Research On Multifunctional Software Defined Radar Emitter Identification Technology

In recent years,with the development of radar technology,multifunctional software defined radar comes into being.Multifunctional software defined radar has the characteristics of definable requirements,reconfigurable software and reconfigurable hardware.In the future,multifunctional software defined radar will pose a great challenge to the existing radar signal and radiation source analysis.Aiming at the difficult problem of multifunctional software defined radar emitter analysis,this paper proposes a new method of multifunctional radar source cognition based on deep learning,and divides the cognition of multifunctional software defined radar emitter into two aspects: functional recognition and individual recognition.The main research contents of this paper mainly include three aspects: radar emitter signals and rf power amplifier modeling,radar emitter signal functional identification based on deep learning and radar emitter individual identification based on deep learning.Firstly,this paper simulates radar signal based on several common modulation methods.At the same time,according to the characteristics of multifunctional software defined radar,the nonlinear characteristics of rf power amplifier are taken as the basis for individual identification of radar radiation source?This paper completed the modeling and simulation of five known and one unknown multifunctional software defined radar emitter.In the functional identification of radar emitter signals,three methods are proposed,namely,autocoder network,convolutional neural network and support vector machine,and their recognition rates in the test set are analyzed and compared.The results show that the autocoder network has the highest recognition rate under 15 d B SNR,and the recognition condition is also satisfactory under low SNR.The network structure is simple and the training time is short,so it is the most suitable method for radar emitter signal functional recognition.Three methods of autocoder network,convolutional neural network and support vector machine are still used for individual identification of radar emitter,and the recognition rates of the three methods in the test set are analyzed and compared.The results show that the convolutional neural network has the highest recognition rate for the test set under 15 d B SNR and is the most suitable method for individual identification of radar emitter.Finally,in the case that the autocoder network is used for the functional identification of radar emitter signals and the convolutional neural network is used for the individual identification of radar emitter signals,the individual identification is combined with the functional identification,so as to complete the ecognition of unknown radar emitter.The results show that after the intercepted enemy multifunctional software-based radar emitter signals are input into the trained deep learning network,wo can not only understand its function,but also analyze that it belongs to a known multifunctional software defined radar emitter or an unknown radar radiation source,so as to complete the cognition of multifunctional software defined radar emitter.