Mean-assisted LSTM network spectrum sensing algorithm

Due to the low detection performance of traditional sensing algorithms at low signal-to-noise ratio regions and the large amount of network training and high complexity of deep learning sensing algorithms, this paper proposes a mean-assisted LSTM network spectrum sensing algorithm. This paper first calculates the multi-point average of the received signal, then constructs the feature vector using the calculated average as the LSTM network input to train the network, and finally senses the available spectrum using the trained network. The simulation results show that the detection performance of the proposed algorithm outperforms that of the traditional algorithms, and that the proposed algorithm can achieve lower complexity than the deep learning algorithm trained with the original received sequence.