Channel estimation and carrier frequency offset estimation for wireless OFDM systems.

摘要

Orthogonal Frequency Division Multiplexing (OFDM) has received tremendous interest in recent years because of its high spectral efficiency and robustness against the effects of multipath fading in high-rate wireless systems. This dissertation addresses two main issues in OFDM systems: channel estimation and carrier frequency offset estimation.;Although channel estimation is not required for differentially modulated signals, it helps to improve system performance and channel capacity. The work on channel estimation in this dissertation is focused on the two aspects, Least Squares (LS) estimation and Minimum Mean Square Error (MMSE) estimation, both based on pilot symbol assisted modulation. For the LS estimation, we propose two methods to improve its performance. The first method is to reduce noise influence by ignoring non-significant multipath tap components in channel impulse response. The second method helps to improve transmission efficiency by using a smaller number of pilots. Compared to the LS estimation, the MMSE estimation can provide more accurate channel information based on channel statistical knowledge. However, there typically exist mismatches between a real channel and an assumed channel. Another drawback of MMSE estimation is its computational complexity. In this dissertation, two different approaches are proposed to simplify the structure of the MMSE estimator and reduce the sensitivity to channel mismatches. In the first approach, the simplification of computational complexity is performed in the frequency domain based on some observations on a weight matrix for the linear MMSE estimator. The second approach uses a low-rank channel correlation matrix to replace the full-rank matrix based on the fact that the correlation between two subcarriers is inversely proportional to the distance between them.;As a multi-carrier system, one major disadvantage of OFDM is its sensitivity to carrier frequency offsets (CFO). An accurate estimation of CFO is a crucial task for OFDM systems but should be performed with affordable computational complexity. In this dissertation, two new algorithms are presented to estimate the fractional and integer CFO respectively with the use of null subcarriers (NSC). The proposed estimator for fractional part of CFO has an iterative structure with a desirable feature of insensitive to the initial CFO. Another proposed estimator for integer part of CFO employs a binary random sequence to assist in determining the pilot arrangement and accurately estimating CFOs. Both proposed estimators offer a reduced computational complexity without degrading the estimation accuracy. Analytical expressions for the performance of the proposed methods are derived and their validity has been justified in computer simulations.