Anti-Jamming System Based on Deep Reinforcement Learning in Dynamic Environment

This paper proposes a DQN algorithm for dynamic environments, which evaluates the performance of the current intelligent body in real time, adjusts the “exploration-utilization” strategy of the intelligent body according to the direction of change, accomplishes the anti-interference function in dynamic interference environments, and shows certain advantages compared with the traditional algorithm. We abstract this scenario into a conventional communication system and use an improved deep Q-network (DQN) algorithm to accomplish the establishment of an anti-jamming system. Based on the ε-greedy strategy and Bolzmann exploration, the algorithm adaptively balances the “exploration-exploitation” threshold by evaluating node performance metrics.