Research On Multi-interference Detection And Recognition Method Based On Feature Extraction

Due to the increasingly complex electromagnetic environment and the increasing communication density,there are often multiple interference signals in communication system.It is necessary to detect and recognize the interference signals,and then take corresponding anti-interference measure to ensure the security and quality of communication.Thesis aims to study the multi-interference detection and recognition method based on feature extraction.Aiming at the problem of detection and recognition of multi-interference signals studied in thesis,the Interference signal analysis system is first introduced to locate the position of the work in the system,and then the received signals and single-tone interference,multi-tone interference,linear frequency sweeping interference,pulse interference,narrow band interference and multi-interference signal models are established,and simulate them simultaneously.Based on the simulated waveform characteristics of the interference signals,the features of the interference signals in time,frequency and time-frequency domains are extracted,including time-domain moment skewness,time-domain envelope fluctuation,high-order cumulant features,carrier factor,spectrum kurtosis,frequency domain moment kurtosis,0.5 times bandwidth,frequency domain moment skewness,frequency domain envelope fluctuation and fractional fourier domain maximum value,and simulate the change of each feature with the jamming signal ratio.It is the basis for the detection and classification of interference signals.For STI,MTI,LFSI,PI and NBI,the interference detection and classification algorithms based on decision tree and support vector machine are studied.Then we simulate and compare their performance.The detection and recognition accuracy rate is high,and the performance of the interference classification algorithm based on support vector machine is 6d B better than that of the decision tree algorithm.In the end,two algorithms are used to detect and identify multiple interference,and the recognition performances of one interference,two interference stacking and three interference stacking are simulated and compared.Simulation results show that the more jamming signals there are,the more serious the deterioration of the identification performance of the two algorithms.Therefore,it is necessary to study the detection and recognition algorithm for multiple interferences.For the multi-interference signals formed by the simultaneous existence of four kinds of STI,MTI,LFSI,PI and NBI,the multi-interference detection and recognition algorithms based on decision tree and support vector machine are respectively studied,and simulation comparisons are made.Through the performance analysis of two algorithms,the multi-interference detection and recognition algorithm based on support vector machine has 5d B performance improvement over that based on decision tree.Finally,the simulation analysis of the detection and identification performance of the two algorithms for one interference and the multi-interference signals formed by twointerference superposition and three-interference superposition respectively shows that the multi-interference detection and recognition algorithm is still applicable to one interference,and the less the number of interference signals.It indicates that the multiinterference detection and recognition algorithm studied in thesis is a generalization of a interference detection and recognition algorithm,which has universal meaning.In addition,the algorithms of thesis and other literatures are compared and analyzed.