6G communication systems, particularly massive Internet of Things (IoT), face critical security challenges in safeguarding transmissions against eavesdropping attacks. These challenges are exacerbated by the presence of intelligent eavesdroppers capable of exploiting impairments in wiretap channels. Traditional physical layer security (PLS) techniques, such as friendly jamming (FJ), typically rely on the differentiability and accurate availability of channel state information (CSI) to optimize performance. However, in real-world scenarios, non-differentiable channels (NDCs) resulting from hardware imperfections, mobility, and complex multi-path fading, pose significant obstacles to conventional gradient-based optimization methods. In this paper, we propose a novel deep learning-based FJ approach tailored specifically for NDC environments, where gradient-based techniques prove ineffective. Leveraging the Deep Deterministic Policy Gradient (DDPG) algorithm, our framework dynamically generates jamming signals to optimize secrecy rates while simultaneously minimizing the block error rate (BLER) at the legitimate receiver. Through extensive evaluation of realistic channel models, including both line-of-sight (LoS) and non-line-of-sight (NLoS) conditions, our proposed approach demonstrates superior security enhancements and robust performance against eavesdropping threats. The results highlight its effectiveness in securing communications under the challenging and dynamic conditions inherent to NDC environments.
6G communication systems, particularly massive Internet of Things (IoT), face critical security challenges in safeguarding transmissions against eavesdropping attacks. These challenges are exacerbated by the presence of intelligent eavesdroppers capable of exploiting impairments in wiretap channels. Traditional physical layer security (PLS) techniques, such as friendly jamming (FJ), typically rely on the differentiability and accurate availability of channel state information (CSI) to optimize performance. However, in real-world scenarios, non-differentiable channels (NDCs) resulting from hardware imperfections, mobility, and complex multi-path fading, pose significant obstacles to conventional gradient-based optimization methods. In this paper, we propose a novel deep learning-based FJ approach tailored specifically for NDC environments, where gradient-based techniques prove ineffective. Leveraging the Deep Deterministic Policy Gradient (DDPG) algorithm, our framework dynamically generates jamming signals to optimize secrecy rates while simultaneously minimizing the block error rate (BLER) at the legitimate receiver. Through extensive evaluation of realistic channel models, including both line-of-sight (LoS) and non-line-of-sight (NLoS) conditions, our proposed approach demonstrates superior security enhancements and robust performance against eavesdropping threats. The results highlight its effectiveness in securing communications under the challenging and dynamic conditions inherent to NDC environments.
Wireless communications are particularly vulnerable to eavesdropping attacks due to their broadcast nature. To effectively deal with eavesdroppers, existing security techniques usually require accurate channel state information (CSI), e.g., for friendly jamming (FJ), and/or additional computing resources at transceivers, e.g., cryptography-based solutions, which unfortunately may not be feasible in practice. This challenge is even more acute in low-end IoT devices. We thus introduce a novel deep learning-based FJ framework that can effectively defeat eavesdropping attacks with imperfect CSI and even without CSI of legitimate channels. In particular, we first develop an autoencoder-based communication architecture with FJ, namely, AEFJ, to jointly maximize the secrecy rate and minimize the block error rate (BLER) at the receiver without requiring perfect CSI of the legitimate channels. In addition, to deal with the case without CSI, we leverage the mutual information neural estimation (MINE) concept and design a MINE-based FJ scheme that can achieve comparable security performance to the conventional FJ methods that require perfect CSI. Extensive simulations in a multiple-input-multiple-output (MIMO) system demonstrate that our proposed solution can effectively deal with eavesdropping attacks in various settings. Moreover, the proposed framework can seamlessly integrate MIMO security and detection tasks into a unified end-to-end learning process. This integrated approach can significantly maximize the throughput and minimize the BLER, offering a good solution for enhancing communication security in wireless communication 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.
The escalating demand for secure wireless communication, driven by the proliferation of connected devices, has made information security a paramount concern. Traditional encryption methods struggle against eavesdropping and malicious jamming in open wireless channels, necessitating innovative solutions. This paper explores physical layer security, leveraging intrinsic wireless channel properties, enhanced by reconfigurable intelligent surface (RIS). We propose an RIS-assisted multiple-input single-output system to counteract eavesdroppers and jamming threats through an artificial noise (AN) cooperation scheme. AN is injected during legitimate beamforming to bolster transmission security. Jointly optimizing the base station’s precoding vector, AN vector, and RIS reflection coefficients under power, quality, and phase constraints, we formulate a non-convex secrecy rate (SR) maximization problem. Addressing its complexity, we model it as a Markov decision process and develop a deep deterministic policy gradient algorithm within a deep reinforcement learning framework. Simulation results demonstrate the proposed method’s superior performance, reduced complexity, and robustness against multiple eavesdroppers and high-power jamming, underscoring the critical roles of RIS and AN in securing wireless communications. It consistently outperforms traditional algorithms such as alternating optimization with semidefinite programming, offering up to 7% higher SR and achieving nearly 120% SR gains as RIS elements scale from N =4 to N =10.
As one of the indoor communication technologies, visible light communication (VLC) has drawn great attention for its advantages such as ultra-wide unlicensed spectrum, power saving and low complexity. The nature of the visible light propagation is an open channel, which is vulnerable to wiretapping. This paper investigates a secure VLC mechanism enabled by multiple light fixtures acting as friendly jammers. The goal of the friendly jammers is to diminish the capability of the eavesdropper to infer the undisclosed information, on the premise of causing minimal impact on the legitimate receiver. For this reason, an algorithm based on reinforcement learning is proposed to dynamically optimize the friendly jamming policy in realistic nonstationary environments. In order to resolve the difficult problem of the dimensional curse and to effectively represent the continuous state and action spaces, an algorithm based on deep reinforcement learning is devised, which utilizes deep convolutional neural networks to accelerate the convergence rate of the learning process. A differentiable neural dictionary is introduced to make full use of the experiences in similar anti-eavesdropping scenarios to improve the learning capability. Simulation results demonstrate that, the proposed schemes can achieve a higher secrecy rate and a lower bit error rate than some state-of-the-art schemes.
Traditional approaches for providing confidentiality of wirelessly transmitted signals are often based on cryptographic techniques. While these techniques ensure that messages are encrypted and secured against eavesdropping, they are not sufficient to protect important transmission attributes at the physical (PHY) layer. Specifically, several elds in the headers of PHY and medium access control (MAC) frames are typically sent unencrypted to maintain proper protocol functionality (e.g., sender/receiver identification). As a result, they leak side-channel information (SCI), including payload size, frequency and phase oset, modulation and coding scheme, identities of communication nodes, frame type, transmission rate, etc. These SCI can be exploited by an adversary to launch various passive (eavesdropping) and active (jamming) attacks. To complement cryptographic techniques used at upper layers and prevent the leakage of SCI at the PHY/MAC layers, PHY-layer security techniques have been introduced. These techniques exploit the characteristics of the wireless channel along with multiple-input multiple-output (MIMO) technologies, in which multiple transmitting and receiving antennas are utilized to prevent information leakage to eavesdroppers and/or avoid jamming attacks. In particular, if the signal-to-interference-plus-noise ratio (SINR) at an adversary is lower than the one at a legitimate receiver, secure communication can be achieved through the so-called wiretap coding. MIMO systems enable the transmitter and/or the receiver to generate artificial noise, called friendly jamming (FJ), so as to reduce the SINR at the adversary without impacting the SINR at the legitimate receiver. In this dissertation, we focus on exploring the security threats to next-generation wireless systems. We develop MIMO-based PHY-layer security methods to achieve reliable and secret communications in these systems. First, we consider a broadcast channel, in which a multi-antenna transmitter (Alice) sends K confidential information signals to K legitimate receivers (Bobs) in the presence of L eavesdropping devices (Eves). Alice uses MIMO precoding to generate the information signals along with her own Tx-based friendly jamming (TxFJ). Interference at each Bob is removed via a technique called MIMO zero-forcing. This TxFJ, however, leaves an eavesdropping \vulnerability region" in the proximity of each Bob, which can be exploited by a nearby Eve. Specically, because of the high correlation in the channel state information (CSI) at a Bob and nearby Eves, the TxFJ becomes less effective in preventing eavesdropping at these Eves. We address this problem by augmenting TxFJ with receiver-based friendly jamming (RxFJ), generated by each Bob. Specifically, we allow each Bob to use self-interference suppression (SIS) techniques to transmit a friendly jamming signal while simultaneously receiving an information signal over the same channel. These SIS techniques exploit circulators to attenuate the self-interference power of RxFJ at Bobs and utilize analog/digital cancellation methods to clean out any remaining residual self-interference. Considering the resulting broadcast channel, we rst study a two-user scenario and develop a secrecy encoding scheme for constructing the signals intended to two Bobs. We characterize the corresponding achievable secrecy sum-rate and formulate an optimization problem to maximize this metric. We then extend our approach to K-user scenarios and study a power minimization problem. Particularly, in our analysis we minimize the powers allocated to information and friendly jamming signals (both TxFJ and RxFJ) while guaranteeing individual secrecy rates for various Bobs. The problem is solved for the cases when the eavesdropper's CSI (ECSI) is known or unknown (but with a known distribution). Simulations show the effectiveness of the proposed solutions. Furthermore, we discuss how to schedule transmissions when the rate requirements need to be satised on average rather than instantaneously. In low and high power regime, a scheduling algorithm that serves only the strongest receivers is shown to outperform the one that schedules all receivers. Next, we consider a single-cell massive MIMO system in which a base station (BS) with a large number of antennas transmits simultaneously to several single antenna users. The BS acquires the CSI for various receivers using uplink pilot transmissions. We demonstrate the vulnerability of the CSI estimation process to a pilot-contamination (PC) attack that aims at minimizing the sum-rate of downlink transmissions by contaminating uplink pilots. We rst study PC attacks for two downlink power allocation strategies under the assumption that the attacker knows the locations of the BS and its users. Later on, we relax this assumption and consider the case when such knowledge is probabilistic. The formulated problems are solved using stochastic optimization, Lagrangian minimization, and game-theoretic methods. A closed-form solution for a special case of the problem is obtained. Furthermore, we analyze the \achievable individual secrecy rates" under PC attacks, and provide an upper-bound on these rates. We also study this scenario without a priori knowledge of user locations at the attacker by introducing chance constraints. Another attack model is considered, where the attacker generates jamming signals in both the CSI estimation and data transmission phases by exploiting inband full-duplex techniques. Our results indicate that such attacks can degrade the throughput of a massive MIMO system by more than 50%, and reduce fairness among users significantly.
Physical layer security (PLS) has been widely employed in the Internet-of-Things (IoT). Since most IoT devices are energy-constrained, the power efficiency of PLS schemes attracts increasing attention. Distinguished with existing power optimization-based PLS schemes, we propose a novel PLS scheme by optimizing the transmit phase to guarantee the security in an energy efficient way. On one hand, we add an initial phase shift at the legitimate transmitter to enlarge the difference between the received signal at the legitimate receiver and the eavesdropper. On the other hand, we employ a reconfigurable intelligent surface (RIS) to enhance the interference at the eavesdropper. Aiming at increasing the bit error rate (BER) at the eavesdropper with low energy cost, the phase optimization problem is solved by using a deep deterministic policy gradient (DDPG) algorithm, which is based on a deep reinforcement learning framework. Simulation results demonstrate that our proposed phase optimization scheme can effectively improve the security performance and reduce the power consumption at the transmitter.