DPSS基下多带流信号的恢复

发布时间：2018-05-07 20:17

本文选题：压缩感知 + 流信号　；参考：《信号处理》2017年04期

【摘要】：压缩感知技术是近年来信号处理领域最热门的技术。传统的压缩感知理论并未考虑到稀疏信号本身可能具有的某种结构,块稀疏就是其中的一种。本文针对压缩感知的多带块稀疏流信号,将稀疏信号重构算法与调制的DPSS(Discrete Prolate Spheroidal Sequence)基扩展相结合,建立了多带块稀疏模型,并利用压缩感知AIC结构,在远低于奈奎斯特速率下对多带模拟信号进行采样。结合压缩感知获得的观测方程和利用前后窗内信号的相关性建立信号状态转移方程,采用降阶的卡尔曼滤波算法恢复原始信号。相对于傅里叶基扩展,DPSS基扩展在降低采样结构复杂度的同时,克服了频谱泄露的问题。仿真结果表明,多带信号在DPSS基下的恢复性能优于多带信号在FFT基下的重构。
[Abstract]:Compression sensing is the most popular technology in the field of signal processing in recent years. Traditional compression sensing theory does not consider the structure of sparse signal itself, and block sparsity is one of them. In this paper, we combine the sparse signal reconstruction algorithm with the modulated DPSS(Discrete Prolate Spheroidal Sequence) base expansion for compressed multi-band block sparse flow signals, and establish a multi-band block sparse model, and use the compressed sensing AIC structure. Multiband analog signals are sampled at a rate well below Nyquist rate. Combined with the observation equation obtained by compression perception and the correlation of the signal in the front and rear windows, the signal state transfer equation was established, and the reduced order Kalman filter algorithm was used to restore the original signal. Compared with Fourier base expansion, DPSS spread reduces the complexity of sampling structure and overcomes the problem of spectrum leakage. The simulation results show that the recovery performance of multi-band signal under DPSS is better than that of multi-band signal under FFT.
【作者单位】：北京邮电大学泛网无线通信教育部重点实验室;
【基金】：国家自然基金(NSFC 61271181) 教育部国家重点实验室开放基金(Grant 9140C170203140C17084)
【分类号】：TN911.7

【相似文献】