<p>An improved proportionate type adaptive filter based on modified Champernowne cost function (CMAF) named as improved proportionate CMAF (IPCMAF) algorithm is suggested for system identification and impulsive noise removal from electrocardiogram (ECG) signals. The Champernowne adaptive filter was proven to be resilient in an environment with impulsive noise but the proportionate term added in the Champernowne adaptive filter capitalizes on the sparse characteristics of the system and sparse nature of the input ECG for faster adaptation. Therefore, the suggested IPCMAF algorithm converges about 29% faster and turns out to be effective for system identification whereas a correlation coefficient value near one is achieved to address the suppression capability. Taylor series expansion is used to get the excess mean square error (EMSE) and steady-state mean square deviation (MSD) closed-form expressions. Additionally, simulations conducted for sparse system identification and noise cancellation demonstrate that the suggested approach performs better and is more robust in an environment with impulsive noise.</p>