A dynamic spectrum intelligent jamming algorithm based on deep reinforcement learning
With the advancement of artificial intelligence technology,reinforcement learning has shown great potential in enhancing electromagnetic spectrum control and jamming efficiency.Given the robust anti-jamming capability of frequency-hopping communication systems and the inadequacy of traditional jamming methods,this paper intends to utilize deep reinforcement learning(DRL)for intelligent electromagnetic jamming in dynamic spectrum environments.First,a partially observable Markov decision process(POMDP)is introduced to model the communication counteraction process between jammers and frequency-hopping communication users.Second,a jamming decision network capable of mining spectrum features and performing memory backtracking is designed,based on convolutional neural networks(CNNs)and long short-term memory networks(LSTMs).This network implements a dynamic spectrum intelligent jamming(DSIJ)algorithm grounded in deep reinforcement learning.Simulation results indicate that compared to the traditional deep Q network(DQN)algorithm,the proposed DSIJ algorithm increases the jamming success rate by approximately 18％;compared to traditional sweeping jamming methods,the success rate is further increased by about 68％.These demonstrate that the proposed algorithm holds effectiveness and significant advantages in implementing intelligent jamming strategies in dynamic spectrum environments.
deep reinforcement learning(DRL)frequency-hopping communicationintelligent jamming decisionpartially observable Markov decision processes(POMDP)
国家科技期刊平台
NSTL
万方数据
