Robust Proportionate Subband Adaptive Filter Algorithms With Optimal Variable Step-Size

In this brief, we address the classical tradeoff between the convergence rate and steady-state error in the proportionate subband adaptive filter (PSAF) algorithm by designing an optimal variable step-size (OVSS) strategy that minimizes the mean square deviation at each iteration. To keep low computational complexity, we also present the vector implementation of OVSS. Furthermore, we extend the OVSS mechanism to the robust PSAF algorithm, to achieve fast convergence and low steady-state error in the impulsive noise. Simulations in an acoustic echo cancellation scenario have illustrated that the proposed algorithms are superior to existing state-of-the-art algorithms.