基于压缩感知的宽带频谱感知方案设计

发布时间：2018-04-26 12:02

本文选题：认知无线电 + 频谱感知　；参考：《南京邮电大学》2016年硕士论文

【摘要】：认知无线电是频谱资源利用效率低下问题的有效解决方案之一。在该系统中,主用户可优先使用分配好的授权频谱进行通信,次用户则可在不干扰主用户通信的前提下,动态地接入并使用主用户的授权频谱。次用户在接入授权频谱之前,需要检测并估计周围可以利用的频谱资源,这是认知无线电中的一个重要的问题。在宽带频谱感知中,压缩感知理论利用信号的稀疏性,以远低于奈奎斯特采样速率完成对信号的压缩采样,显著地降低了宽带频谱感知高速率采样所带来的硬件压力。本文主要基于压缩感知对认知无线网络中的频谱感知和动态频谱接入技术展开研究,主要对以下两个方面进行了深入的研究:(1)实际通信中频谱占用具有一定的稀疏性,根据频谱占用的特点,提出了一种基于预测的差分信号压缩感知算法,该算法利用信道占用时间上的相关性,建立了一种信道占用情况的预测模型,依此模型预测出信道占用的变化情况;基于预测结果,在重构信号时可减少频点的搜索范围,两次降低重构算法的运算量。仿真结果表明,在保证感知性能的前提下,新算法可大幅降低迭代次数,减少算法复杂度。(2)频谱预测技术可以有效地指导频谱感知技术,提高系统的感知效率。本文在分析了相关文献的基础上,提出一种基于支持向量机(Support Vector Machine,SVM)的宽带频谱预测方案。在本方案中,通过学习信道的的历史占用状态信息,预测出在下一个感知周期信道的占用状态。该方案充分的利用宽带无线通信在频谱上的稀疏性和信道占用情况的时间相关性,将压缩感知和支持向量机有效的结合起来进行频谱预测,显著地提高了频谱预测的性能。仿真结果表明基于SVM的宽带频谱预测方案在较低的信噪比下,依然具有较高的检测概率和更低的虚警概率。
[Abstract]:Cognitive radio is one of the effective solutions to the inefficient use of spectrum resources. In this system, the primary user can preferentially use the assigned authorized spectrum for communication, while the secondary user can dynamically access and use the primary user's authorized spectrum without interfering with the primary user's communication. It is an important problem in cognitive radio that the secondary users need to detect and estimate the spectrum resources available around them before accessing the authorized spectrum. In wideband spectrum sensing compression sensing theory utilizes the sparsity of signal to complete the compressed sampling at a rate far lower than that of Nyquist which significantly reduces the hardware pressure caused by the high rate sampling of broadband spectrum sensing. In this paper, the spectrum sensing and dynamic spectrum access technology in cognitive wireless networks are studied based on compressed sensing, and the following two aspects are studied deeply. According to the characteristics of spectrum occupation, a predictive differential signal compression sensing algorithm is proposed. Based on the correlation of channel occupancy time, a prediction model of channel occupancy is established. Based on the prediction results, the search range of frequency points can be reduced and the computation of reconstruction algorithm can be reduced twice. Simulation results show that the new algorithm can significantly reduce the number of iterations and reduce the complexity of the algorithm. The spectral prediction technology can effectively guide the spectrum sensing technology and improve the sensing efficiency of the system. Based on the analysis of related literatures, this paper presents a wideband spectrum prediction scheme based on support vector machine (SVM) support Vector Machine (SVM). In this scheme, the historical occupancy state information of the learning channel is used to predict the occupancy state of the channel in the next perceptual period. This scheme makes full use of the sparsity of wideband wireless communication on the spectrum and the time dependence of channel occupancy. It combines compressed sensing and support vector machine effectively to predict the spectrum, and improves the performance of spectrum prediction significantly. The simulation results show that the broadband spectrum prediction scheme based on SVM still has higher detection probability and lower false alarm probability at lower SNR.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN925

【相似文献】