微型电脑应用

针对传统认知无线电频谱感知时间长、计算量大的问题，提出一种改进LSTM的无线电频感知算法。具体改进则是在三个门输入部分增加一个上一时间步的记忆单元作为输出状态值，从而提高神经网络的学习能力；以改进的LSTM为基础，构建改进LSTM的OFDM频谱感知模型。通过仿真实验对算法性能进行验证，仿真结果表明，提出的改进LSTM的OFDM信号频谱感知算法相较于传统频谱感知算法，可提高对认知无线电频谱的获取能力，具有更好的频谱感知性能。

Aiming at the problems of long time and heavy computation of traditional cognitive radio spectrum sensing methods in short wave channel, this paper proposes a cognitive radio frequency sensing algorithm based on improved LSTM. Firstly, the improvement of LSTM network is studied, and a memory unit is added to output state value of the last time step in the input part of the three gates of LSTM. Then, taking the widely used OFDM wireless communication technology as an example, the OFDM signal spectrum sensing algorithm model based on the improved LSTM is constructed, and training and testing are completed. Finally, the algorithm is verified by simulation experiments. The performance of the proposed method is verified and compared with the traditional signal detection methods LS algorithm and MMSE algorithm. The results show that the proposed spectrum sensing algorithm based on improved LSTM can improve the spectrum acquisition ability of cognitive radio and has better spectrum sensing performance.