Abstract: ‍ ‍As an effective anti-jamming approach， frequency-hopping （FH） technology has been widely applied to tactical communication system to improve communication network's throughput under strong confrontation environment. Key challenges for tactical wireless communication network face are the follower jammer with responsive spectrum reconnaissance and frequency agility. In response to improving the anti-fragile performance and throughput of tactical FH communication system， this paper investigates a double deep Q-network （DDQN） based anti-jamming scheme. The interactions between a radio transmitter and a follower jammer are formulated as a Markov decision process （MDP）， in which the jammer adjusts to improve the interference performance， and the radio terminal feeds the state feedback information into algorithm input parameters， then selects the transmit power and hopping rate according to the output of policy network. With the aim of maximizing the system throughput the proposed approach updates network parameters through interacting with environment simultaneously without knowing the environment state and jamming model， and converges to an optimal solution with joint power and hopping rate control. Simulations results demonstrate that the proposed scheme can provide better effective interference resistance in follower jamming environment， comparing with other model free learning methods in contrast.