Modified automatic digital modulation recognition algorithm

[WT9.BZ]In the past, most digital modulation recognition algorithms focused on only memoryless signals, such as PSK, ASK, FSK. A modified algorithm is presented with MSK signals added to the existing set of signals that can be recognized. The method uses simultaneous statistical moments of received signals as their features and uses multi-layer perceptron (MLP) as classifiers. The algorithm's performance has been evaluated by simulating different types of band-limited digital signals corrupted by white Gaussian noise. It is found that the overall success rate is over 96% at the signal to noise ratio (SNR) of 15 dB. Furthermore, when the signal to noise ratio(SNR) is over 10 dB, the overall success rate is not lower than 90% .