Modified Trimmed Mean FxLMS for Impuslive Noise Reduction in ANC

• Main Authors: Bo-Wei Jiang, 江柏緯
• Other Authors: Dah-Chung Chang
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/00796010472772582528

碩士 === 國立中央大學 === 通訊工程學系 === 104 === The Filtered-X Least Mean Square (FxLMS) algorithm has been widely used for Active Noise Control (ANC), which has a satisfying convergence performance in the Gaussian noise. Hence, there have been many researches studying how to improve its performance in the literature. Nevertheless, in practical applications, the ANC system can receive an unusual signal suddenly generated and called the impulsive noise (IN),which can result in serious performance degradation or convergence failure to the conventional FxLMS algorithm. Some references can be found discussing this issue, but many of the proposed solutions highly relies on the setup of pre-determined parameters dependent of special stituations for the modified FxLMS algorithms. In this thesis, the IN problem is dealt with mainly applying the trimmed mean method, which is adapted from order statistics. With the new FxLMS algorithm, there is no prerequisite to determine the parameters as required in the previous works. Through simulation results, we will show that the proposed method has better average noise reduction (ANR) performance compared to previous works.