Abstract
Radio frequency interference (RFI) is the most threatening interference in very high-speed digital subscriber line (VDSL) transmission systems. Recently, a parametric approach is used for RFI estimation and suppression. In this kind of approach, the narrowband characteristic of RFI is utilized to facilitate the construction of a closed-form model and the RFI components spilled over the modulated tones can be known after the model parameters are estimated. The suppression is performed in the frequency domain after the discrete Fourier transform (DFT) demodulation. Theoretically, the parameters in the RFI model can be determined by replacing two measured unmodulated tones near the RFI center frequency into the closed-form equations. However, this seems to be impractical since the corresponding two equations are nonlinear. Although each of the two equations can become linear by some approximations, the modeling accuracy is sacrificed. In other words, there exists compromise between accuracy and linearity. In this thesis, we propose a modified parameter estimation approach to not only enhance the accuracy but also have the linear form simultaneously. Additionally, we propose an RFI-detection method to locate the RFI automatically since the information about the RFI is unknown in advance. Moreover, we try to handle the multiple RFI case with an extension from the single RFI case under some assumptions. Finally, some factors crucial to the RFI suppression performance are examined and discussed.