On The Blind And Compensation Based Frequency Offset Estimation Techniques In Mobile Communications

Frequency offset between the received carrier and local reference at the receiver may occur due to Doppler shifts and/or oscillator instabilities after down conversion in digital communication system. The primary goal of this paper is to find efficient and precise methods to estimate the frequency offset, which is also a pre-requisite problem in wireless communication research. The frequency offset estimation is one of the key synchronization technologies in coherent detection in burst mode transmission. Since the precision of frequency offset estimation will directly affect the performance of receivers, it is important to investigate the frequency offset scheme with improved accuracy. And this is exactly the research motivation of this thesis.Firstly, the fundamental knowledge of frequency offset and those possible sources resulting in the frequency offset are reviewed. Meanwhile, the principle and the significance of estimation frequency offset are presented. Then the details of two kinds of pilot format in TD-SCDMA and LTE standards are introduced to highlight the pilot signal design utilized for both frequency offset and channel estimation in practical burst mode wireless communication system. Also, the performance metric as well as the theorectical Cramer-Rao low bound are presented for the comparision of different frequency offset estimation schemes. Some classic algorithms for frequency offset estimation over AWGN channel are presented and compared through simulations to show their applicabilities in terms of estimation range, estimation accuracy and calculation complexity. At the same time, the pilot signal design and its influence on the achieved frequency offset estimation performance are addressed and it is disclosed that the pilot placement scheme should be taken into consideration to achieve the optimal estimation performance. After a brief survey of the existing blind frequency offset estimation scheme, the blind frequency offset estimation problems for the MPSK signals are investigated. Basically, the M-power based nonlinear transformation could be utilized to eliminate the effect of the unknown data symbols on the frequency offset estimate. By combining Kim's algorithm with the existing estimation methods over AWGN channel, a blind frequency offset estimation scheme is proposed in this thesis. It is shown that the proposed blind frequency offset estimation scheme is able to achieve the reasonable accuracy close to the Cramer-Rao bound for a SNR as low as 0 dB. The improved estimation accuracy, a lower SNR threshold, together with the less computational complexity involved make the proposed scheme appealing for medium/low SNR communication systems.Finally, the frequency offset estimation problem over the flat-fading and the frequency-selective fading channel are discussed. Different from the frequency offset estimation problem over the AWGN channel, now the received signals become time varying. Basically, the algorithms suitable for the AWGN channel are no longer suitable for the estimate problem over the fading channel. Some of the mostly commonly utilized frequency offset estimation algorithm for the flat fading channel, such as the KF, MKF, uMMV, NLS and Simplified NLS algorithms are reviewed and compared through simulations as well. And It is show that the achieved estimation performance for all of these algorithms are not good enough to approach the CRB over the flat-fading channel. An estimation scheme based on channel estimation and compensation over flat-fading channel is proposed to enable the following two step estimation scheme. Firstly, using the Simplified NLS method to estimate the initial frequency offset. Secondly, the fading distortion is evaluated and compensated to transform the frequency estimation issue over the AWGN. Simulation results show that the proposed scheme can approach the CRB for AWGN channel.