Welch FFT Segment Size Selection Method for FFT Based Wide Band Spectrum Measurement
Welch FFT Segment Size Selection Method for FFT Based Wide Band Spectrum Measurement
复制标题
DOI：
10.1587/transcom.2017ebp3069
复制
发表时间：
2018-07-01
影响因子：
0.7
通讯作者：
Narieda, Shusuke
中科院分区：
文献类型：
--
作者：
Iwata, Hiroki;Umebayashi, Kenta;Narieda, Shusuke
在智能频谱接入(SSA)环境下，提出了一种基于FFT的宽带频谱测量的Welch FFT段大小选择方法，其中主用户(PU)的统计频谱使用信息(DC)将被次用户(SU)利用。基于韦尔奇FFT的能量检测器可以有效地检测宽带环境中PU信号的存在，如果设置合适的韦尔奇FFT段大小，则可以正确估计DC。在韦尔奇FFT段大小方面，存在检测性能和频率分辨率之间的权衡。最佳分段大小取决于信噪比，这使得实际的最佳分段大小设置变得困难。针对这一问题，我们以前提出了一种利用噪声底限(NF)估计输出与没有信噪比信息的分段大小之间的关系的分段大小选择方法。由于它使用穷举搜索来选择合适的分段大小，因此可以以相对较高的计算复杂度为代价来实现准确的频谱感知。在本文中，我们提出了一种分段大小的选择方法，由于使用了有限的搜索，因此该方法在提供合理的频谱感知性能的同时具有较低的计算复杂度。仿真结果表明，该方法的频谱感知性能与传统方法相当，复杂度降低70%或更少。
We introduce a Welch FFT segment size selection method for FFT-based wide band spectrum measurement in the context of smart spectrum access (SSA), in which statistical spectrum usage information of primary users (PUs), such as duty cycle (DC), will be exploited by secondary users (SUs). Energy detectors (EDs) based on Welch FFT can detect the presence of PU signals in a broadband environment efficiently, and DC can be estimated properly if a Welch FFT segment size is set suitably. There is a trade-off between detection performance and frequency resolution in terms of the Welch FFT segment size. The optimum segment size depends on signal-to-noise ratio (SNR) which makes practical and optimum segment size setting difficult. For this issue, we previously proposed a segment size selection method employing a relationship between noise floor (NF) estimation output and the segment size without SNR information. It can achieve accurate spectrum awareness at the expense of relatively high computational complexity since it employs exhaustive search to select a proper segment size. In this paper, we propose a segment size selection method that offers reasonable spectrum awareness performance with low computational complexity since limited search is used. Numerical evaluations show that the proposed method can match the spectrum awareness performance of the conventional method with 70% lower complexity or less.