Xu Xiguang, Qu Hua, Zhao Jihong, Yan Feiyu, Wang Weihua
School of Electronic and Information Engineering, Xi'an Jiaotong University, Xi'an 710049, China.
Suzhou Caiyun Network Technologies Co., Ltd., Suzou 215123, China.
Entropy (Basel). 2018 Apr 3;20(4):246. doi: 10.3390/e20040246.
Spectrum sensing is the most important task in cognitive radio (CR). In this paper, a new robust distributed spectrum sensing approach, called diffusion maximum correntropy criterion (DMCC)-based robust spectrum sensing, is proposed for CR in the presence of non-Gaussian noise or impulsive noise. The proposed distributed scheme, which does not need any central processing unit, is characterized by an adaptive diffusion model. The maximum correntropy criterion, which is insensitive to impulsive interference, is introduced to deal with the effect of non-Gaussian noise. Simulation results show that the DMCC-based spectrum sensing algorithm has an excellent robust property with respect to non-Gaussian noise. It is also observed that the new method displays a considerably better detection performance than its predecessor (i.e., diffusion least mean square (DLMS)) in impulsive noise. Moreover, the mean and variance convergence analysis of the proposed algorithm are also carried out.
频谱感知是认知无线电（CR）中最重要的任务。本文针对存在非高斯噪声或脉冲噪声的认知无线电，提出了一种新的鲁棒分布式频谱感知方法，即基于扩散最大相关熵准则（DMCC）的鲁棒频谱感知。所提出的分布式方案不需要任何中央处理单元，其特点是采用自适应扩散模型。引入对脉冲干扰不敏感的最大相关熵准则来处理非高斯噪声的影响。仿真结果表明，基于DMCC的频谱感知算法在非高斯噪声方面具有出色的鲁棒性能。还观察到，在脉冲噪声中，新方法比其前身（即扩散最小均方（DLMS））具有明显更好的检测性能。此外，还对所提算法进行了均值和方差收敛性分析。
