毫米波系统中基于压缩感知的波束空间信道估计

Compressive Sensing Based Beamspace Channel Estimation in Millimeter Wave Systems

波束空间信道估计中大多为窄带信道模型,因此,将基于天线开关网络(Antenna Switching Network, ASN)的导频分组方案应用到宽带信道中,将信道估计构建为稀疏信号恢复问题。针对稀疏信号重构算法中基于广度优先准则的多径匹配追踪算法(Breadth First Based Multipath Matching Pursuit, MMP-BF)需要已知稀疏度的问题,使用分段弱正交匹配追踪算法(Stagewise Weak Orthogonal Matching Pursuit, SWOMP)的阈值准则代替MMP-BF算法的相关性准则,并在每一次迭代结束后根据估计值再次采用阈值准则进行回溯以提高原子选择的准确性。仿真结果表明,在信噪比为0 dB时,改进算法在低导频开销下相比正交匹配追踪(Orthogonal Matching Pursuit, OMP)等传统算法估计精度提高了至少2 dB,相比MMP-BF算法在不同路径个数下具有更好的鲁棒性。

Most of the traditional beamspace channel estimation methods are designed for narrowband systems,therefore,the pilots grouping scheme based on antenna switching network(ASN)is applied to wideband systems,and the channel estimation is constructed as a sparse signal recovery problem.To address the problem that the Breadth First based Multipath Matching Pursuit(MMP-BF)needs to know the sparsity,the threshold criterion of Stagewise Weak Orthogonal Matching Pursuit(SWOMP)algorithm is used instead of the correlation criterion.Also,the threshold criterion is used again at the end of each iteration to improve the accuracy of the atoms’selection based on the estimated values.Simulations show that the improved algorithm improves the estimation accuracy by at least 2 dB compared with conventional algorithms such as Orthogonal Matching Pursuit(OMP)at a signal-to-noise ratio of 0 dB with low pilots overhead,and better robustness than the MMP-BF algorithm for different number of paths.