Defend Against Jamming Attacks Using Deep Reinforcement Learning

Cognitive radio network has the learning ability to adjust transmission behavior to adapt to dynamic electromagnetic environment, and it has become a research hotspot in the field of the new generation of intelligent wireless communication systems. However, recent research results show that cognitive radio networks are vulnerable to interference attacks, resulting in a serious decline in spectrum utilization. In this paper, an intelligent anti-jamming scheme based on deep reinforcement learning is proposed, and the anti-jamming decision of the communication party is modeled as a Markov decision chain. This method can help the communication transmitter to learn a combination of communication parameters that can maximize the probability of communication success. In the specific implementation, we use Double DQN as the communication decision-making brain, and specially designed a convolutional neural network suitable for the communication spectrum as input to accurately estimate the action value function. The simulation results show that our method has better performance than the traditional anti-jamming method based on reinforcement learning.