基于矢量近似消息传递的智能反射面辅助毫米波信道估计

Millimeter-wave Channel Estimation with Intelligent Reflecting Surface Assisted Based on Vector Approximate Message Passing

毫米波属于一种典型的视距传输方式,其受大气吸收影响严重。针对毫米波的非视距传播受限,该文通过智能反射面(IRS)辅助毫米波通信,提出结合Khatri-Rao积的矢量近似消息传递(KR-VAMP)算法来提高毫米波通信系统的信道估计质量。该算法基于Khatri-Rao积将级联信道问题转换为稀疏信号恢复问题,并结合VAMP的矢量和迭代阈值算法的优势,使得整个IRS辅助毫米波系统在减少训练迭代次数的同时,降低了整个系统的信道估计误差。最后通过仿真结果对比,分析了各变量对信道估计的均方误差(MMSE)的影响,以及MMSE随着迭代次数的收敛情况,验证了此算法对比其他近似消息传递(AMP)算法具有更好的性能。

Millimeter-wave is a typical line-of-sight transmission method,which is seriously affected by atmospheric absorption.Aiming at the limited non-line-of-sight propagation of millimeter waves,Intelligent Reflecting Surface(IRS)is used to assist millimeter-wave communications,and the Khatri-Rao product combined with the Vector Approximate Message Passing(KR-VAMP)algorithm is proposed,which can improve the channel estimation quality of the millimeter-wave communication systems.By adopting the Khatri-Rao product,the cascaded channel problem is transformed into a sparse signal recovery problem.The proposed algorithm combines with the advantages of the VAMP’s vector and iterative threshold algorithm.The number of training iterations and the channel estimation error are reduced in the IRS-assisted millimeter-wave system.Finally,based on simulation results,the influence of each variable on the Mean Square Error(MMSE)of channel estimation and the convergence of MMSE with the number of iterations are analyzed.It also verifies that the algorithm has better performance than other Approximate Message Passing(AMP)algorithms.