Bias Compensated Linear Constrained Affine Projection Sign Algorithm

This paper presents a bias compensated linear constrained affine projection sign algorithm (BC-CAPSA) for linear constrained system identification in the presence of input noise and impulsive observation noise. The input reusing concept and l1-norminvolved in constrainedaffine projection sign algorithm (CAPSA) makes it suitable for impulsive observation noise for colored input. However, the presence of input noise hampers the estimation performance of CAPSA by producing a bias in the weight updateequation. In this work, weaddabias compensatorin the weight update equation of linear constrained affine projection sign algorithm (CAPSA) for making the estimation unbiased for noisy colored input. Simulations carried out in linear phase system identification problem ratify the betterment in the estimation behavior of the proposed BC-CAPSAin context ofsteady-statenormalized misalignment and convergence rate.