Intelligent Cognitive Anti-Jamming Algorithm Based on Long Short-Term Memory Network

Aiming at the weak anti-intelligent jamming ability of maritime terahertz communication network in complicated communication environment, an intelligent cognitive anti-jamming algorithm is proposed based on the original asynchronous advantage actor-critic (A3C) algorithm and long short-term memory (LSTM). The terahertz communication ship is regard as an agent, and the environment spectrum state is used as the prior information of A3C algorithm. LSTM neural network is used to reduce the training time and improve convergence speed. Taking advantage of both A3C algorithm and LSTM network, the proposed algorithm can realize rapid and effective anti-intelligent jamming channel selection. Simulation shows the effectiveness and practicality of the proposed algorithm. Compared with deep Q learning (DQN) algorithm and original A3C algorithm, our algorithm can realize higher cumulative throughput and channel decision-making success rate.