Modulation Recognition Based on Combined Feature Parameter and Modified Probabilistic Neural Network

Criterions that weigh the performance of modulation recognition algorithm were presented according to the modulation recognition process and purpose. In terms of these criterions, the combined feature parameters were extracted exploiting the propriety of the cyclic spectral correlation algorithm, these parameters consisted of parameters of cyclic spectral correlation and frequency field. The modified probabilistic neural network was used as a classifier algorithm. The modulation types can be identified dynamically availing oneself of the presented combined feature parameters and the classifier algorithm. The new algorithm presented in this paper can improve the recognition performance and expand the range of recognition, but it does not increase the computational complexity simultaneously. The simulation results obtained by Monte-Carlo method proved these