Transceiver Designs For Secure Communications In MIMO Systems

Nowadays,the security of communication systems has been further enhanced.Physical layer security techniques,as supplement to traditional encryption technology,have become a hot re-search topic.Recently,researches on physical layer security mainly focus on three aspects:secre-cy beamforming and precoding schemes,secrecy transmission in relay systems and artificial noise(AN).This thesis studies the combined application of these techniques in the area of physical layer security and proposes efficient transceiver design schemes for specific communication systems.The research background and developing status are introduced at the beginning of this thesis.Researches on information security theory,the classical wiretap channel model and convex opti-mization have been carried out before the design.This thesis summarizes the applications of the beamforming techniques,relays and AN in physical layer security.Advices on further exploration are given after analysis on the existing researches.On this base,we study two different multi-input multi-output(MIMO)communication systems and propose specific transceiver design schemes.Beamforming and relay techniques are jointly used in the first system while secrecy beamforming scheme with artificial noise is applied in the second system.In this thesis,a MIMO relay communication system with one legitimate receiver and multiple eavesdroppers is studied.The beamforming vector and the relay amplify and forward(AF)matrix are jointly designed to maximize the signal to noise ratio(SNR)of the legitimate receiver under power constraints and a set of secrecy constraints.A novel iterative algorithm based on alternating optimization(AO)and a switched relaying(SR)scheme are proposed when the eavesdroppers'channel state information(CSI)is available for the design.Robust algorithms are developed by using the S-procedure when the eavesdroppers' CSI is imperfect.In addition,a simplified SR algorithm with reduced complexity is derived while the codebook of permutation matrices for the SR scheme is constructed using an efficient off-line approach.Extension to the scenario of colluding eavesdroppers is also discussed in this thesis as a complement.Simulation results show that the proposed AO and SR algorithms outperform the traditional algorithms with significantly reduced computational complexity and signaling overhead.A MIMO secure communication system with multiple users and multiple eavesdroppers is also studied.Two assumptions of the legitimate users and the eavesdroppers'CSI are discussed in this thesis.The beamforming vector and the AN are jointly designed to maximize the sumrate of the legitimate receivers under a power constraint and a set of intercept probability constraints.The original optimization problem is transformed into the equivalent weighted mean-square error minimization problem before the intercept probability constraints are rewritten as a set of linear matrix inequality(LMI)and second order cone(SOC)constraints.Robust algorithms based on AO are proposed under the two different CSI assumptions.Simulations confirm the security and robustness of the proposed design approach against eavesdropping with low complexity.