软件通信适配器的调制模式识别算法

Modulation classification algorithm for software-designed communication adapter

在异构卫星网络动态组网时,为了解决星上软件通信适配器对物理层调制模式识别率低的问题,提出了一种适合低信噪比和贫先验知识的自动调制模式识别算法.该算法以高斯白噪声信道作为信道模型,选取信号高阶累积量和经典统计量作为特征参数,采用引力搜索算法对径向基神经网络基函数中心进行优化,并在引力搜索算法中引入粒子群的信息熵来调节算法执行过程中探索与开采的关系,进一步提高了算法的分类和泛化能力.然后,利用仿真试验测评了该算法对6种卫星常用调相调制信号的识别效果.仿真试验结果表明,没有先验知识的情况下,该算法在调制信号信噪比大于4 d B时就可以达到100%的识别率,从而证明了该算法在低信噪比和贫先验知识条件下的有效性,说明算法满足星上软件通信适配器对物理层调制模式的识别要求.

To overcome the low recognition rate of the SDCA(software-designed communication adapter)on the satellite for the physic layer modulation during heterogeneous satellite dynamic networking,a novel AMC(automatic modulation classification)algorithm is proposed for low SNR(signal noise ratio)and poor previous knowledge scenario.In this algorithm,the AWGN(additive white Gaussian noise)is used as the channel model.The high-order cumulants and the classical statistics are selected as the features.The classification and generalization capability of the algorithm is enhanced by the IEGSA(information entropy improved gravitational search algorithm)to optimize the basis function center of the RBFNN(radical basis function neural netw ork),using the information entropy of agents to balance exploration and exploitation in iteration.Then,the recognition effects of the proposed algorithm on six kinds of satellite phase modulation signals are evaluated by simulation.The experimental results show that without previous knowledge of received signals,the recognition rate of the proposed algorithm can achieve100%when SNR is above4dB,proving the effectiveness of the algorithm under the condition with low SNR and poor priori knowledge.This algorithm can meet the requirements of the SDCA to the classify modulation mode.