectively proposed. The method based onconstellation diagram firstly estimates the carrier frequency by using cyclic spectrum.Secondly signals’ constellation diagram can be reconstructed b using subtractive clustering,and then extracts the number of cluster centers as characteristic parameter to achievecategories judgment. The method based on cyclic spectrum is aiming at the mixed signalwhose carrier frequency and symbol rate are the same, and in this moment we can not shieldthe other signal by selecting one of the signals’ characteristic, so new characteristicparameters are constructed in both frequency domain and cyclic frequency domaincharacteristics to realize categories judgment.