Research On Synchronization Techniques For OFDM Systems

OFDM is a high-speed multi-carrier transmission scheme, which is robust to multipath fading and has high spectrum efficiency. As a B3G or 4G technology, it attracts wide attention and has been adopted in many wide-band wireless digital communication standards. OFDM system has a number of orthogonal sub-carriers. Synchronization errors would cause severe Inter-Channel Interference (ICI) and Inter-Symbol Interference (ISI). Therefore, synchronization is extremely crucial to OFDM systems. The synchronization techniques for burst packet transmission mode OFDM systems is studied, including timing, frequency offset estimation and channel estimation.The relationship between the performance of least sum of squared errors (LSSE) channel estimation based on time-domain training sequence and the auto-correlation property of the training sequence is analyzed. To improve the performance of the channel estimation, a new method based on complementary code (CC) is proposed, which utilizes the ideal auto-correlation property of the complementary code. It achieves the low bound of the variance of LSSE channel estimation based on time-domain training sequence, and has lower computational complexity. It is also found that the channel estimate is sensitive to the frequency offset.To employe the new complementary training sequence and alleviate the impact of frequency offset on the channel estimate, a scheme including coarse timing, frequency offset estimation/compensation, channel estimation and fine timing is designed. Before channel estimation, coarse timing and coarse frequency offset estimation/compensation are performed to reduce frequency offset. Therefore the advantage of the new channel estimation method is fully exploited. The proposed scheme is evaluated in MATLAB. Simulation results validated its superiority over the conventional synchronization based on time-domain training sequence.Focusing on frequency offset in OFDM systems, a novel multistage frequency offset estimation algorithm based on time-domain training sequence is proposed. This algorithm avoids the conflict between estimation accuracy and estimation range by removing the phase ambiguity in every stage, and utilizes the best linear unbiased estimation (BLUE) to further improve the estimation accuracy. Numerical analysis and simulations show that this algorithm reduces the computational complexity significantly for the estimation of large frequency offset while maintains a good performance. The main operations of this algorithm can be carried out in parallel, thus it can be applied in high data rate communication systems.