Adaptive Second-Order Volterra RLS Algorithms with Dynamic Selection of Channel Updates

This paper proposes novel adaptive Volterra recursive least square (RLS) algorithms, which dynamically choose Volterra channels for coefficient updates in order to reduce computational complexity while still maintaining the compromised performance degradation for nonlinear active noise control. The developed algorithms employ a channel selection scheme, which compares an adaptive threshold to the estimated norm (energy) of each channel input vector and then sets the corresponding channels active or inactive at each iteration step. Our experimental results show that both developed Volterra filtered-X and filtered-error RLS algorithms with a dynamic selection of channels (VFXRLS-DS and VFERLS-DS) gain the same performance as compared to their full sequential updates. In addition, both proposed algorithms could significantly reduce the computational complexity of the standard VFXRLS and VFERLS algorithms with the compromised performance degradation.