Dimensionality reduction and improving the performance of automatic modulation classification using genetic programming

Modulation recognition is one of the main parts of modern communication receivers. The automatic modulation recognition of received signal is considered as the intermediate step between signal detection and its demodulation. In most military and communication systems, modulation detection is considered as a part of whole system. The existing methods for this purpose are based on modulated signal’s components. In this paper, the features of modulation classification were decreased using genetic programming. The performance of modulation classification can be improved using genetic programming. The performance improvement was compared between Multi-layer neural network perceptron, and decision tree. The purposed method was designed for 16QAM ، 64QAM ،2FSK ،4FSK ،2PSK ،4PSK modulations. The simulation results show that this method has high accuracy and good convergence rate in presence of noise.

کلیدواژه ها:

modulation automatic detection ، genetic programming ، entropy ، Multi-layer neural network perceptron ، and decision tree