Comparative Study Of Adaptive Filter In Noise Cancellation

Wong, Pooi Mun (2017) Comparative Study Of Adaptive Filter In Noise Cancellation. Project Report. Universiti Sains Malaysia, Pusat Pengajian Kejuruteraan Mekanik. (Submitted)

Download (1MB) | Preview


Adaptive filters have been widely used in adaptive noise cancellation (ANC) applications, including telecommunication. Various adaptive filters that uses least mean square (LMS) algorithm as basis are available with each performance varies in terms of convergence rate and accuracy in estimation of noise for noise reduction. This paper compares the performance parameters between three adaptive filters: single LMS, cascaded LMS and cross-coupled LMS by evaluating the mean square error (MSE), improved signal-to-noise ratio (SNR) and convergence rate. Simulation models of the respective adaptive filter were built in LabVIEW. Using these models, ANC was simulated by cancelling noise from corrupted speech at the optimum step-size of each respective adaptive filter. The simulation results are validated through measurements carried out in real-time using myRIO 1900 real time (RT) platform. It was found that cascaded LMS filter has the highest improved SNR, smallest average MSE at its respective optimum step-size and the fastest convergence rate at the same step-size as the other adaptive filter. Cross-coupled LMS albeit able to perform when the noise reference input was corrupted by the desired speech, has the lowest improved SNR, largest average MSE and the lowest convergence rate. This meant that the ascending order of the most accurate and effective adaptive filter was cross-coupled LMS, single LMS and cascaded LMS.

Item Type: Monograph (Project Report)
Subjects: T Technology
T Technology > TJ Mechanical engineering and machinery > TJ1-1570 Mechanical engineering and machinery
Divisions: Kampus Kejuruteraan (Engineering Campus) > Pusat Pengajian Kejuruteraan Mekanikal (School of Mechanical Engineering) > Monograph
Depositing User: Mr Mohamed Yunus Mat Yusof
Date Deposited: 20 Jul 2022 03:00
Last Modified: 20 Jul 2022 03:00

Actions (login required)

View Item View Item