This paper investigates the multi-user power control problem in relay-assisted anti-jamming systems. Because of the hierarchical confrontation characteristics between users and jammer, we take the incomplete information and observation error into consideration and formulate an anti-jamming Bayesian three-layer Stackelberg game, in which primary users act as leaders, relay users act as vice-leaders, and jammer acts as follower. Both users and jammer have the ability to sense others’ transmission power and choose optimal power to realize the maximum of utility. Based on the backward induction method, we propose a multi-user hierarchical iterative algorithm (MHIA) to obtain the Stackelberg Equilibrium (SE), and prove the existence and uniqueness of SE. Finally, simulation results are compared to Nash Equilibrium (NE) to verify the effectiveness of the proposed game. Moreover, both the influence of incomplete information and observation error on utility are analysed.

In this article, we study the problem of anti-jamming through channel selection in fading environment. Different from the most existing works which ignored the fading characteristic of the channels, we use a Markov channel model to capture the influence of variable channel transmission rate in fading environment. Then, we model the anti-jamming channel selection problem as a Markov decision process. On this basis, an online reinforcement learning algorithm (Q-learning) is proposed to select the optimal channel intelligently. Simulation results show that the proposed algorithm outperforms the sensing algorithm. The reason is that it can learn both the pattern of jamming signal and the condition of channel, so as to select the channel with better condition for data transmission. Based on the proposed algorithm, we develop a real-life dynamic spectrum anti-jamming testbed based on the USRP platform in the indoor environment to demonstrate the effectiveness of the algorithm.

In this letter, we study the problem of jamming pattern recognition (JPR) when some patterns have no training samples. Existing works were designed for recognition of known jamming patterns only, hence they failed to deal with unknown jamming patterns. To address this challenge, a zero-shot learning (ZSL) based JPR scheme is proposed. Firstly, a supervised training process is introduced to learn the latent feature representation of known jamming patterns. Then, an unsupervised classification approach is proposed to recognize different jamming patterns. Finally, both known and unknown jamming patterns are classified in latent feature space. Simulation results show that the proposed scheme achieves excellent performance when dealing with the open-set JPR tasks.

This letter investigates the problem of combating the intelligent reactive jammer through a deep reinforcement learning (DRL) based hidden strategy. Existing anti-jamming researches always assume a capacity-limited adversary which is non-reactive or only launches attacks while detecting activities of the transmitter. In this letter, we consider an intelligent reactive jammer that releases track jamming once activities of the transmitter are detected, otherwise it launches indiscriminate jamming (e.g., sweep or comb jamming). A DRL-based hidden strategy is proposed to resist this powerful jamming via adjusting power and accessing idle channels simultaneously. Moreover, the parallel policy network and individual reward function are introduced to enhance the anti-jamming performance. Simulation results confirm that the proposed algorithm can significantly elude the jammer’s detection and achieve dynamic spectrum access in a dynamic and unknown environment.

In this letter, we investigate an intelligent jamming problem in a wireless confrontation scenario where both the user and the jammer can dynamically adjust spectrum access or the jamming pattern. To realize accurate jamming attacks, we first propose a deep learning-based pattern recognition method to recognize the user’s spectrum access patterns. Then, based on the recognition results, we design targeted jamming methods including conventional jamming patterns and a deep reinforcement learning-based jamming pattern. Moreover, a practical intelligent jamming demonstration system is designed and built based on the software-defined radio (SDR) platform for verification. The simulation and platform verification results demonstrate the effectiveness of the proposed method in the dynamic spectrum confrontation scenario.

In this paper, we analyze the anti-eavesdropping and anti-jamming performance of D2D communications with a full-duplex active eavesdropper (FAE). We consider the scenario when the FAE intrudes the D2D underlaying cellular networks, it can passively wiretap confidential messages in D2D communications and actively jam all legitimate links. A hierarchical and heterogeneous power control mechanism with multiple D2D user equipments (DUEs) and one cellular user equipment (CUE) is proposed to combat the intelligent FAE. Moreover, a multi-tier Stackelberg game is formulated to model the complex interaction among them and the existence of Stackelberg equilibrium (SE) is proved. A best response (BR) based hierarchical power control algorithm with perfect information, and a robust learning method with imperfect information are proposed to obtain the SE. The numerical results illustrate the convergence of the two proposed hierarchical power control algorithms, which are also compared with random selection algorithm (RSA).