Adaptive Jamming Stragety Based On Reinforcement Learning

With the rapid development of Artifical Intelligence technology,the new generation electronic information equipment with cognitive ability has also improving.In the field of electronic countermeasures,facing the gradually intelligent target system,traditional electronic countermeasures equipment not effective,needing to develop cognitive electronic countermeasures based on Artifical Intelligence technology,adaptive jamming stragety is one of the core of this.Advanced target system usually have multiple working states,eg.digital phased array radar.Adaptive jamming stragety allows the jamming party to form an optimal jamming stragety against the different states of the target system to adapt to the complex electromagnetic environment of modern electronic countermeasures.Adaptive jamming stragety establish the relationship between jamming resourses and target status through behavioral learning ability,learning conclusion can support the construction of dynamic expert knowledge base,it also can be used as the basis for scheduling intelligent jamming resourses.This paper proposes adaptive jamming stragety technology based on reinforcement learning.First,to build a relationship between reinforcement learning and adaptive jamming stragety,completing the work of system modeling.Then,introduce the principles of typical algorithm of reinforcement learning,and simulate its process,and add the mechanism of eligibility trace based on the traditional algorithm to improve the learning efficiency of the algorithm.In order to solve the limitation of reinforcement learning only applicable to finite state space,combining deep learning with reinforcement learning,applying value function approximation by deep reinforcement learning,so that reinforcement learning can be extended to high dimensional or even continuous state space,and present the workflow of adaptive jamming stragety based on deep reinforcement learning.Finally,simulate the process of electronic countermeasures in an intelligent jamming software,validating the correctness and effectiveness of adaptive jamming stragety.