Subband IPNLMS for blind adaptive MIMO filtering with sparse impulse response systems

In acoustic signal processing, the improved proportionate normalized least mean square (IPNLMS) is known as an effective adaptive algorithm for sparse impulse response systems. And it is known that the subband adaptive filter provides fast convergence rate, because of its data prewhitening characteristic. In this paper, we propose a subband blind adaptive algorithm for multi-inputs multi-outputs (MIMO) systems. The proposed algorithm (subband IPNLMS) combines subband filtering technique with IPNLMS algorithm. In this approach, subband adaptive filtering is employed to overcome the problems in long adaptive filters such as computational complexity and slow convergence rate. Simulation results show that the subband IPNLMS performs better than the subband NLMS, when the blind channel impulse response is sparse.