最小相关性和最大-最小相关性MWC重构算法

Minimum and Maximum-Minimum Correlation MWC Reconstruction Algorithms

  • 摘要: 针对调制宽带转换器(Modulated Wideband Converter, MWC)现有重构算法高概率恢复所需的通道数高的问题,提出了MinCor(Minimum Correlation)算法和Max-MinCor(Maximum-Minimum Correlation)算法。MinCor算法利用噪声子空间与感知矩阵的最小相关性求支撑集。Max-MinCor算法联合利用信号子空间与感知矩阵的最大相关性和噪声子空间与感知矩阵的最小相关性求支撑集。实验结果表明,MinCor和Max-MinCor算法在理论条件下,重构所需的通道数均已接近理论极限值。MinCor算法可实现低通道数下的快速重构,但在信噪比小于10dB时,该算法的恢复性能下降;Max-MinCor算法在MinCor算法的基础上提高了抗噪性,但其重构速率低于MinCor算法。以上两种算法均可用于节约硬件成本、减少系统体积。

     

    Abstract: The required fewest channels in the existing reconstruction algorithms of modulated wideband converter(MWC) was far from the theoretical limit, so we proposed minimum correlation(MinCor) algorithm and maximum-minimum correlation(Max-MinCor) algorithm. In MinCor, the noise subspace was regarded as the measurement matrix, and the support was calculated based on the minimum correlation between the measurement matrix and the sensing matrix. In Max-MinCor, we treat the noise subspace and signal subspace as the measurement matrix respectively, and the support was calculated based on the maximum correlation and the minimum correlation. The experimental results showed that both MinCor and Max-MinCor performed better in fewer channels, and the required number of channels is just one more than the theoretical limit without noise. MinCor was much less time-consuming, but its performance decreased if SNR<10dB, while Max-MinCor outperformed MinCor in the anti-noise capability, but did worse in reconstruction speed. The above two algorithms can be used to save hardware cost and reduce system volume.

     

/

返回文章
返回