最大似然准则下的随机信号非重构压缩检测与分析  

Non-reconstruction Compressive Detection of Random Signal using Maximum Likelihood Criterion and its Analysis

作　　者：李斌武[1,2] 李永贵 朱勇刚 

机构地区：[1]解放军理工大学通信工程学院研究生二队,南京210007 [2]南京电讯技术研究所,南京210007

出　　处：《信号处理》2013年第8期996-1002,共7页Journal of Signal Processing

基　　金：江苏省博士后基金项目(1201076C);解放军理工大学预研基金青年项目(KY63ZLXY1203);通信信息控制和安全技术重点实验室基金项目(9140C13030111DZ4603)

摘　　要：压缩采样能够较好的保持原信号的结构和信息,可以在不重构原信号条件下,直接处理采样值实现对原信号的检测。已有的信号压缩检测技术只对确知信号有效,而实际中信号多为随机信号。本文在建立基于非重构压缩采样的信号检测模型的基础上,详细推导了随机信号的最大似然检测算法,并在此基础上分析了算法的检测性能。理论分析与仿真结果表明:在虚警概率一定的条件下,检测概率随压缩比的减小呈对数增加,并且能够在获得高检测概率的同时保持较低的虚警概率;最后进一步验证了算法适用于宽带信号的检测。Compressive sampling theory can effectively maintain structures and information of the original signal, so de-tection tasks of the original signal could be solved by directly processing the samples without reconstructing the original sig-nal. The existing signal detection theory based on CS is directed at the deterministic signal;however, most practice signal is random signal. This paper firstly set up a model of signal detection based on compressive sampling without signal recovery, and then derives the maximum likelihood detection algorithm of random signal in detail, and analysis the performance of this detection algorithm lastly. Theoretical analysis and simulation show that detection probability is increased logarithmic pro-gression with the decrease of the compression ratio for given false alarm probability, in addition, it can achieve very high de-tection rates while simultaneously keeping the false alarm rate very low. Lastly, the applicability of broadband signal detec-tion is verified.

关 键 词：压缩采样 随机信号 最大似然检测 非重构 

分 类 号：TN911.7[电子电信—通信与信息系统]