Modulation recognition based on constellation diagram for M-QAM signals

This paper proposed a modulation recognition algorithm for M-QAM signals by the constellation diagram which does not require the prior information. Firstly, this scheme estimates the modulation parameters. Secondly, it reconstructs the received signals´ constellation and uses k-means cluster algorithm to compute the number of the signal constellation points which are as a recognition feature used for classification. The experimental simulation results proves that this method is effective for M-QAM signals, and the correct recognition ratio reaches 100% under the condition that the SNR is higher than 15dB. So it has relatively preferable practical value.