Optimized Full-Duplex Multi-Antenna Relay in Single-Input Single-Output Link

Abstract

This thesis studies the performance evaluation and optimization of full-duplex multiple-input multiple-output (MIMO) relaying systems in single-input single-output (SISO) link, based on signal-to-interference-plus-noise ratio (SINR). Relays are transceivers which can improve the throughput of a system by coverage extension in a power-efficient manner, whereas full-duplex (FD) systems are point-to-point communication systems, in which transmission and reception occurs simultaneously on a single frequency band. Deploying relaying systems in the full-duplex mode, however, causes self-interference, because the signal transmitted from the transmitter side of the relay couples at its receiver side. This interference causes performance degradation in these systems. In this thesis, a one-way SISO communication link with a MIMO relay connecting the source and the destination nodes is studied. The relay is considered to be implementing either amplify-and-forward (AF) or decode-and-forward (DF) protocol. First, the end-to-end SINR of the system is derived. With the knowledge of SINR, numerical evaluation is made via computer simulations. The numerical results are reached by introducing different assumptions to the general system, as well as by keeping the system intact. Although the numerical solutions provide high performance, they require much time and computational power. Hence, this thesis offers some computationally efficient analytical solutions to the problem. For example, after setting the transmit filter of the relay, minimum mean square error (MMSE) method is applied on the first hop to optimize the system; or by assuming the relay self-interference channel is a rank-one matrix, a closed-form solution for the transmitter and receiver relay filters eliminating the self-interference is derived. Then, the performance of these methods are compared and discussed in different aspects; such as high SINR and computational requirement. The results indicate that each scheme has certain benefits over the others depending on the system design requirements.