【作者】 王臣昊；

【导师】 杨震；

【作者基本信息】 南京邮电大学 ， 信号与信息处理， 2012， 硕士

【摘要】 众所周知,如今无线电频谱资源在全球范围内都很稀缺。为了提高频谱的利用率,1999年学者J.Mitola提出了认知无线电这个概念。在认知无线电中,认知用户可以使用频谱检测技术在授权信道寻找空闲信道进行通信。可见,频谱检测技术是整个认知无线电系统中保证成功的关键技术之一。此外,当我们对宽带信道进行频谱检测时,使用传统的奈奎斯特采样定律会产生海量的采样数据,而现在的硬件水平很难满足这一信号的快速处理需求。幸运的是,近年出现的压缩感知理论给这一困境带来了转机:一方面,压缩感知理论允许稀疏信号以低于奈奎斯特采样定理的速率进行采样,大大减轻了硬件的处理压力;另一方面,无线信号在频域上天然的稀疏特性又符合压缩感知理论的前提之一——原始信号的稀疏性。因此基于压缩感知的认知无线电频谱检测技术是一个可行且很有价值的研究方向。本文在第一章中介绍了本课题的研究背景,介绍了认知无线电的相关理论,特别是对频谱检测技术做了详细的阐述。在第二章中本文详细的论述了压缩感知理论和其当前的研究成果。在第三章中本文研究了压缩感知重构算法之一的最小l₁范数法,结合迭代加权和约束条件l₁范数化,提出了自己的最小l₁范数法的改进算法,并引入频谱检测,用仿真验证其在频谱检测中的效果。在第四、五章中本文研究了贝叶斯压缩感知的理论和引入优化高斯随机观测矩阵贝叶斯压缩感知理论,并将两种算法分别引入频谱检测技术进行仿真研究;最后以能量检测法和使用能量判决的BCS频谱检测方法为代表分析了传统频谱检测法与压缩重构频谱检测法的优劣。更多还原

【Abstract】 As we all know, the wireless spectrum resource is very scarce globally. In order to improve the utilization of the wireless spectrum, the conception of Cognitive Radio （CR） was proposed by J. Mitola in 1999. In CR theory, a cognitive user can first search an unoccupied authorized channel, and then use it for transmitting signals. So obviously, the spectrum detection is one of the key technologies in CR. And as we also know, when it comes to detection for the wideband spectrum, the samplings should be two times of the bandwidth according to the Nyquist’s Theorem, which made the hardware hard to burden. Luckily, the appealing of the Compressed Sensing （CS） theory changes the situation. On one hand, the CS theory permits the hardware to deal with the signal far below the Nyquist’s Sampling Rate, which can release the pressure on the hardware’s processing ability. On the other hand, the radio signals are sparse in the frequency domain, which satisfied one of the preconditions to apply the CS theory on. Thus, the research on the CS-based spectrum sensing technology is available and quite meaningful.The first chapter of the paper introduces the background of our research and the fundamental knowledge of CR. The second chapter of the paper introduces the CS theory detailedly. The third chapter of the paper focuses on the l₁ norm minimum algorithm and its improved algorithms. We propose a new improved algorithm, which combines the iterative weighted l₁ norm algorithm with the constraints of l₁ norm. The simulations show that the improved algorithm works well in the spectrum detection. In the forth and fifth chapter of the paper, our research is focused on the Bayesian CS （BCS） algorithm and an optimized BCS （OBCS） algorithm. Then we simulate them in the spectrum detection scene. At last we take energy detection and BCS spectrum detection which judged by energy as the examples to compare the traditional spectrum detection and the compressed sensing spectrum detection.更多还原