基于YOLOX的跳频信号检测

现有跳频信号检测算法大多针对固定参数的跳频信号检测，而面对可变参数的跳频信号检测时，其每跳时长及带宽的不确定性导致这些算法的适用性下降。为此，提出了一种基于YOLOX的跳频信号检测算法。该算法适用于固定参数跳频信号和可变参数跳频信号，判断跳频信号的存在性。首先将观测时间内的信号进行短时傅里叶变换获得灰度时频谱图；然后将时频谱图经过目标检测网络YOLOX,获得各个信号的预测框；最后根据跳频信号各跳点的时间连续性，筛选出跳频信号各个跳点，根据连续跳点个数判断跳频信号的存在性。对算法的检测流程进行了仿真，以验证算法的可行性。实验结果表明，该算法可以较好地完成盲检测任务，且能够通过增加较少的先验信息以提高检测性能。

Frequency hopping signal detection based on YOLOX

Most of the existing frequency hopping(FH) signal detection algorithms are suitable for fixed parameter FH signal detection.For detecting variable parameter frequency hopping signals,the unpredictability of both the length and bandwidth of each hop can significantly reduce the effectiveness of these algorithms.Therefore,the anthors present a YOLOX-based detection algorithm for FH signal,which is suitable for fixed parameter FH signal and variable parameter FH signal to judge the existence of signal.First,the signal is processed by short-time Fourier transform to obtain the gray-scale time-frequency spectrum.Then the spectrum is passed through the target detection network YOLOX to get the prediction box of each hop.Finally,according to the continuity among the hops,these hops are selected,and the existence of the signal is judged by the number of consecutive hops.The algorithm is simulated to verify its feasibility.The experiment results show that this algorithm can complete the blind detection task well and improve the detection performance by adding less prior information.