摘要
为了解决LTE-A系统中eNodeB端的同频干扰问题,提出了一种基于自然梯度算法的干扰信号盲分离算法。首先构建eNodeB、蜂窝用户、D2D用户之间的干扰场景,详细分析了复用蜂窝上行资源时三者接收到的混合信号组成。其次针对eNodeB端的MIMO技术同时结合混合信号的构成,给出了干扰信号盲分离的系统框图。仿真结果表明,步长μ=0.001时,收敛速率较慢,随着迭代次数的增加,串音误差曲线未达到稳定状态;当μ=0.006时,曲线震荡严重,稳定性无法保证;当μ=0.002时,曲线趋于稳定,串音误差下降到1 dB,且分离信号与源信号的相关系数均达到了99%以上,算法具有很好的分离效果。
In order to solve the same frequency interference of eNodeB terminal in LTE-A system, a blind separation algorithm based on the natural gradient algorithm is proposed. Firstly, the interference scene among eNodeB, cellular user and D2D user is constructed, and the received hybrid signal of those is analyzed when multiplexing cellular uplink resource. Secondly, the MIMO technology of eNodeB terminal is combined with the composition of the received hybrid signal, the block diagram of the blind separate is proposed. The simulation results show that, convergence speed is slow when μ= 0. 001 and crosstalk error curve is not reach steady state with the increase of the number of iterations ; stability isn' t guaranteed when μ = 0. 006 ; the crosstalk error curve tends to be stable and crosstalk error fells to 1 dB when μ= 0. 002 ,the correlation coefficient of separation signal and source signal reach more than 99% , the algorithm has a good separation effect.
出处
《电视技术》
北大核心
2016年第3期60-64,82,共6页
Video Engineering
基金
河北省自然科学基金项目(F2014201168)
保定市科学技术研究与发展指导计划项目(14ZG036)
河北大学第八批教学改革研究课题项目("基于工程教育认证的通信工程专业实践能力培养机制研究")
河北大学自然科学研究计划项目(2014-303)
河北大学实验室开发项目(sy2015009
sy2015057)
2015年度河北省科技计划自筹经费项目(15210409)
关键词
LTE-A
D2D通信
干扰分离
自然梯度算法
LTE-A
D2D communication
interference separation
natural gradient algorithm