Filtered weight FxLMS adaptation algorithm: Analysis, design and implementation

摘要

In the Filtered-x Least-Mean-Square (FxLMS)-based Active Noise Control (ANC), the convergence speed of the adaptation process has a direct relationship to a scalar parameter, called the step size. There is a theoretical upper-bound for the step size beyond which the system becomes unstable. However, the step size is usually set to a number smaller than its upper-bound in practice. This is because for relatively large step sizes, the adaptation process becomes very sensitive to any non-stationary change in acoustic noise. Owing to this trade-off, real-time implementation of high-performance ANC systems becomes challenging. To overcome this problem, this paper develops a novel ANC algorithm in which a recursive filter compensates for influences of the step size increase on the system performance. It is shown that this filter can efficiently increase the step size upper-bound; consequently, the performance of the system is improved. This improvement is demonstrated using computer simulation. Also, experimental results shows the preference of the proposed algorithm to the traditional FxLMS-based ANC algorithm in practice.