Modulation Recognition Of Digital Communication Signals In Multipath Channels

With the rapid development of the communication technology, the system of communication signal and the modulation mode are becoming increasingly diversified, and the communication environment tends to be complex. That also made communication signals modulation recognition become more difficult than ever. Now most of methods of modulation recognition are based on the ideal condition, gaussian white noise interference or the small forward of multipath interference. So far, under the condition of low SNR and jam strong of multipath, the classification of modulation is still difficult.In the multipath channl with the Gaussian noise, a modulation identification method based on signal circulation spectrum is presented in this paper. The method has extracted six anti-multipath characteristics in the signal spectrum domain. First in circulation spectrum domain, we set a threshold to reduce the interference of the noise to the characteristic. Then using the nature of the signal's circulation spectrum, the correlation coefficients of some sections of frequency and circular circulation are calculated which overcome multipath interference. Theoretical analysis and experiments demonstrate that these characteristics parameter can eliminate the effects of parameters multipath channel, which are the effective and stable signal modulation recognition characteristics. After this paper proposed an improved genetic algorithm to selection a congregate from the above six characteristics whith is the most suitable for modulation recognition of digital signal. The method can adaptive adjusts crossover and mutation probability regulation by the change of outstanding population size which caused by a choice, elimination, then selects the optimal combination of the characteristic for the recognition. Finally by the threshold hard decision tree and RBF neural network to distinguish all the signals, the simulation results illuminate that this method has good performance, and is validity in multipath and low SNR channel.