Collaborative Anti-jamming Algorithm Based on Q-learning in Wireless Communication Network

Abstract

Aiming at defending against the malicious jamming attacks and considering the interference among users in the multi-user wireless networks, a collaborative anti-jamming algorithm based on Q-learning in wireless communication network (CAAQ) is proposed in this paper. Specifically, since there exists the competition and collaboration among the users, the metric is first applied to determine whether there has interference among users by adding the distance threshold, which can significantly decrease both the training time and the complexity of multi-agent Reinforcement Learning (RL). Then, through the user-to-user collaboration at the information interaction level, a collaborative anti-jamming algorithm based on Q-learning is proposed to optimize the spectrum allocation for all users. Numerical results verify the superiority and substantive of the proposed CAAQ, which can simultaneously avoid the interference among the users and overcome the malicious jamming attack.