An advanced feedback Filtered-x least mean square algorithm for wideband noise removal

Adaptive filters are different from conventional filter in the sense that the adaptive filters are dynamic in nature. These filters are run by algorithms known as adaptive algorithms. Numerous adaptive algorithms have been developed since the development of DSP chips in the 1980s. These adaptive algorithms were used to tackle various issues like system identification, noise and echo cancellation. Least Mean Square (LMS) algorithm, Normalised Least Mean Square (NLMS) algorithm and Filtered-x-LMS (FXLMS) algorithm were the adaptive algorithms initially developed. Various hybrid algorithms were developed later to improve various characteristics like speed of convergence, stability and robustness. Many structures like the Volterra structure, Lattice structure and modular structure were also developed. Active Noise Control (ANC) is a method to cancel noise signals using these adaptive filters which work on the principle of optimization. In this paper a hybrid adaptive algorithm has been developed in a feedback based ANC system improving the offline modelling of the system. Simulations of the proposed algorithm have also been obtained using MATLAB as the platform. A comparative study between the proposed algorithm and the conventional feedforward FXLMS algorithm has been performed. From the obtained results it was shown that the proposed algorithm has improved characteristics.