Adaptive set-membership normalized least mean squares: An adaptive filter for the systems with bounded noise

A new set-membership filtering method, adaptive set-membership normalized least mean squares (ASM-NLMS), is presented that introduces a forgetting factor into the conventional set-membership normalized least mean squares (SM-NLMS) method. The proposed ASM-NLMS method is more effective in dealing with non-stationary systems compared to the SM-NLMS method. The performance of the proposed ASM-NLMS method and the conventional SM-NLMS method in reducing the effect of motion artifacts in the impedance signals measured on the arms are compared. It is shown that the proposed ASM-NLMS has better accuracy and less computational complexity compared to the conventional SM-NLMS.