Abstract
In this thesis, we study the timing and frequency synchronization and channel estimation for multicarrier mobile communication systems. Two typical multicarrier systems, MC-CDMA and OFDMA are considered. The sensitivity of multicarrier systems to the timing and frequency offsets are studied, and system performance is deduced by analytical methods. Analysis shows that both MC-CDMA and OFDMA are very sensitive to the timing and frequency offsets. The OFDMA system with the subband-based subcarrier assignment scheme is more sensitive to the synchronization errors of the reference user than that with interleaved subcarrier assignment scheme, while less sensitive to the synchronization errors of the interference users. We investigate the timing and frequency synchronization methods for multicarrier systems in both the downlink and uplink transmissions. The uplink case is more challenging because the uplink synchronization has to cope with independent frequency and timing offsets for each individual user. A novel timing and frequency offset estimation scheme for the uplink is proposed exploiting the principle of best linear estimation. The timing and frequency offsets are estimated by identifying the differential phases of the training pilots in frequency and time dimensions respectively. In contrast to other methods, the proposed scheme has moderate complexity and allows flexible subcarrier assignment schemes for the OFDMA system. The analysis and simulation results show that the proposed method performs well in the uplink channels. The initial synchronization tasks in the downlink including the cell identification, and the timing and frequency synchronization are studied. Several typical synchronization methods are presented and compared with each other. It shows that the MMSE method achieves the best performance for the timing estimation. In order to reduce the plateau of the metrics, a time domain smooth window is used with which a single peak is observed for all metrics. After the timing and frequency synchronization, a frequency domain correlation algorithm is applied for the cell identification. The overall performance of the initial synchronization was evaluated under multiple access interference, which gives the reference preamble power for designing the system. Finally, channel estimation algorithms are investigated. MMSE interpolation, FFT-based interpolation and polynomial interpolation methods are presented and compared with each other. Simulation results show that MMSE interpolation is superior to other methods in terms of MSE performance. However considering the trade-off between performance and complexity, the polynomial interpolation is a good approach. The effect of residual synchronization errors on channel estimation performance is studied. The modified channel estimation method to compensate residual synchronization errors is proposed. It shows that our compensation methods considerably improve the performance of channel estimators in imperfect synchronization conditions.