Modulated wideband converter (MWC) was a sub-Nyquist sampling system for sparse multiband analog signals.In the actual design of the MWC,a variety of factors should be considered,such as the hardware implementation and the accurate reconstruction,which might result in high redundancy in the sampled data.This paper deeply analyzed the reasons why the MWC might cause redundancy in the sampled data and the characters of the sampled data for the first time,and proposed a compression method for the sampled data based on the principal component analysis (PCA).The proposed method firstly used PCA transform to concentrate most of the energy of the MWC sampled data on few principal components,and then compressed the sampled data by only retaining,quantizing and encoding these principal components.The experimental results show that the ability of the PCA transform to concentrate the energy of the MWC sampled data outperforms the wavelet transform and the DCT.On the premise of ensuring the reconstruction accuracy higher than 90％,the proposed method can compress the sampled data to less than 1/8 of the original one.%调制带宽转换器(modulated wideband converter,MWC)是针对稀疏多频带信号提出的一种亚奈奎斯特采样方法.实际MWC设计中需考虑硬件实现、重构精度等问题,导致采样数据中存在一定冗余.对MWC系统产生采样冗余的原因及采样数据特点进行深入分析,并提出一种基于主成分分析(principal component analysis,PCA)的MWC采样数据压缩方法.对MWC采样数据进行PCA变换,将其中的大部分能量集中在少量主成分中,进而通过保留少量主成分,并对其进行进一步量化编码的方式来实现对MWC采样数据的压缩.实验结果表明,PCA变换对MWC采样数据的能量集中效果优于小波变换和离散余弦变换,提出的压缩方法能够在保证重构精度高于90％的前提下将MWC系统的采样数据量压缩至原来的1/8以下.
展开▼
