Optimal Jamming Frequency Selection for Cognitive Jammer based on Reinforcement Learning

In this paper, a cognitive jammer is developed which can adaptively and optimally jam the radar and protect the target from being detected. The interaction of the cognitive jammer and the environment is modeled as a finite Markov Decision Process based on the framework of reinforcement learning. Q-learning algorithm is used to solve the optimal jamming frequency selection problem. After several interactions, the jammer can learn the radar's strategy and optimize its jamming frequency to achieve a larger jamming-plus-noise to signal ratio (JNSR) during the whole process. Numerical results are given to illustrate the effectiveness of the proposed method. Compared with a random jamming frequency selection method, the JNSR of the proposed method is significantly larger.