Effects of Channel Estimation and Implementation on the Performance of MIMO Wireless Systems

Bit-rate and quality of service demands of new wireless communication standards are pushing signal theory and algorithm implementation to their limits. One of the main strategies which are being used to achieve the demanded rates is the multiple input-multiple output (MIMO) technique, which employs multiple antennas, both at transmission and reception. This PhD dissertation concentrates on the analysis of the effects of channel estimation, specially complex due to the number of parameters to estimate, on the performance of MIMO detectors, focusing on both practical and theoretical aspects. The practical analysis has been addressed by designing and developing a real-time wireless MIMO communication platform. A whole 2 X 2 system has been implemented which has allowed to evaluate the eects of a real hardware implementation on the performance of the MIMO receiver. A zero-forcing (ZF) detector and a sphere decoder (SD) have been implemented in order to evaluate their degradation in bit error rate (BER) performance in a realistic environment including synchronization, channel estimation or quantization effects. On the other hand, a simulation-based analysis has been carried out to evaluate the effects of channel estimation on the comparative performance of the main detectors in fundamental MIMO systems, including narrowband, orthogonal frequency-division multiplexing (OFDM) and single-carrier with frequency-domain equalization (SCFDE) schemes. The effects of channel estimation on iterative Turbo-MIMO systems have also been analyzed, proposing a performance prediction tool based on analytical extrinsic information transfer (EXIT) functions. Finally, a parallel decision-directed semi-iterative channel estimation architecture has been proposed which is specially designed to t in eld-programmable gate array (FPGA)devices and solves some of the problems of other iterative approaches with a moderate complexity.