一种改进的认知无线电调制识别方法研究

Modified Modulation Recognition Method for Cognitive Radio

调制识别对认知无线电系统感知和决策具有重要的研究意义。因此,提出了一种认知无线电调制识别的改进方法,适用于低信噪比通信环境,且计算复杂度较小,具有较好的稳健性。该方法基于高阶累积量的特征提取手段,达到了理想的噪声抑制效果;采取堆叠稀疏自编码器作为调制识别分类的新思路,实现了对多层结构的优化,对特征阈值的自动判决也相对简单,可识别多种信号调制样式。最后,仿真结果验证了新方法的有效性,且其适用性较强,识别性能也优于传统的一些算法。

Modulation recognition has important research significance for perception and decision of cognitive radio systems. Therefore, a modified method for cognitive radio modulation recognition suitable for low signal-to-noise ratio communication environment and with less computational complexity and better robustness is proposed. This method based on the feature extraction of high-order cumulants could achieve an ideal noise suppression effect, and with stacked sparse autoencoder as a new idea for modulation recognition classification, realize the optimization of multi-layer structure. And in addition, it has relatively simple auto-decision on characteristic threshold, and can recognize a variety of signal modulation patterns. Finally, the simulation results indicate that this new method is feasible and effective, and strong in applicability, and also better than certain traditional algorithms in recognition performance.