Research On Jamming Decision Based On Partially Observable Markov Decision Processes

With the continuous development of radar technology,radar plays an increasingly important role in modern electronic warfare.In order to ensure targets can not detected by the radar,the enemy radar must be interfered and reduced the probability of finding targets.Usually,there is a dynamic game between the radar and the jammer.In a complex electromagnetic environment,radar and jammer are non-cooperative parties.It is very difficult for the jammer to obtain the radar strategy accurately.Therefore,the traditional jamming decision technologies are not good to interfere with the radar.In this paper,a jamming decision technology based on Partially Observable Markov Decision Processes(POMDP)is proposed.This method can ensure that the jamming can still get good decision result under the premise that the enemy radar strategy cannot be obtained.This paper mainly studies the method from the following aspects.The jamming decision technology has been analyzed in detail firstly,and the basic principle and working characteristics of the jamming decision technology are discussed.The jamming principle of suppressive interference and deceptive interference is expounded.And the feature of the time-frequency domain of several typical jamming modes is simulated and analyzed.Then the intelligent jamming decision technology and the jamming decision technology based on game theory are studied.The jamming decision technology based on game theory is simulated and analyzed.And the jamming decision result is given and the problem of existing jamming decision technology is pointed out.Then POMDP was studied.POMDP is an extension of Markov Decision Processes(MDP).In the article,the MDP is modeled and analyzed,then the performance criteria,decision strategy and belief state meaning of the decision process are redefined according to the characteristics of POMDP.The basic model of POMDP is established and several methods of solving POMDP are studied.Finally,POMDP is applied to the jamming decision technology,this paper discusses the meaning of state space,action space,transition probability function,observation space,observation probability function and real-time benefit module in the jamming decision technology based on POMDP,and establish the model of the jamming decision technology based on POMDP.The model is solved and the results of the jamming decision technology based on POMDP are analyzed.The simulation results are compared with the jamming decision technology based on observation parameters and the jamming decision technology based on game theory.The simulation results show that the results of the jamming decision technology based on POMDP are better than those of the other two jamming decision technologies,and the method can still obtain good jamming decision results when only partial radar observation parameters can be obtained,which effectively solved the problem that when the jammer and the radar are non-cooperative parties,the jammer cannot effectively interfere with the radar.