【作者】 李广威；

【导师】 郑侃；

【作者基本信息】 北京邮电大学 ， 电子与通信工程（专业学位）， 2013， 硕士

【摘要】 压缩感知理论是一种新的信息获取方法,能够以远低于耐奎斯特定律的采样速率对稀疏信号进行采样,并精确地进行重构。无线信道的冲激响应通常呈现出稀疏的分簇多径特性。利用此性质并考虑OFDM系统本身的一些特性,如最大相对时延小于循环前缀长度,将压缩感知理论应用到OFDM系统的信道估计中,能够减小导频开销,降低导频图样限制,具有较高的估计精度。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结构,块稀疏就是其中的一种,即非零元素成块出现。利用这一特点,可以进一步的降低采样速率,并提高重构精度。然而现有的块稀疏贪婪重构算法,如BOMP,只能够针对块长相同,且非零块间的距离为块长的整数倍这一特例进行很好的重构。因此针对这个问题,文章中提出了一种改进的BOMP算法,它能够适用于更加一般的块稀疏信号,并且获得更好的重构效果。本文还将块稀疏信号重构算法与DPSS基扩展相结合,进行时延多普勒域信道信息的估计。相对于傅里叶基扩展,DPSS基扩展在降低复杂度的同时,克服了能量泄露的问题,并降低了对多普勒频移的敏感度更多还原

【Abstract】 Compressed sensing (CS) is a new signal acquisition method. It enables signals to be sampled below the Nyquist rate given that the signal is sparse, with an accurate reconstruction. Because of the multipath effect, the channel state information in delay domain has sparsity. When we apply CS theory into channel estimation of OFDM system, the pilots’ number decreases, and the accuracy increases.The traditional CS theory doesn’t concern the natural property of the sparse signal, such as block-sparstiy. Applying with this, we can improve the recovery accuracy and promote the efficiency. However, most of the exist reconstruction algrithms, such as BOMP can only get wonderful performance when the lengths of nonzero blocks are the same, and the intervals between them are integral multiple of the length. So, we propose a modified BOMP algorithm for the more general block sparse singals in this paper. And we combine the reconstuction of block-sparsity singal with DPSS basis expand model to estmate the channel state information in time-delay domain. In this way, we can overcome the energy leak of Fourier basisi expand model, and reduce the sensitivity to Doppler shift.更多还原