基于卷积神经网络的串行空时分组码盲识别算法

针对多输入单输出(multiple input single output, MISO)系统中的空时分组码(space-time block code, STBC)盲识别问题, 提出了一种基于卷积神经网络(convolutional neural network, CNN)的串行STBC识别方法。首先, 结合STBC识别问题提出了基本CNN (CNN basic, CNN-B)框架; 然后在分析STBC相关性的基础上, 针对空间复用和Alamouti信号混叠问题, 设计了基于相关性的CNN (CNN based on correlation, CNN-BC)模型; 最后将STBC数据集输入到网络模型中, 完成网络的训练和识别测试。仿真结果表明, 相比于基于特征提取的传统算法, 该方法将可识别的STBC扩展到了6种, 并且在低信噪比下的识别准确率更高, 识别过程可控制在微秒级别, 具有较高的工程应用价值。

Blind recognition algorithm of serial space-time block code based on convolutional neural network

Aiming at the problem of the blind recognition of space-time block code (STBC) in multiple input single output (MISO) system, a recognition method of serial STBC recognition method based on convolutional neural network (CNN) is proposed. Firstly, the basic CNN-B network framework is proposed for STBC recognition. Then, on the basis of STBC correlation analysis, a correlation-based CNN-BC network model is designed for the problem of special multiplexing and Alamouti signal aliasing. Finally, the STBC dataset is input into the network model to complete the training and recognition test of the network. The simulation results show that compared with the traditional algorithm based on feature extraction, this method extends the recognized STBC to 6 kinds, and has higher recognition accuracy under low signal to noise ratio, besides, the recognition process can be controlled at the level of microsecond, which has higher engineering application value.