Research On Interference Perception And Recognition Technology Of UAV Measurement And Control System

Complex electromagnetic environment has become an important factor to restricting the reliable and effective communication of Unmanned Aerial Vehicle(UAV)measurement and control system.Jamming cognition is to realize the perception and recognition of the electromagnetic environment by sensing the surrounding electromagnetic environment,the use of jamming detection,jamming identification,jamming parameter estimation and other technical means to achieve the electromagnetic environment perception and recognition,it provides the basis for the cognitive decision and anti-jamming transmission of the communication system so that the system able to continuously adjust the transmission parameters and decision parameters to achieve the best transmission results.Jamming cognition has become an important guarantee for UAV work and efficient transmission of data.In this paper,we focus on the common types of Jamming in UAV measurement and control links,from the aspects of jamming cognitive strategy development,jamming detection,jamming multi-dimensional feature parameter extraction and jamming classification algorithm to study the jamming detection and jamming classification technology of UAV measurement and control system,The main contents are as follows:The first part mainly introduces the research background and significance of this paper,and elaborates the basic situation of UAV measurement and control links and the current jamming perception and classification technology.The second part mainly studies the cognitive strategy and jamming cognitive process of UAV measurement and control link.It includes dividing the common jamming of UAV measurement and control links into three types: time domain,frequency domain and non-stationary,giving the four factors that need to be considered in the development of jamming cognitive strategy,establishing the jamming cognitive strategy of UAV measurement and control link and giving the jamming cognitive process.The third part mainly studies the time domain jamming detection algorithm and the frequency domain jamming detection algorithm.For the time domain jamming,the time domain packet detection algorithm is mainly studied,the threshold factor of packet detection is deduced,the algorithm realization flow is analyzed,on this basis,the periodic and duty cycle estimation algorithms for impulsive noise jamming are further studied,for non-silent period of frequency domain jamming detection,the improved FCME detection algorithm based on Welch spectrum is studied,the simulation results show that the time domain packet jamming detection algorithm and the pulse jamming parameter estimation algorithm have better performance,the improved FCME algorithm based on Welch spectrum is superior to the traditional FCME algorithm,having better detection performance,and the probability of false detection is relatively low,it is suitable for non-silent period jamming detection.The fourth part mainly studies the extraction algorithm of multi-dimensional jamming characteristic parameters from several dimensions of time domain,frequency domain and fractal domain.The characteristic parameters include Time Domain Peak To Average Ratio(TDPAR),Time Domain Kurtosis(TDK),Jamming Bandwidth Factor(JBF),Frequency domain Peak To Average Ratio(FDPAR),Normalized Spectrum Standard Deviation(NSSD),Box Dimension(BD),LZ Complexity,Time Frequency Position Standard Deviation,(TFPSD).The time-domain characteristic parameter values of impulse noise jamming are analyzed theoretically.In order to apply to non-stationary jamming,we propose a method to extract the LZ complex by using the short time Fourier transform to extract the time-frequency binary graph,and extract the standard deviation characteristic parameters of the jamming frequency,Finally,all the jamming characteristic parameters are simulated and analyzed.Simulation results show: time domain characteristic parameters have better ability to distinguish impulse jamming,the jamming bandwidth factor and the frequency domain peak ratio are better than the aiming jamming,partial standard deviation of normalized spectral impulses has a good ability to distinguish between multi-tone jamming,Fractal dimensions are better robust at JSNR> 10 dB and have better ability to distinguish between partial jamming types.The fifth part mainly studies the decision tree(DT)and the support vector machine(SVM)applying to jamming classification.Based on the multi-dimensional characteristic parameters,the decision tree and the non-silence decision tree are constructed,a set of support vector machines is developed to identify the structure of multiple jamming types,simulation results show that the jamming recognition performance is different under different classification algorithms,in the case of non-silence,the performance of support vector machine is better than that of decision tree,when JSNR> 10 dB,the use of decision tree or support vector machine in the silence period and non-silence period is almost 98% of the recognition performance,all jamming have better performance.