Reducing the Computational Complexity of an RLS-Based Adaptive Controller in ANVC Applications

Abstract

In this paper a fast array adaptive IIR filter in active noise and vibration control setup is presented. This fast array implementation is an extended form of the fast array algorithms for FIR filter which is studied in literature before. Since the original algorithm derived for ANVC applications was based on RLS recursion its computational complexity was of order O(n 2) and it was also vulnerable to round-off and finite precision errors that may occur in real-time implementation of the algorithm. The proposed fast array solution of this algorithm not only reduces its computational complexity to the order of O(n) with the same performance, but also because of its matrix nature it has good numerical stability in real-time applications which is a necessity in active noise and vibration control applications.