Performance analysis of sub-band adaptive algorithms

• Main Authors: Chien-Hsin Liu, 劉建新
• Other Authors: 李仲溪
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/23455298017808993027

碩士 === 元智大學 === 電機工程學系 === 94 === As the long time of development, adaptive algorithms are used in many applications such as acoustic echo cancel, system identification and channel estimation . The least mean square (LMS) based algorithms are most popular due to their simplicity. However, they still have their drawbacks such as suffering from the eigen-value spread of input signal and long length of filter. Sub-band adaptive filter can reduce the complexity and the eigen-value spread by sub filters and split the input signal in sub-bands. In this thesis, we introduce some sub-band adaptive algorithms – SAF, NSAF and compare with NLMS and GMDF. We can see the performance of NSAF is between GMDF and NLMS by observing the experiments. In some applications (e.g. channel estimation), adaptive filter would face not the time invariant system but time varying system. Therefore we test these algorithms in time variant system to see if they can work. In the result, we found these adaptive algorithms we use can not track the system changing quickly well.