Study On Artificial Noise For Wireless Communication Security

With the commercialization and development of the fifth generation(5G)mobile communication system,the data transmission capability of wireless communication systems has been proliferating.However,due to the nature of broadcast propagation,signal transmission of wireless systems suffers from a critical threat of information leakage,which poses a challenge to improving wireless communications security.In theory,the transmitter is capable of utilizing the spatial degrees of freedom(DOF)of the wireless channel to generate artificial noise(AN)within the null space of the legitimate channel.Due to the difference between the legitimate channel and the wiretap channel,AN is capable of reducing the channel capacity of the eavesdropper,thereby improving the secrecy performance of systems.In addition,from the perspective of communication scenarios in beyond 5G(B5G)systems,on the one hand,the vision of the Internet of Everything demands few radio frequency(RF)chains and high power efficiency,thus AN technology faces challenges of limited spatial DOF and power dissipation.On the other hand,conventional physicallayer security systems often underestimate the interception capability of eavesdroppers and are unable to accurately evaluate the secrecy performance of AN under various interception methods.Addressing the aforementioned challenges,this dissertation delves into the AN technology,the contributions of which are summarized as follows:Firstly,in order to address the issues of limited spatial DOF and power dissipation in scenarios with few RF chains,the generalized spatial modulation-based AN secure transmission technology is proposed.Since conventional AN schemes refuse to use the prior knowledge of the transmitted signal and just use complex Gaussian random vectors without prior knowledge,they suffer from the problems of power dissipation and insufficient jamming intensity.In response to the problem of power dissipation,the power-optimized AN design is proposed.By minimizing the total transmission power,the closed-form expression for the power-optimized AN is derived.Moreover,in response to the problem of insufficient jamming intensity,the jamming intensity-optimized AN design is proposed.By minimizing the Euclidean distance between the transmitted signal and the jamming signal,the closed-form expression for the jamming intensity-optimized AN is derived.Furthermore,the secrecy capacity and the bit-error rate performance under each technical solution are analyzed to quantify the secrecy performance.Simulation results verified the accuracy of our theoretical analysis and proved that the proposed two AN schemes are superior to conventional schemes.Secondly,in order to address the issue of power dissipation in high power efficiency scenarios,the reconfigurable intelligent surface(RIS)-based AN secure transmission technology is studied.From the development of 5G wireless technology,the RIS is capable of reducing transmission power dissipation.However,in order to optimize the power of AN,a large number of RIS phase shift parameters need to be designed,and conventional exhaustive search suffers from a high computational complexity.In response to the above challenges,the phase shift design based on the alternating direction(AD)method is proposed,avoiding exhaustive search and effectively reducing the computational complexity in multiple-input single-output(MISO)systems.Moreover,in order to extend this algorithm to multiple-input multiple-output(MIMO)scenarios,the design for beamforming and phase shifts based on the dual-cycle AD algorithm is proposed.Furthermore,the single-cycle AD algorithm is proposed and closed-form expressions for phase shifts are derived,solving the problem of an excessive number of cycles introduced by the double-cycle AD algorithm.In addition,the RIS-assisted passive beam attack technology is proposed to address the issues that conventional active attack technologies are prone to exposure.This technology aims to construct a self-interference beam between the RIS-reflected signal and the direct signal,achieving attack without sending jamming signals.The closed-form expressions for the RIS location,unit number,and phase shifts are derived by minimizing the achievable rate of the legitimate receiver.Simulation results show that the proposed algorithm has a faster convergence speed in contrast to conventional schemes,while RIS can effectively achieve passive attacks.Thirdly,in order to address the issue of limited spatial freedom in single RF chain scenarios,the stacked intelligent metasurfaces(SIM)-based AN secure transmission technology is proposed.By stacking multiple programmable metasurface layers,SIM is capable of fitting multi-antenna transmission signals,thereby breaking through the constraint of null space for the SISO channel and reducing the hardware requirements of AN technology on RF chains.To begin with,the SIM-based AN secure transmission model is proposed,and the joint optimization problem of signal fitting error and transmission power is formulated.Moreover,in order to reduce computational complexity,closed-form expressions for phase shifts of SIM and transmission power are derived.Furthermore,the upper and lower bounds of the secrecy rate are analyzed by considering the signal bias introduced by SIM.Simulation results demonstrate that SIMs with more than 6 layers are capable of performing joint modulation,beamforming,and AN,while the theoretical bounds of the secrecy rate are tightly close to the actual one at the same time.Fourthly,considering that the interception capability of the eavesdropper may affect the jamming effect of AN,the adversarial techniques that eavesdroppers may adopt against AN are also studied.When eavesdroppers can obtain legitimate channel state information(CSI),the null-space elimination technology is proposed,and its secrecy rate and bit-error rate are derived.Analytical and simulation results prove that when eavesdroppers have sufficient antennas,they can effectively eliminate AN and recover information,while compared to conventional zero-forcing elimination,the proposed technology can reduce the hardware requirements of eavesdroppers and improve their detection quality.Moreover,when eavesdroppers are unable to obtain legitimate CSI,the artificial-noise-to-signal ratio(ANSR)is defined and the goal of minimizing the ANSR is determined.Based on minor component analysis,principal component analysis,and knearest neighbor algorithms,solutions to the problem of artificial noise elimination for single classification,multiple classifications,and unclassified samples are proposed,respectively.Simulation results indicate that compared to the conventional hyperplane clustering algorithm,the proposed algorithms can more effectively eliminate AN.In summary,this dissertation conducts in-depth research on the AN technology to meet the requirements of future wireless communication.From different perspectives of the transmitter,receiver,and eavesdropper,this dissertation improves and evaluates the security performance of AN.The technical methods proposed in this dissertation can provide theoretical analysis and technical reference for the physical-layer security direction of future wireless communication systems represented by B5 G,which have good theoretical significance and application value.