Joint Channel, Power and Bandwidth Optimization for Anti-jamming Communications: A Multi-agent Q-learning Approach

Abstract

To facilitate external malicious jamming and interference in multi-user anti-jamming communication networks, in this paper, a novel joint channel, power and bandwidth optimization model is constructed. Channel reconstruction technology and sleep mechanism are introduced to avoid channel access conflict and jamming. Besides, joint probability distribution is adopted to characterize incomplete channel state information (CSI) caused by complex electromagnetic environment. Based on the system model, a multi-agent Q-learning approach based on joint optimization anti-jamming is adopted. Simulations show that the proposed algorithm can achieve better communication utility and maximum communication transmission rate of the whole network than the comparison algorithms.