No abstract

No abstract

No abstract

By considering the different cumulant combinations of the 2FSK, 4FSK, 2PSK, 4PSK, 2ASK, and 4ASK, this paper established new identification parameters to achieve the recognition of those digital modulations. The deep neural network (DNN) was also employed to improve the recognition rate, which was designed to classify the signal based on the distinct feature of each signal type that was extracted with high order cumulants. The extensive simulations demonstrated the exceptional classification performance for new key features based on high order cumulants. The overall success rate of the proposed algorithm was over 99% at the signal to noise ratio (SNR) of −5 dB and 100% at the SNR of −2 dB. The results of the experiments also showed the robustness of the proposed method for a variety of conditions, such as frequency offset, multi-path, and so on. INDEX TERMS Modulation recognition, high order cumulants, deep learning, wireless communications.

No abstract