Research On Phase Noise Cancellation Algorithm For OFDM Systems Based On Linear Interpolation

Orthogonal Frequency Division Multiplexing(OFDM)has been applied because of its advantages of high bandwidth efficiency and robustness against frequency selective fading ability,which solves the problem of high speed information transmission in wireless channel,and has become the core technology in the field of broadband wireless communications.Phase noise(PN)of the transceiver frequency synthesizer leads to CPE and ICI,which destroys the orthogonality of subcarriers.The elimination of CPE is easy because it is same to each sub-carrier in an OFDM symbol.The estimation of ICI is more difficult because of its randomness.In this thesis,we have analyzed phase noise problem of OFDM systems and provided some effective algorithms,include the following:This thesis proposes a linear interpolation algorithm to solve phase noise in OFDM systems.The power spectral density of Wiener phase noise decays slowly near the carrier frequency,of which the ICI interference caused by the adjacent subcarriers is relatively serious.According to the characteristic,the truncated ICI can be used to estimate the phase noise.At this point,an inherent problem arises from the fact that Truncation operation brings spectrum leakage.The traditional interpolation algorithm is based on two estimations to determine the interpolation interval in the symbol boundaries,thus the precision is limited.We propose a comparison method of the observation point slope to determine the interpolation interval,and compensate phase noise of symbol boundaries by the linear interpo.lation.This thesis proposes the adaptive weighted phase noise elimination algorithm.As the phase noise bandwidth spreads,the CPE interference gradually decays and ICI introduces a greater impact.According to the low-pass characteristics of Wiener's phase noise,the continuous piloted can estimate CPE and ICI by using truncated ICI,of which the performance is better at high SNR.Contrastively,the method based on decision-assisted to estimate a few frequency domain components has a relative advantage under low SNR.According to the characteristics of the two methods,this thesis proposes a self-adaptive weighted algorithm,where the boundary phase noise is reconstructed using linear interpolation in time domain.The method achieves the goal of flexibly adjusting the weighting coefficients to improve the system robustness.