基于局部保持投影的M-QAM调制信号识别

M-QAM Signal Recognition based on Locality Preserving Projection

针对基于统计的QAM信号识别算法,忽略了信号的局部特性,导致算法性能不好等问题,提出了一种基于流形学习的16QAM、32QAM、64QAM信号识别算法。该算法利用高阶累计量特征描述信号,在此基础上利用邻接图描述特征的内在几何属性,较好地刻画了数据的相似性几何属性,最后利用最近邻分类器算法进行分类。实验结果表明,该算法具有好的识别率,尤其在低信噪比下,算法性能比较突出。

Quadrature amplitude modulation （QAM） signal recognition algorithm based on statistics ignores the lo- eal eharacteristies of the signals, which results in the degradation of the performance. To address this issue, a new manifold based algorithm is proposed for the recognition of 16QAM, 32QAM and 64QAM. This method employs higher-order aceumulation to describe the signal features, and uses the adjacency graph to depict the intrinsic geometrieal properties of the features, which can better charaeterize the similarity of the data; then a nearest neighbor classifier is used to achieve classification. The experimental results show that the proposed algorithm has better recognition performance, especially in the event of low signal to noise ratio （SNR）