Research On Synchronization Algorithms In OFDM System

Orthogonal frequency division multiplexing (OFDM) has become the key technique in broadband wireless access and the next generation (Beyond 3G) mobile communication system. Extensive attentions have been paid to OFDM due to its high spectral efficiency, high data rate and well anti-multipath ability. The orthogonal multicarrier modulation, which is the main character of OFDM, determines that it is very sensitive to both timing synchronization error and carrier frequency offset. So it is important for OFDM system to obtain accurate symbol timing and carrier frequency synchronization, and now synchronization in OFDM has become a hotspot in the research of modern communication.In this paper, the evolution history of OFDM system is reviewed, and the fundamental principles in OFDM are briefly introduced. After analyzing the effects of synchronization errors to the system performance, many kinds of synchronization algorithms, such as the cyclic prefix (CP) based, cyclostationary based and training symbol based joint symbol timing and carrier frequency synchronization algorithms, and also the subspace based carrier frequency synchronization algorithms are introduced and discussed in detail. The system simulations and performance analyses are made to each algorithm.Based on analyses of the existing algorithms, some modified synchronization algorithms are proposed in this paper. As to training symbol based algorithms, a modified algorithm based on both CP and training symbol is proposed, which overcomes the defect in timing synchronization of the traditional algorithms, and also improve the carrier frequency synchronization. There are also two modified methods according to the two existing subspace based frequency synchronization algorithms. The proposed algorithms use the inherent CP to construct the subspace model, which would avoid the decrease in spectral efficiency caused by the using of virtual subcarriers in traditional algorithms. Simulation results show that the proposed algorithms can attain comparable performance while the spectral efficiency is dramatically improved. And when the SNR is higher than some fixed value, the synchronization performance of the modified algorithms are better than those of the traditional methods.