Research On Frequency Synchronization For OFDMA Systems

As a ?exible method of multiple access, OFDMA is becoming the most compet-itive air-interface technology in the future broadband wireless mobile communicationsystems. Similarly to OFDM, OFDMA is extremely sensitive to carrier frequencyo?set (CFO), therefore the carrier frequency synchronization is the bottleneck restric-tions of the application of OFDMA. Aiming at resolving this problem, the dissertationpresents some deep research on the frequency synchronization algorithms in OFDMAdownlink, OFDMA uplink, as well as OFDMA ranging process.Firstly, the e?ects of timing o?set, sampling clock frequency o?set and carrierfrequency o?set are analyzed in the dissertation. And the accurate model of the OFDMsignal is given, which provides a theoretical basis and guidelines for the algorithm andperformance analysis.Secondly, the frequency synchronization in OFDMA downlink is investigated inthis dissertation. According to the OFDMA signal model with several synchroniza-tion o?sets, a modified method is proposed by using two sets of symmetrical pilot.With considering the e?ects of multipath and the number of subcarriers, the proposedmethod is robust to time-dispersive channel.Thirdly, this dissertation investigates the multi-user frequency synchronization inthe OFDMA uplink system with interleaved subcarrier assignment. The signal struc-ture is analyzed, then a two-stage CFO-estimation algorithm is proposed. It can es-timate all the CFOs of users simultaneously using only one OFDMA symbol. Sincethe number of users and their corresponding occupied subchannels can be estimated,the proposed algorithm can be easily extended to the random subchannel assignment.Furthermore, we present a novel CFO-compensation method based on adaptive beam-former, which performs CFO compensation and data detection simultaneously for eachactive user in the interleaved-OFDMA uplink.Finally, we focus on the CFO estimation in both initial ranging and periodic rang- ing in the OFDMA uplink. After the analysis about the probability of the number ofranging users in one time-slot, two algorithms for multi-user case are proposed basedon the correlation properties of pseudo noise (PN) sequences in time domain and fre-quency domain respectively. In Method 1, the CFO of each ranging user is estimatedby utilizing the phase of time-domain correlation results between the ranging signaland the received signal samples. Method 2 uses the phase di?erence of frequency-domain correlation results between the ranging signal and two received consecutiveOFDMA symbols. Theoretical analysis and simulation results show that he proposedfrequency-domain cross-correlation method performs better than the proposed time-domain cross-correlation method even in the case of multiple users simultaneously ex-isting in one ranging time-slot. And it is more robust to residual timing o?set resultingfrom the estimation error of symbol timing o?set (STO).