Artificial Intelligence (AI) technology is developing rapidly, permeating every aspect of human life. Although the integration between AI and communication contributes to the flourishing development of wireless communication, it induces severer security problems. As a supplement to the upper-layer cryptography protocol, physical layer security has become an intriguing technology to ensure the security of wireless communication systems. However, most of the current physical layer security research does not consider the intelligence and mobility of collusive eavesdroppers. In this paper, we consider a MIMO system model with a friendly intelligent jammer against multiple collusive intelligent eavesdroppers, and zero-sum game is exploited to formulate the confrontation of them. The Nash equilibrium is derived by convex optimization and alternative optimization in the free-space scenario of a single user system. We propose a zero-sum game deep learning algorithm (ZGDL) for general situations to solve non-convex game problems. In terms of the effectiveness, simulations are conducted to confirm that the proposed algorithm can obtain the Nash equilibrium.
Artificial Intelligence (AI) technology is developing rapidly, permeating every aspect of human life. Although the integration between AI and communication contributes to the flourishing development of wireless communication, it induces severer security problems. As a supplement to the upper-layer cryptography protocol, physical layer security has become an intriguing technology to ensure the security of wireless communication systems. However, most of the current physical layer security research does not consider the intelligence and mobility of collusive eavesdroppers. In this paper, we consider a MIMO system model with a friendly intelligent jammer against multiple collusive intelligent eavesdroppers, and zero-sum game is exploited to formulate the confrontation of them. The Nash equilibrium is derived by convex optimization and alternative optimization in the free-space scenario of a single user system. We propose a zero-sum game deep learning algorithm (ZGDL) for general situations to solve non-convex game problems. In terms of the effectiveness, simulations are conducted to confirm that the proposed algorithm can obtain the Nash equilibrium.
Physical-layer security (PLS) has become an intriguing technology to address eavesdropping issues in Internet of Things (IoT) systems owing to its low complexity and latency. As wireless communication technology and computing capabilities advance by leaps and bounds, eavesdroppers have enhanced eavesdropping capabilities. They can analyse the environment and move to find better locations for eavesdropping. To mitigate the significant impact of multiple intelligent eavesdroppers extremely close to the device-constrained transmitters in the uplink multiple-input-multiple-output IoT systems, we exploit time-domain artificial noise (AN) in the PLS design and formulate a game problem against the intelligent eavesdroppers. First, we derive the optimal closed-form solution for eavesdroppers and propose a low complexity difference of the convex (LCDC) algorithm to obtain the optimal strategy of the legitimate users and the access point equipped with a zero-forcing receiver. For general linear receivers, we propose a successive convex approximation (SCA) algorithm for the game problem. Simulations are conducted to verify the convergence and effectiveness of our proposed algorithms. The system secrecy performance of the time-domain AN is much better than that of the frequency-domain AN with single-antenna transmitters.
Non-orthogonal multiple access technology (NOMA), as a potentially promising technology in the 5G/B5G era, suffers from ubiquitous security threats due to the broadcast nature of the wireless medium. In this paper, we focus on artificial-signal-assisted and relay-assisted secure downlink transmission schemes against external eavesdropping in the context of physical layer security, respectively. To characterize the non-cooperative confrontation around the secrecy rate between the legitimate communication party and the eavesdropper, their interactions are modeled as a two-person zero-sum game. The existence of the Nash equilibrium of the proposed game models is proved, and the pure strategy Nash equilibrium and mixed-strategy Nash equilibrium profiles in the two schemes are solved and analyzed, respectively. The numerical simulations are conducted to validate the analytical results, and show that the two schemes improve the secrecy rate and further enhance the physical layer security performance of NOMA systems.
Physical layer security has emerged as a promising approach to counter eavesdropping threats in the Internet of Things (IoT). With the advancement of technology, eavesdroppers (Eves) are increasingly sophisticated, exacerbating the difficulties in ensuring the security of uplink transmissions. In situations where transmitters are unable to deter Eves from approaching at close range, these intelligent Eves can attain substantial channel gains, thereby posing significant security threats. To address these challenges, we first propose a time-domain artificial noise (AN) scheme to eliminate interference at the legitimate receiver in Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) systems. Then, we model the adversarial interaction between legitimate transceivers and intelligent Eves as a game-theoretic problem. To solve the dynamic game problem, we introduce a deep deterministic policy gradient (DDPG)-based framework, where an optimal beamforming fractional programming algorithm (OBFP) and a lower-complexity zero-forcing beamforming difference of convex algorithm (ZFDC) are designed to ascertain the most effective strategy for legitimate transceivers. Simulation results validate the effectiveness of our proposed algorithms and highlight the critical role of time-domain AN in combating intelligent eavesdropping.
This dissertation investigates the physical layer security of wireless communication networks using game-theoretic and optimization models. Considered is a wireless network of parallel channels that may be under eavesdropping attack. Game models are played between a friendly interferer (he) who can broadcast artificial noise and a strategic eavesdropper (she). The friendly interferer wants to maximize the secrecy level of the wireless communication network by decreasing the eavesdropper's listening capability with intentionally generated noise signals, while the eavesdropper wants to maximize her own capability and avoid interference. The game-theoretic models reflect the strategic confrontation between the friendly interferer and the eavesdropper in different communication environments. These models address some of the limitations in literature. Most existing resource allocation strategies for friendly interference adopt the conservative assumption that all parallel channels are under attack. Game-theoretic approaches provide more efficient strategies to the friendly interferer by anticipating the channels to be attacked by the eavesdropper. The first model considers the case in which the friendly interferer can flexibly allocate cooperative jamming power to all communication channels in the network under a total power constraint. When all channels used for secret communication are under the threat of eavesdropping attacks, the optimal power allocation policy results from solving a convex optimization problem. In this case, the optimal policy is unique and can be obtained via a water-filling scheme. When the eavesdropper can not attack all channels, the eavesdropper should behave strategically and may select the targets probabilistically. In this case, a non-zero-sum game helps the friendly interferer predict and concentrate on the targets selected by the eavesdropper. When the eavesdropper can only attack a single channel, an analytical form of the NE strategies is presented; while an iterative numerical method that guarantees the convergence to NE strategies is proposed if the eavesdropper can attack multiple channels. Compared to naive power allocation strategies, such as splitting the power equally to each channel, the game-theoretic power allocation strategy always delivers better expected total secrecy capacity. Next, an extension of the power allocation game for cooperative jamming models the case in which the state information of the eavesdropping channels may not be known completely. New utility functions for the friendly interferer and the eavesdropper are designed to include the distributional information of eavesdropping channels' state information. A non-zero sum game using the new utility functions is analyzed between the friendly interferer and the strategic eavesdropper who can only attack a limited number of channels. Note that the friendly interferer may not have the capability to send cooperative jamming signals to multiple channels simultaneously. For such a friendly interferer, a probabilistic cooperative jamming scheme is devised such that the defender focuses on protecting randomly selected channels. In the mean time, the eavesdropper selects the most vulnerable channels to attack. A linearly constrained non-zero-sum bimatrix game models the competition between the defender and the eavesdropper. The Nash equilibrium is shown to be equivalent to the solution of a quadratic programming problem. A closed form solution is derived for the special case in which the eavesdropper can only attack a single channel. We show that there exists a cut-off index determining the set of channels that will be covered in the equilibrium strategies. Furthermore, the adverse effect of the interference signal plays an important role in the value of the cut-off index and it represents the vulnerability of wireless communication networks against eavesdroppers Wireless communication networks of parallel channels actually share a lot of similarities with general cyber networks. Thus, the final model in this dissertation is a cost-efficient network security game for defensive resource allocation in a heterogeneous cyber network of N nodes. A defender, by allocating defensive resources to a networked node, can decrease the node's vulnerabilities. At the same time, the cost of using defensive resources is taken into consideration. Two non-zero-sum Nash games model the competition between the defender and two types of attackers, respectively. The first type tries to maximize damage based on the value of security assets related to networked nodes, while the second type aims at infiltrating the network. Consideration of the case that the defender does not know the attacker's type leads to a Bayesian game. For all three games, we derive conditions for the existence of a unique Nash equilibrium (NE), and obtain the equilibria and the value functions in closed form.
This paper considers an adversarial scenario between a legitimate eavesdropper and a suspicious communication pair. All three nodes are equipped with multiple antennas. The eavesdropper, which operates in a full-duplex model, aims to wiretap the dubious communication pair via proactive jamming. On the other hand, the suspicious transmitter, which can send artificial noise (AN) to disturb the wiretap channel, aims to guarantee secrecy. More specifically, the eavesdropper adjusts jamming power to enhance the wiretap rate, while the suspicious transmitter jointly adapts the transmit power and noise power against the eavesdropping. Considering the partial observation and complicated interactions between the eavesdropper and the suspicious pair in unknown system dynamics, we model the problem as an imperfect-information stochastic game. To approach the Nash equilibrium solution of the eavesdropping game, we develop a multi-agent reinforcement learning (MARL) algorithm, termed neural fictitious self-play with soft actor-critic (NFSP-SAC), by combining the fictitious self-play (FSP) with a deep reinforcement learning algorithm, SAC. The introduction of SAC enables FSP to handle the problems with continuous and high dimension observation and action space. The simulation results demonstrate that the power allocation policies learned by our method empirically converge to a Nash equilibrium, while the compared reinforcement learning algorithms suffer from severe fluctuations during the learning process.
Physical layer security has emerged as a promising technique to safeguard the information security in wireless networks. In this paper, we investigate the physical layer security issue for a wireless network where there coexist multiple users with security concerns. Specifically, we tackle the problem from a distributed perspective and formulate the secure transmissions at different users as a non-cooperative game. Consider the practical situation that the legitimate transmitter may not always have the perfect information regarding the channel state information of the eavesdropper, we adopt the robust secrecy rate to combat the potential worst cases. Accordingly, the robust Nash equilibrium is employed as the solution to the resource competition game among the users. Further, we analyze properties of the equilibrium and derive the optimal transmission strategy for each individual user to maximize its own robust secrecy rate, following which the distributed algorithm is proposed for the network-wide competition to reach the equilibrium. Finally, simulation results are provided to corroborate our theoretical findings.