Physical Layer Security In Multiuser MIMO Downlink Systems

Wireless physical layer security aims at developing effective secure communication schemes by exploiting physical properties of wireless channels. This new technique can strengthen the security of existing wireless communication systems by introduing a level of information theoretic security, which provides a new idea to solve the security problem of wireless communications and has broad research and application prospects. Multiuser multiple-input multiple-output(MIMO) systems can greatly enhance system capacity by utilizing spacial resources, which in multiuser MIMO downlink systems, permits a central base station to transmit to several users simultaneously via flexible space-time processing and precoding techniques. In this thesis, physical layer security techniques in multiuser MIMO downlink systems are studied aiming at guaranteeing user quality of service and maxmizing system secrecy capacity. This thesis studies communication means which take account of security, reliablity and efficiency of information transmission, and emphasizes transmission strategies to reduce eavesdroppers’ ability to obtain information while guaranteeing transmission reliability and efficiency of all legitimate users.From the user quality of service(Qo S) perspective, linear precoding designs for the general MIMO wiretap channels are first studied. Utilizing mean square error(MSE) as performance criteria of Qo S, joint linear minimum mean square error(MMSE) transceiver optimization designs are proposed. Assuming perfect channel state information of the legitimate user is available at the transmitter, two physical layer security transmission approaches are proposed based on whether the knowledge of the eavesdropper’s channel is available or not. The first approach considers that the knowledge of the eavesdropper’s channel is unavailable and employs artificial noise based transmission strategy, in which the transmitter maximizes the transmit power of the artificial noise for degrading the reception quality at the eavesdropper to the utmost extent while guaranteeing the given MSE requirement of the legitimate user. The second approach assumes that the eavesdropper’s channel state information is available and uses the optimal linear MMSE precoding scheme to transmit secret data streams which can maximizes the eavesdropper’s MSE while meeting the required MSE of the legitimate user. Simulation results show that the proposed physical layer security transmission approaches based on MSE criteria can degrade the eavesdroppers’ reception performance to the utmost extent while guaranteeing the required reception quality of the legitimate user, thus providing security and reliablity simultaneously.Masked beamforming schemes for secure communication of downlink multiuser MIMO systems without knowledge of the eavesdroppers’ channel are further developed respectively, according to the channel state information of the legitimate users are perfect or not at the base station. The aim of maked beamforming scheme is to maximize the transmit power of artificial noise to confuse the eavesdropper while meeting the given MSE requirements of the legitimate users. For time division duplex(TDD) systems, channel state information at the transmitter(CSIT) can be obtained from uplink training exploiting channel reciprocity. With perfect CSIT, MMSE balancing algorithm can be adopted to minimize the transmit sum-power required for successful communications in the desired user links, and uses the remaining resources to transmit an artificial noise to jam the eavesdropper. With imperfect CSIT, the orthogonality between the intended signal and the artificial noise will be lost, thus causing noise leakage problem. To solve the problem, two of the main sources of the CSIT imperfection: delay and estimation error are considered and a stochastic channel uncertainty model is first proposed. An average MSE uplinkdownlink duality with imperfect CSI is then derived by adopting a Bayesian approach.Using the duality, a robust masked beamforming algorithm is proposed. Simulation results show that the proposed robust secure communication approach can significantly reduce the sensitivity to CSIT errors to achieve the required MSE of the legitimate uses while transmitting maximum power for the artificial noise for degrading eavesdroppers’ channels.The idea of artificial interference is extended to cooperative jamming, where a friendly helper is introduced to provide the jamming signal to degrade the reception quality at the eavesdropper. For frequency division duplex(FDD) systems, the effect of quantized channel state information on the achievable secrecy rate is investigated in the MISO channel with cooperative jamming. Given a fixed total number of feedback bits available at the legitimate user, the problem how to allocate feedback bits to the transmitter and helper is studied. The strategy of maximizing the ergodic secrecy rate is first studied and an analytic expression for the ergodic secrecy rate is derived with the knowledge of statistical distribution information about the evesdropper channel. The strategy of minimizing the secrecy rate loss is then studied and an approximate upper bound for the mean loss in secrecy rate is derived for fixed power allocations at the transmitter and helper. A closedform solution with adaptively allocation of feedback bits is obtained by minimizing the upper bound. The strategy of minimizing the secrecy rate loss is next extended to the MISO broadcast wiretap channels with multiple legitimate users. Numerical simulations demonstrate that optimally allocating the feedback bits between the transmitter and helper can lead to a significant improvement in secrecy.The above three studies focus on the scenarios with single user or multiple users where the number of users is less than the number of transmit antennas at the base station.However, the number of users in practical systems is far larger than the number of transmit antennas, therefore user selection procedure is necessary before designing physical layer security based precoding schemes. The user dimension of multiuser MIMO systems can be further exploited for enhancing secrecy. Finally, the scenario where one multiantenna base station wants to communicate with a large number of single-antenna users in the presence of an external eavesdropper is considered. It is first assumed that the base station communicates with only a single user at a given time, and that the base station has knowledge of only the instantaneous SNR of all legitimate users, and only statistical information about the eavesdropper channel. Two opportunistic scheduling schemes exploiting multiuser diversity, i.e., M-SNR and A-PFS, are investigated. For the two scheduling schemes, new closed-form expressions for the ergodic secrecy rate and interception probability are, respectively, derived. Numerical simulation results demonstrate that the derived analytical results exactly match the simulation results, and proper user scheduling schemes are beneficial to secure communications. The throughput performance of orthogonal random beamforming for a realistic number of users(e.g., K < 100) is known to be severely impaired. To circumvent that, several beam selection and user scheduling algorithms featuring different complexity levels are investigated for different scenarios with eavesdropper’s channel state information. Simulation results show that the proposed dynamical beam selection and user scheduling strategies can greatly enhance secrecy.