Self-interference cancellation for full-duplex MIMO transceivers

Abstract

In recent years, there has been enormous interest in utilizing the full-duplex (FD) technique with multiple-input multiple-output (MIMO) systems to complement the evolution of fifth generation technologies. Transmission and reception using FD-MIMO occur simultaneously over the same frequency band and multiple antennas are employed in both sides. The motivation for employing FD-MIMO is the rapidly increasing demand on frequency resources, and also FD has the ability to improve spectral efficiency and channel capacity by a factor of two compared to the conventional half-duplex technique. Additionally, MIMO can enhance the diversity gain and enable FD to acquire further degrees of freedom in mitigating the self-interference (SI). The latter is one of the key challenges degrading the performance of systems operating in FD mode due to local transmission which involves larger power level than the signals of interest coming from distance sources that are significantly more attenuated due to path loss propagation phenomena. Various approaches can be used for self-interference cancellation (SIC) to tackle SI by combining passive suppression with the analogue and digital cancellation techniques. Moreover, active SIC techniques using special domain suppression based on zero-forcing and null-space projection (NSP) can be exploited for this purpose too. The main contributions of this thesis can be summarized as follows. Maximum-ratio combining with NSP are jointly exploited in order to increase the signal-to-noise ratio (SNR) of the desired path and mitigate the undesired loop path, respectively, for an equalize-and-forward (EF) relay using FD-MIMO. Additionally, an end-to-end performance analysis of the proposed system is obtained in the presence of imperfect channel state information by formulating mathematically the exact closed-form solutions for the signal-to-interference-plus-noise ratio (SINR) distribution, outage probability, and average symbol-error rate for uncoded M-ary phase-shift keying over Rayleigh fading channels and in the presence of additive white Gaussian noise (AWGN). The coefficients of the EF-relay are designed to attain the minimum mean-square error (MMSE) between the transmission symbols. Comparison of the results obtained with relevant state-of-the-art techniques suggests significant improvements in the SINR figures and system capacity. Furthermore, iterative detection and decoding (IDD) are proposed to mitigate the residual self-interference (SI) remaining after applying passive suppression along with two stages of SI cancellation (SIC) filters in the analogue and digital domains for coded FD bi-directional transceiver based multiple antennas. IDD comprises an adaptive MMSE filter with log-likelihood ratio demapping, while the soft-in soft-out decoder utilizes the maximum a posteriori (MAP) algorithm. The proposed system’s performance is evaluated in the presence of AWGN over non-selective (flat) Rayleigh fading single-input multiple-output (SIMO) and MIMO channels. However, the results of the analyses can be applied to multi-path channels if orthogonal frequency division multiplexing is utilised with a proper length of cyclic prefix in order to tackle the channels’ frequency-selectivity and delay spread. Simulation results are presented to demonstrate the bit-error rate (BER) performance as a function of the SNR, revealing a close match to the SI-free case for the proposed system. Furthermore, the results are validated by deriving a tight upper bound on the performance of rate-1=2 convolutional codes for FD-SIMO and FD-MIMO systems for different modulation schemes under the same conditions, which asymptotically exhibits close agreement with the simulated BER performance.