Adaptive IIR notch filters based on new error criteria

Modified indirect and direct plain gradient (MIPG and MDPG) algorithms using new error criteria for second-order adaptive IIR (infinite-impulse response) notch filters (ANFs) with constrained poles and zeros are proposed in this paper. The proposed techniques are based on modifying error functions of the standard indirect and direct ANFs. In this work, the proposed error criteria are modified as the mean value of the product of the two signals which are, respectively, produced by the FIR (finite-impulse response) and IIR sections of the ANFs. Due to a non-zero slope straight line-like characteristic of the new cost functions, the gradient based adaptive algorithm rapidly converges to an optimum solution resulting in faster convergence speed than that of the conventional method. The closed form expressions for steady-state estimation bias and mean square error (MSE) of the proposed algorithms are also carried out. In addition, extensive simulations of the proposed algorithms compared to other methods are provided to support the theoretical analyses. Although, there are some limitations in the proposed algorithms, however, the demonstrated results obtained from tested algorithms at the same condition confirm that the proposed MIPG and MDPG algorithms are superior to others in convergent speed.