Study On Synchronization Technology For OFDM

Orthogonal Frequency Division Multiplexing (OFDM) is a kind of multi-carrier modulation technology with high spectrum efficiency and the ability in coping with inter-symbol interference and multi-path fading, which make it be suitable for high speed wireless transmission, and it is considered as a key technology in 4G communication. But it has some shortcomings such as being sensitive to synchronization error. In detail, the high frequency utilization and the robust transmission of OFDM are founded on the orthogonality between the sub-carriers, once the frequency offset is generated and orthogonality between the sub-carries is destroyed, that will cause inter-carrier interference (ICI), the timing error also introduce inter-symbol interference (ISI) .Thus to study the synchronization techniques of OFDM and to do effective estimation and compensation of timing and frequency offset are very important. Synchronization technology for OFDM is studied deeply in this paper, the main work will be described as followings:At the beginning of this thesis, the basic principle of OFDM system is researched deeply, including the orthogonality between the sub-carrier for OFDM system, the realization of OFDM system based on IDFT/DFT and the structure of OFDM symbol with Cyclic Prefix (CP). And next the effects of symbol timing error and the carrier frequency offset on the performance of OFDM system are discussed and deduced in theory, and the importance of synchronization is illustrated.Secondly, the principle of ML algorithm which based on cyclic prefix is analyzed, ML algorithm, which exploits the redundant information contained within the cyclic prefix and adopts the method of maximum likelihood estimation, is a joint symbol timing and carrier frequency offset estimation. And some improved algorithms that have been existed are also analyzed. Base on the analysis, an improved ML scheme is proposed, the fore feedback structure which makes the system have a memory is used to optimize the ML algorithm. Simulation results show that the improved scheme is better than the traditional ML algorithm.In the end, the characteristic of training symbol and the principle for SC algorithm and improved schemes are researched. The advantage of the training symbol with symmetric structure in estimating timing offset is analyzed, and the definition of the timing metric is modified. Base on the analysis, an improved training symbol with symmetric and repetitive structure is proposed, and the estimate range of the frequency offset is enlarged by increasing the number of repetitive sequence and shortening the interval of repetitive sequence. Aiming at the characteristic of the training symbol, a new training symbol weighted by the pseudo-noise (PN) sequence is designed. The symmetric property of the training symbol and the correlation property of the PN weighted factor are utilized in the timing offset estimation to improve the accuracy. Based on the timing synchronization, the PN weighted factor in the training symbol is removed, and the renewed training symbol is used to estimate the frequency offset. Simulation results show that the accuracy of timing synchronization and the estimate range of the frequency offset for the improved joint timing and frequency synchronization scheme based on a training symbol ourperform the conventional SC algorithm.