Research On Frequency Hopping Signal Sorting And Location Technology Based On Constrained Clustering Algorithm

Frequency-hopping communication,as a typical multi-access technology,is widely used in commercial and military fields due to its characteristics of anti-interference,antifading,low interception probability and security.Especially for the military field,as a third party to intercept and track the target frequency-hopping signals,the first task is to sort out the frequency-hopping signals of different network stations in the frequencyhopping network,and then conduct positioning according to the sorting results to achieve target identification and tracking.So frequency hopping signal sorting and location technology is of great significance in the military field.On the basis of frequency-hopping network type and parameter feature extraction technology,the sorting and location of multiple frequency-hopping signals are carried out.Therefore,this thesis also focuses on these two aspects,and the specific research content is as follows:This thesis first studies the principle of frequency hopping system and its networking theory,because the frequency hopping signal is a non-stationary signal,so in order to extract its characteristic parameters,it is necessary to carry out the time-frequency analysis of the frequency hopping signal.This thesis mainly studies different time-frequency analysis methods,including short-time Fourier transform,Wegener distribution and its derivative methods.Considering the time-frequency resolution and calculation amount,the short-time Fourier transform method is selected for time-frequency analysis.Then,this thesis considers that frequency-hopping signals will encounter a lot of noise and interference signals in the actual electromagnetic environment.In order to eliminate the impact of noise,this thesis studies different image segmentation algorithms including edge detection method,watershed detection method and threshold segmentation method by means of morphological processing.After considering the characteristics of different algorithms,the threshold detection method which can still maintain a high accuracy of parameter estimation under the condition of low signal-to-noise ratio is selected as the main segmentation algorithm.In view of the interference signal in the actual frequency hopping signal reception,the extraction and parameter estimation of frequency hopping signal are affected.The effect of common interference signals on time-frequency graphs is removed by erosion expansion operation and frequency-hopping signals are extracted successfully.The simulation results show that this method can maintain high accuracy of parameter estimation.Finally,this thesis studies the sorting and positioning work on the basis of the above,selects the K-Means algorithm with simple calculation and good effect for the clustering operation in the aspect of frequency-hopping signal sorting,and adds the initial point selection distance maximization and constraint pair on the basis of the original algorithm to improve the accuracy of sorting,and proposes a new emission source number estimation method and type judgment method.The experiment shows that COP-Kmeans algorithm has higher sorting accuracy than the original algorithm,even in the case of large parameter estimation error,the algorithm can still maintain good clustering effect.In the time difference positioning scene,the signal matching and positioning of multi-target frequencyhopping network signal is proposed by using time-frequency analysis and signal sorting.In the direction finding positioning scene,the false points in the multi-target scene are eliminated by using the datum line minimum distance method and the correct target position estimation is obtained.The simulation results show that good results are obtained.