Affine projection algorithm with coefficient vector reusing

The recently proposed normalized least mean squares algorithm reuses estimated past coefficient vectors for the finite impulse response model and effectively reduces the misadjustment. The affine projection algorithm, which is an extension of the normalized least mean squares algorithm, generally suffers from relatively high misadjustment in the steady state. To decrease the misadjustment, we adopt the method of reusing the past coefficient vectors for the normalized least mean squares algorithm to the affine projection algorithm. Through computer simulations, we show the effect and performance of the affine projection algorithm which uses the coefficient vector reusing.