Physical-Layer-Security-Oriented Frequency Allocation in Ultra-Dense-Networks Based on Location Informations

Abstract

In this paper, we investigate location-based frequency allocation schemes in a two-layer ultra-dense network (UDN) with a wideband eavesdropper to efficiently enhance the macro layer security in the whole working bandwidth. The cross-tier interference, treated as evil by traditional wisdom, is employed to confuse the malicious node as well as to tackle the conflict between the secrecy and traditional performances through the prudent spectrum allocation among the small cells. Three games are designed to progressively give insight into the frequency assignment problems under increasingly strict scenes. With the help of game theoretic stochastic learning approaches and location information, small cell base stations (SBSs) are endowed with the ability to distributedly select the subchannel with no requirement on the eavesdropper's CSI. First, more specifically, we focus on the security problem, aiming to promote the safety transmission of macro users by leveraging a state field, and thus formulate the spectrum selection of SBSs as a state-based potential game, which guarantees a budget-balanced utility design. The existence of a recurrent state equilibrium point is proved, and that it is able to maximize the total safety transmission probability of all subchannels. A step forward, we take the service delay of SBSs into consideration, nowadays, as numerous new services are delay-sensitive. To this end, an exact potential game, in which the equilibrium always exists, is built to help SBSs strike a balance between altruistically helping the macro users and selfishly keeping their own performances. Furthermore, a fully distributed non-cooperative game that requires no exchange among SBSs is put forward. The proposed scheme may work well even when the backhaul is limited or even unavailable since all the SBSs only depend on the observation of their own instantaneous performances. Finally, the numerical results validate the effectiveness of the proposed games on improving the safety transmission probability while guaranteeing a better service within small cells. INDEX TERMS Ultra-dense networks, physical layer security, delay, state-based game, reinforcement learning.

Physical Layer Authentication and Security Design in the Machine Learning Era

“…show the applications of ML in enhancing the security of wireless systems. Among the most popular approaches are neural networks (NNs) and reinforcement learning [21], [30]- [34]. However, there are still some contributions considering other ML types to perform secure PHY transmission, for example, support vector machine (SVM) and Bayesian solutions [35].…”

“…As another development, Li et al [32] design a Q-learningbased power control strategy for secure transmission by considering a powerful attacker having a high number of antennas. Upon using reinforcement learning, Xiao et al [33] propose an optimal beamforming scheme for visible light communication In [34], reinforcement learning is used by Miao and Wang to handle the frequency allocation problem without requiring any information exchange among base stations. In contrast to the above-mentioned papers, He et al [35] do not use NNs or reinforcement learning in designing secure transmissions.…”

“…More recently, Liu et al [174] consider an RIS-based sensing and communication system and propose an optimal NN Employing a pair of neural networks for designing wiretap codes that can minimize both the block error rate and the amount of information leakage [30] NN Considering an energy-harvesting system in the presence of an eavesdropper and using a neural network for learning system parameters that maximizes the security throughput [31] NN Using a neural network for finding power allocation coefficients that enhance the secrecy rate of a relaying system, under the impractical assumption of having the eavesdropper's CSI [32] RL Considering the Nash equilibrium in a zero-sum game between a transmitter and an attacker, and finding the Qlearning-based power control strategy for achieving the equilibrium [33] RL Using reinforcement learning for finding the optimal beamforming policy that improves the secrecy rate of a visible light communication system [34] RL Considering a two-layer ultra-dense network with the focus of securing the macro layer and proposing a reinforcementlearning-aided game-theoretic model for enhancing the security and delay performance [35] SVM, naive Bayes Using SVM and naive Bayes for selecting the optimal transmit antenna that maximizes the secrecy performance [173] not to be named Proposing learning-based attacks in favour of an attacker, where the self-defined supervised learning method utilizes the packet preamble and header as training data secure PHY transmission scheme by jointly optimizing the transmit beamforming and RIS coefficients. To deal with the proposed security optimization problem, [174] employs an RL algorithm that is built on the basis of an actor NN and a critic NN.…”

Physical Layer Security: Detection of Active Eavesdropping Attacks by Support Vector Machines

“…As for ML-aided PHY security design, neural networks and reinforcement learning constitute the most popular approaches [17], [26]- [30]. Having said that, there are some contributions considering the potential use of other ML types, for example, support vector machine (SVM) and naive Bayesian solutions [31].…”

“…He et al [26] exploit both beamforming and artificial noise [28] design a Q-learning-based power control strategy for secure transmission by considering a powerful attacker having a high number of antennas. Upon using reinforcement learning, Xiao et al [29] propose an optimal beamforming scheme for visible light communication In [30], reinforcement learning is used by Miao and Wang to handle the frequency allocation problem without requiring any information exchange among base stations. In contrast to the above-mentioned papers, He et al [31] do not use neural networks or reinforcement learning in designing secure transmissions.…”