Abstract: DoS attacks signal has non-stationary and time-varying property.It is lost in the complex network environment with color noise background,and it is difficult to detect.Traditional methods use Hough transform impulse response method to detect the non-stationary signal.Due to the edge effect of frequency distribution,detection performance is not good.A new spectrum detection method of DoS attack signal was proposed based on envelope extension and intrinsic wave matching filtering.The DoS attack signal is processed with hyperbolic frequency modulated signal decomposition,and mathematical evolution model is constructed.Signal envelope intrinsic wave features are extracted.The bilinear Hough transform method is used to analyze the spectrum distortion,instantaneous frequency estimation is obtained,and single pulse response amplitude frequency response is calculated.In time frequency feature space,the envelope extension path search is optimized.Intrinsic wave matching filter is designed based on minimum mean square error criteria.DoS frequency shift is controlled,and the spectrum detection is obtained.Simulation results show that the algorithm can improve the detection performance,and the interference of strong colored noise can be suppressed.The detection probability is higher than traditional methods.It can accurately estimate the parameters,and the active defense ability of network security is improved.

Key words: Envelope extension,Intrinsic wave matching,DoS attack,Signal detection,Network security