Research On Wideband Spectrum Sensing Technology And Its FPGA Implementation

Electromagnetic countermeasures in modern warfare are becoming increasingly fierce,and communication systems are often subject to malicious jamming attacks from the enemy.Based on the cognition of the electromagnetic spectrum environment,the intelligent anti-jamming communication system adaptively adjusts the transmission waveform according to the electromagnetic environment,and intelligently selects the best anti-jamming mode.With the increasing bandwidth of communication systems,wideband spectrum recognition technology has become the foundation of intelligent anti-jamming communication systems.This thesis focuses on the research of wideband spectrum cognition technology,including wideband jamming detection,jamming recognition,and jamming parameter estimation.The FPGA implementation of the wideband spectrum cognition system is completed,and the main research contents include:Firstly,for the wideband jamming detection problem,a multi-subband joint jamming detection algorithm based on Forward Consecutive Mean Excision（FCME）is proposed.The thesis first divides the wideband spectrum into multiple subbands and preliminarily determines whether there is jamming in the subbands based on the relative size of subband power.Then,the common initial threshold is calculated using the subband with the minimum power,and the jamming frequency point is detected using the FCME algorithm.The leakage problem of single subband jamming detection is solved while effectively reducing the complexity of wideband jamming detection.Simulation results show that when the Jamming to Noise Ratio（JNR）is greater than 2d B,the probability of missing detection is less than10^-2,and when the JNR is greater than 8d B,the probability of missing detection is less than10^-5,and the probability of false alarm is not more than10^-2.Secondly,in response to the jamming recognition problem,a convolutional neural network（CNN）-based jamming recognition algorithm is studied,and a jamming recognition scheme based on reconstructed jamming spectrum is designed.In this thesis,the reconstructed jamming signal spectrum is used to construct four-channel data of the real part,imaginary part,frequency domain amplitude spectrum,and Hadamard product of time-domain data.The CNN automatically extracts jamming features and classifies them.Simulation results show that for the silent period sample data,when JNR is greater than-4 d B,the recognition accuracy of the six typical suppression-type jamming is greater than 99%;for the non-silent period sample data,when JNR is greater than-2 d B,the recognition accuracy of the six typical suppression-type jamming is greater than 99%.Thirdly,for the jamming parameter estimation problem,a method for estimating jamming location based on descending FCME is proposed.Unlike traditional FCME,this thesis first sorts the spectrum to be detected in descending order,and then takes the frequency points with the maximum amplitude as the initial jamming points.Finally,the jamming frequency point detection is completed using the idea of FCME iteration,effectively solving the problem of inaccurate jamming location estimation when jamming and DSSS signals coexist during the non-silence period.Simulation results show that for the silence period sample data,when the JNR is greater than 2d B,the NRMSE of the estimated center frequency and bandwidth of jamming are in the order of10^-2 and10^-3,respectively,and for the non-silence period sample data,when the JNR is greater than6d B,the NRMSE of the estimated center frequency and bandwidth of jamming are in the order of10^-2 and10^-3,respectively.The fourth part of the study implemented the circuit of the wideband spectrum cognition system on the hardware platform based on XCVU11P FPGA chip,and conducted system testing.Specific FPGA implementation schemes were provided for the key modules of the wideband spectrum cognition system,and the functionality of the modules was verified through ModelSim simulation.Performance testing was then conducted on the hardware platform,and the results showed that the error of jamming recognition accuracy between the circuit test and the simulation result did not exceed 2%,and the error of NRMSE between the circuit test and the simulation result was not more than10^-1when JNR=2d B.Finally,the wideband spectrum cognition system was embedded in an intelligent anti-jamming communication system and subjected to field testing.The results showed that for 18 different jamming scenarios,the correct cognition probability of the wideband spectrum cognition system was 100%,and the processing delay did not exceed 4.2ms,which met the requirements of the system.