摘要
针对多跳频信号参数估计问题,利用跳频信号频率在时频域上的稀疏性,在块稀疏贝叶斯学习（b SBL）的基础上,提出了一种T-SBL稀疏学习算法,通过重构后信号的时频图完成跳频信号的跳周期、中心时刻、跳时刻等参数的估计.首先将接收信号进行重叠分割得到观测矩阵;然后根据跳频信号时频域的稀疏性建立信号的多测量（MMV）稀疏模型,在块稀疏贝叶斯学习算法的基础上将多测量模型转化为单测量（SMV）模型;最后通过重构后信号的清晰时频图进行参数的快速估计.为了进一步提高低信噪比下的估计性能,采用形态学滤波的方法对获得的时频图进行修正.理论分析和仿真实验表明了该算法的有效性和良好的估计性能.
Aiming at the multi-frequency hopping signal parameter estimation, a T-SBL sparse learning algorithm was proposed based on the block sparse Bayesian learning（b SBL） by using the sparsity of frequency in time-frequency domain.The hop period,central time and start hopping time were estimated through the time-frequency spectrum of the reconstructed signal. Firstly, the observation matrix was obtained by segmenting the received signals into overlapped segments. Secondly, the sparse model of multiple measurement vectors（MMV） model was established according to the sparsity of frequency hopping signals in time-frequency domain,and then the MMV model was transformed into the single measurement vector（SMV） model based on the b SBL algorithm. Lastly, the fast estimation of parameters was made by the clear time-frequency spectrum of the reconstructed signal. At the same time, the time-frequency spectrum was amended via morphological filtering method in order to enhance performance of low signal-to-noise ratio（SNR） algorithm.Theoretical analysis and simulation results show that this algorithm has good effectiveness and estimation performance.
作者
郭英
于欣永
张坤峰
李雷
Guo Ying1,2 Yu Xinyong1 Zhang Kunfeng1 Li Lei1(1 Institute of Information and Navigation, Air Force Engineering University, Xi＇an 710077, China; 2 Science and Technology on Information Transmission and Dissemination in Communication Networks Laboratory, Shijiazhuang050081, Chin)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2018年第2期95-99,共5页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(61601500)
关键词
跳频信号
块稀疏贝叶斯
时频图
重叠分割
形态学滤波
frequency-hopping signal
block sparse Bayesian learning （bSBL）
time-frequency spectrum
overlapped segments
morphological filtering
