Detection Of LFM Signals In Impulse Noise

A linear frequency modulation(LFM) signal, as the typical non-stationary signal, is widely used in radar, communication, sonar, seismic prospecting and biology. In the field of electronic countermeasures, to capture the radar signals of the enemy and to analyze the information obtained is an important mission for electron reconnaissance. Therefore, it is very important to conduct research work on LFM signal detection. Gaussian distribution is often used as the background noise model in traditional analysis, however, actual noise, such as lightning, sea clutter and low-frequency atmospheric noise, tends to be more complex and has impulses. Alpha stable noise can model the actual noise better. To address the problem that the traditional analysis methods degrade severely in Alpha stable noise environment, the following robust methods which can effectively suppress the impulse noise based on the traditional time-frequency analysis are proposed:LFM signal analysis method based on L-estimation is proposed. This kind of methods can effectively suppress the impulse noise by the weighted function. The L-power weighted method is used to filter the noise out in the frequency domain, and the power function of 2 is used as the weighted function to suppress the impulse noise. The optimal L-Cauchy weighted method based on the Cauchy distribution, which is the only Alpha stable distribution that has closed-form probability density function when a(27)2, and can match not only the impulsive but also the heavy-tailed characteristics, is selected as the weighted function due to its effectiveness in suppress the impulse noise. The simulation results show that the two methods proposed in this paper have good robustness.LFM signal analysis method based on the M estimation is proposed. The robust iterative method of LFM signal is established based on the M estimation and re-described as an optimal problem, in which the optimal filtering result is reached by iteration. The mean filter and the linear weighted filter based on robust iterative analysis method are proposed. In addition, an LFM signal analysis method based on the extending of the Cauchy distribution tail parameter is proposed, which can improve the detection performance and broaden the application scope. Finally, a method based on the cost function optimization is proposed. A new tailing parameter is introduced, a cost function being able to suppress the noise is constructed, and a unified structure of robust weighted filtering is derived in this method, within the unifying framework, the weighted Myriad, the weighted Merid and the generalized Cauchy distribution based weighted filtering methods are interpreted. The simulation results show that the proposed methods can effectively suppress the Alpha stable distribution noise, and it is robust to the impulse noise.