Closed-form Design and Application of Adjustable Fractional Delay Filter

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 95 === In many applications of signal processing, there is a need for a delay which is a fraction of a sampling period. These applications include synchronization in digital Modem, time delay estimation, beam steering of antenna array, position of voice source, spe...

Full description

Bibliographic Details
Main Authors: Cheng-Sheng Chung, 鍾承昇
Other Authors: Chien-cheng Tseng
Format: Others
Language:zh-TW
Published: 2007
Online Access:http://ndltd.ncl.edu.tw/handle/65434792372565974234
Description
Summary:碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 95 === In many applications of signal processing, there is a need for a delay which is a fraction of a sampling period. These applications include synchronization in digital Modem, time delay estimation, beam steering of antenna array, position of voice source, speech coding and synthesis, modeling of music instrument, comb filter design, A/D conversion and sampling rate conversion. Generally speaking, the design method can be classified into two categories. One is fixed fractional delay (FFD) and the other is adjustable/variable fractional delay (VFD) filter. In this thesis, we design one dimension (1D) and two dimension (2D) VFD filters using Taylor series expansion and weighted least-square (WLS) method, and the closed-form solution is desired. The details of above methods are described in chapter 2 and chapter 3. In chapter 2, three VFD filter design methods are presented. First, we design 1D VFD filter using 1D Taylor series expansion and then the WLS method is applied to design VFD filter. Finally, the WLS with filter coefficient symmetry is used to design VFD filter. In chapter 3, we extend the methods in chapter 2 to design 2D VFD filter. The 2D Taylor series expansion and 2D WLS design method have been used to design 2D VFD filter, and finally, we use digital signal interpolation experiments which included fractional sample period delay in 1D digital signal, resampling in 1D digital speech signal and 2D digital image interpolation to demonstrate that the performance of the applications using adjustable\variable fractional delay filter is better than traditional bilinear method.