采用支持向量机的宽带频谱感知算法

发布时间：2018-06-23 02:15

本文选题：频谱感知 + 压缩感知　；参考：《信号处理》2014年12期

【摘要】：认知无线电系统中,压缩感知理论已广泛运用于宽带频谱检测。但是,压缩感知中的重构问题造成频谱检测算法计算复杂度高,且在低信噪比下检测效果不佳。本文提出了采用支持向量机的宽带频谱感知算法,该算法利用支持向量机建立频谱检测分类器,代替信号的重构与检测过程。根据系统对实时性的要求,分别设计了多级二元分类器感知算法和单级多元分类器感知算法。前者适用于分级数有限且实时性要求不高的场景,后者可大幅降低系统的算法复杂度,降低感知时间,适用于实时检测系统。仿真结果表明,与基于重构的能量检测算法相比,本文提出的两种算法均可以有效改善系统对噪声的鲁棒性,提高在较小信噪比下的检测性能。
[Abstract]:In cognitive radio systems, compressed sensing theory has been widely used in broadband spectrum detection. However, the reconstruction problem in compressed sensing results in high computational complexity of spectrum detection algorithm and poor detection performance under low SNR. In this paper, a wideband spectrum sensing algorithm using support vector machine (SVM) is proposed, which uses support vector machine (SVM) to build spectrum detection classifier instead of signal reconstruction and detection process. According to the real time requirement of the system, the multilevel binary classifier perception algorithm and the single level multivariate classifier perception algorithm are designed respectively. The former is suitable for scenes with limited hierarchical number and low real-time requirements, while the latter can greatly reduce the algorithm complexity and perceptual time of the system, and is suitable for real-time detection systems. Simulation results show that compared with the energy detection algorithm based on reconstruction, the two algorithms proposed in this paper can effectively improve the robustness of the system to noise and improve the detection performance in the case of small signal-to-noise ratio (SNR).
【作者单位】：南京邮电大学"宽带无线通信与传感网技术"教育部重点实验室;
【基金】：国家自然科学基金项目(61271335,61471202) 国家重点基础研究发展计划(973计划)(2011CB302903) 江苏省高校自然科学研究项目(13KJB510020) 江苏省高校自然科学研究重大项目(14KJA510003)
【分类号】：TN925

【参考文献】