基于测量向量转换的MWC支撑集恢复算法

A Joint Support Set Recovery Algorithm of MWC Based on Measurement Vectors Transform

针对调制宽带转换器欠采样重构算法中连续到有限模块计算复杂度高、计算速度较慢的缺陷,提出一种基于测量向量转换的支撑集恢复算法,通过对测量向量进行频率积分,将无限测量向量转换为单测量向量进而实现支撑集的恢复。该方法只需对频率积分便可实现无限测量向量到单测量向量的转换,避免了以往方法中将无限测量向量转换为多测量向量时,矩阵特征值分解的高复杂度运算,而且后续对单测量向量模型求最稀疏解的过程也比以往算法中求解多测量向量模型更简单快速。实验表明提出的方法在适当的采样数据量下至少可以提高3倍的恢复速率。

A new method called MVT was proposed based on measurement vectors transform to get the support set. By integrating in frequency domain,the infinite measurement vectors model was transformed into single measurement vector model which avoids the high computational procedure in original method like eigenvalue decomposition. Moreover,the procedure solving for sparsest solutions in multiple measurement vectors model is much more time-consuming compared with single measurement vector model. The experimental results showed that MVT increases the recovery rate at least three times under appropriate number of sampling data.