具低複雜度與強健性相位雜訊補償

Abstract

正交分頻多工系統因為有著高頻譜使用效益與抵抗多重路徑通道所產生之符元間干擾的強健性，近年來被許多通訊標準所採用。但是因正交分頻多工系統其子載波正交的特性，故其對於頻率同步相當敏感，其中相位雜訊為影響其系統表現最甚的因素之一，相位雜訊主要會產生兩種效應:共同相位錯誤與子載波間干擾，造成系統效能下降且產生錯誤位階(Error Floor)的現象。 本論文藉由離散傅立葉係數來得到相位雜訊的資訊，利用連續干擾消除來補償其相位雜訊產生之效應，並以Cholesky 分解與Blockwise Matrix Inversion Lemma兩種理論提出能有效降低連續干擾消除過高複雜度的問題，最後再以符元間線性內插的方式改善相位雜訊估計值在邊際可能產生偏差的現象，更進一步壓抑相位雜訊產生的錯誤位階現象。由模擬結果顯示我們所提出的方法確實對於補償相位雜訊的效應表現良好。

Orthogonal Frequency Division Multiplexing systems have been adopted in many transmission standards in recent years due to its high spectral efficiency and robustness for inter-symbol interference (ISI) in multi- path fading channel. However, it is very sensitive to frequency synchro- nization because of its requirements of the orthogonality between the subcarriers. Phase noise (PHN)is one of the factors that effect the system performance most significantly .It will induce Common Phase Error (CPE)、 Inter-carrier interf- erence(ICI) and error floor . In this thesis, we approximate PHN by discrete Fourier series and then suppress the phase noise effect by implementing successive interference cancellation. Based on the schemes, we propose a complexity reduction method by Cholesky decomposition and blockwise matrix inversion lemma. Moreover, we use the linear interpolation between two OFDM symbols to mitigate the edge problems on the edges in PHN estimation samples. It is shown by simulation results that the proposed schemes performs well to suppress the error floor.