MIMO-OFDM System Performance and Parameter Estimations in Presence of Design Imperfections

Most of the MIMO-OFDM analyses in the literature assume idealized receiver conditions in terms of time and frequency synchronization, and sampling clock offset. Moreover, often the cyclic prefix (CP) length is considered to be longer than the channel delay spread, and hence contributions from possible inter-symbol interference (ISI) in demodulation are not accounted for. This paper provides an aggregated framework to space frequency block coded (SFBC) MIMO-OFDM system performance by including major channel impairments and design issues such as insufficient CP against excess delay spread, sample timing error, frequency error, and time-varying fading. Transmit and receive filtering operations are also modeled, and bit error rate (BER) expressions are explicitly derived. We investigate different receiver detection schemes, diversity orders, and other key system parameters on BER. We also investigate the impact of excess delay spread on both channel and frequency-offset estimation. Cramer-Rao lower bounds (CRLB) are additionally derived and compared to assess the accuracy of the estimates. Along with a simple frequency offset detection technique, we derive channel and frequency offset estimation through the novel Recursive Nonlinear Dynamic Data Reconciliation (RNDDR) approach. Numerical results show improved performances of the RNDDR method compared to the traditional Kalman filter method for regular as well as step changes in parameters.