基于最大特征值的拟合优度检验频谱感知算法

发布时间：2019-03-01 12:29

【摘要】：如今频谱资源稀缺问题已成为热点问题之一,认知无线电技术是解决该问题的一种重要手段。基于拟合优度(Go F)检验的频谱感知算法是一种优秀的感知算法,它不需要知道任何主用户的信息,而且能够在较少的采样点条件下达到较优的检测性能。已有的基于拟合优度检验的频谱感知算法虽然能够在静态信号下表现出优异的性能,但是在检测动态信号时性能急剧下降。针对这个问题,本文提出了一种基于最大特征值的拟合优度检测算法利用随机矩阵理论分析了信号协方差矩阵的最大特征值分布,通过Go F检验来感知主用户的存在性,在检测动态信号时仍能保持优异的检测性能。此外,在所提算法中,设计了一种低复杂度的拟合准则,它能够降低Go F算法拟合统计量的计算复杂度,并提高算法检测性能。仿真结果表明了所提检测算法和拟合准则的有效性。
[Abstract]:Nowadays, the scarcity of spectrum resources has become one of the hot issues. Cognitive radio technology is an important means to solve this problem. The spectrum sensing algorithm based on goodness-of-fit (Go F) test is an excellent sensing algorithm. It does not need to know the information of any master user, and it can achieve better detection performance under the condition of fewer sampling points. Although the existing spectral sensing algorithms based on goodness-of-fit test can perform excellent performance under static signals, the performance of detection of dynamic signals decreases dramatically. In order to solve this problem, a goodness-of-fit detection algorithm based on maximum eigenvalues is proposed in this paper. The maximum eigenvalue distribution of signal covariance matrix is analyzed by using random matrix theory, and the existence of master users is perceived by Go F test. In the detection of dynamic signals can still maintain excellent detection performance. In addition, in the proposed algorithm, a low complexity fitting criterion is designed, which can reduce the computational complexity of the fitting statistics of Go-F algorithm and improve the detection performance of the algorithm. The simulation results show that the proposed detection algorithm and fitting criterion are effective.
【作者单位】：大连理工大学信息与通信工程学院;
【分类号】：TN925

【相似文献】