基于压缩感知的宽带频谱检测

Wideband Spectrum Detection Based on Compressed Sensing

压缩感知是近年来提出的一种针对稀疏信号处理的新方法,其核心是将压缩与采样同步进行,由于信号的投影测量数据量远小于传统方法的数据量,突破了香农采样定理瓶颈从而使得高分辨率信号采集成为可能。频谱感知技术是认知无线电中关键技术之一,它要求次用户在短时间内快速检测出主用户的频谱占用情况。利用认知无线电中频谱的稀疏性,将压缩感知技术用于宽带信号频谱检测,通过少量的压缩数据能够判断频谱是否空闲是一种有效解决这个问题的方法。文中首先建立宽带频谱压缩感知的模型,并提出一种多感知节点多尺度检测算法。该方法将频谱检测分为两个步骤,即粗检测和细检测。在第一步的宽带粗检测过程中,文中分别就高、低信噪比环境下做出讨论,提出了相应的去噪办法;在进一步子带细检测过程中,推导并论证了压缩感知非重构检测算法。仿真结果证实了文中算法的有效性与可行性。

Compression Sensing is a new theory in sparse signal processing field which has been proposed in recent years. It combines the signal compression and sampling simultaneously. It breaks through the bottlenecks of Shannon sampling theorem and makes the high resolution signal acquisition possible since the signal measurement data is far less than those by using the conventional method. Spectrum detection in cognitive radio technology is one of key technologies, it requires the second user to quickly detect whether the primary user occupies the interested spectrum. Because the spectrum is sparse in cognitive radio, by u- sing the compressed sensing technology on wideband spectrum detection with a small amount of com- pressed data, it can find out the idle band. This paper firstly establishes a model of wideband spectrum on compressed sensing, and proposes an algorithm of multi-scales with multi-sensors in rough wideband spec- trum detection process. In this paper, we discuss detection performance in high and low SNR environment and propose the corresponding denoising approach. In the further process of sub-band detection, we de- duce and demonstrate the non-reconstruction compressed sensing detection algorithm. Simulation results verify the effectiveness and feasibility of the algorithm.