Adaptive system identification is an active area of research in the field of signal processing due to the fact that it updates the unknown parameter of the filter using a suitable adaptive algorithm. Due to great advancements in digital signal processors in terms of high speed and low power consumption, adaptive algorithms are able to calculate the parameters of interest very fast with reduced complexity. Taking one of the practical aspect of adaptive algorithm such as convergence speed, many algorithms have been found in the literature that quickly converge to some steady state value. But presence of outliers in the data limit the performance of these algorithms. Though minimum Wilcoxon norm [1] has been proposed to overcome the effect of outliers, yet the speed of convergence can still be improved. A convex combination of two adaptive filters have been taken into account to tackle the above limitation. Results show that convex combination of filters work very efficiently against conventional adaptive algorithms, giving faster convergence. But in the presence of outliers its effectiveness is not up to the mark. A normalized median based approach is proposed, which is applied along with the convex combination in order to improve the performance. Simulation results defend the above statement and verify the authenticity of the research. © 2019 IEEE.