Precoding and Equalisation for Broadband MIMO Systems

Joint precoding and equalisation can help to effectively exploit the advantages of multi-input multi-output (MIMO) wireless communications systems. For broadband MIMO channels with channel state information (CSI) such techniques to date rely on block transmission where guard intervals are applied to mitigate inter-block (IBI) and inter-symbol interference (ISI) but reduce spectral efficiency. Therefore, this thesis investigates novel MIMO transceiver designs to improve the transmission rate and error performance. Firstly, a broadband MIMO precoding and equalisation design is proposed which combines a recently proposed broadband singular value decomposition (BSVD) algorithm for MIMO decoupling with conventional block transmission techniques to address the remaining broadband SISO subchannels. It is demonstrated that the BSVD helps not only to remove co-channel interference within a MIMO channel, but also reduces ISI at a very small loss in channel energy, leading to an improved error performance and transmission rate. Secondly, a design for jointly optimal precoding and block decision feedback equalisation that can operate at low redundancy is proposed and demonstrated to outperform existing similar design approaches. Thirdly, an approach to improve error performance in block transmission systems is proposed. This approach involves a modification of the precoder and an elimination of IBI either by sharing the guard interval between transmitter and receiver or by using a BDFE. Lastly, in order to assess the proposed BSVD methods in more realistic communications scenarios, an order reduction of matrices within the BSVD is proposed in order to contain the system complexity. It is shown that the introduced error can be controlled and confined below a pre-set threshold. The effect of such an error — which can also arise from non-perfect CSI — on the BSVD-based system design is investigated. Simulation results show that the proposed BSVD-based designs are more robust to estimation or truncation errors than state-of-the-art block-based transmission system.